The Magnitude of Menu Costs: Direct Evidence from Large U. S. Supermarket Chains
-
Upload
spanalumni -
Category
Documents
-
view
0 -
download
0
Transcript of The Magnitude of Menu Costs: Direct Evidence from Large U. S. Supermarket Chains
THE MAGNITUDE OF MENU COSTS DIRECT EVIDENCEFROM LARGE U S SUPERMARKET CHAINS
DANIEL LEVY
MARK BERGEN
SHANTANU DUTTA
ROBERT VENABLE
We use store-level data to document the exact process of changing prices andto directly measure menu costs at ve multistore supermarket chains We showthat changing prices in these establishments is a complex process requiring doz-ens of steps and a nontrivial amount of resources The menu costs average$105887year per store comprising 070 percent of revenues 352 percent of netmargins and $052price change These menu costs may be forming a barrier toprice changes Specically (1) a supermarket chain facing higher menu costs (dueto item pricing laws that require a separate price tag on each item) changes pricestwo and one-half times less frequently than the other four chains (2) within thischain the prices of products exempt from the law are changed over three timesmore frequently than the products subject to the law
ldquoIn principle xed costs of changing prices can be observed andmeasured In practice such costs take disparate forms in differentrms and we have no data on their magnitude So the theory canbe tested at best indirectly at worst not at allrdquo [Alan Blinder 1991p 90]
I INTRODUCTION
The costs of changing nominal prices also known as ldquomenucostsrdquo have important macroeconomic implications First menucosts can be a source of price rigidity and thus can provide a
Address all correspondence to the rst author We are especially indebtedto Peter Aranson Nathan Balke George Benston Robert Chirinko Leif DanzigerAnil Kashyap John Leahy Jeffrey Sandgren the discussants John Driscoll at theAmerican Economic Association meetings in San Francisco January 1996 andRobert Hall at the NBER Economic Fluctuations Program meeting in CambridgeMA July 1996 the editor Olivier Blanchard and an anonymous referee for pro-viding valuable comments and suggestions We are also grateful to Martin J Bai-ley Hashem Dezhbakhsh Xavier Dreze David Lilien Paul Rubin EytanSheshinski Daniel Tsiddon and seminar participants at the 1996 American Eco-nomic Association meetings the marketing and the macroeconomics workshopsat the University of Chicago the economics workshops at Emory Southern Meth-odist and Texas AampM Universities and the July 1996 NBER Economic Fluctua-tions Program meeting for useful discussions Michael Caldwell Pinaki Mitraand Georg Muller provided research assistance The second and third authorswould like to thank the Graduate School of Business of the University of Chicagofor funding All authors contributed equally to the work The usual disclaimerapplies
q 1997 by the President and Fellows of Harvard College and the Massachusetts Instituteof TechnologyThe Quarterly Journal of Economics August 1997
micro-based explanation for monetary nonneutrality Secondeven small menu costs may be sufcient to generate substantialaggregate nominal rigidity and large business cycles1 Conse-quently menu costs have received considerable attention in thetheoretical macroeconomics literature as many predictions gener-ated by traditional Keynesian and more recent new Keynesianmodels crucially depend on the existence of some form of pricerigidity2
Despite the theoretical importance of menu costs howeverlittle is known about their actual magnitude as the above quota-tion from Blinder succinctly reects Because of the practical dif-culty of measuring menu costs directly a common feature of theexisting empirical studies of menu costs is that they all provideindirect evidence3 Yet many authors including Blinder [1994]Kashyap [1995] and Slade [1996a] have emphasized the impor-tance of assessing the empirical relevance of menu costs at thelevel of individual rms For example according to Slade [1996ap 19] ldquoGiven the large number of theoretical papers that evalu-ate the implications of [price] adjustment costs obtaining directevidence that such costs are present seems crucialrdquo
Our primary contribution in this paper is providing directmeasures of menu costs at ve large U S retail supermarketchains Using a unique store-level data set we show that chang-ing prices in these establishments is a complex process requiringdozens of steps and a nontrivial amount of resources The menucosts reported in this study are made up of (1) the labor cost ofchanging shelf prices (2) the costs of printing and delivering newprice tags (3) the costs of mistakes made during the price changeprocess and (4) the cost of in-store supervision of the pricechange process4 We nd that the measurable components ofmenu costs for the four chains that are not subject to an item
1 See Akerlof and Yellen [1985] Mankiw [1985] Parkin [1986] Blanchardand Kiyotaki [1987] Caplin and Leahy [1991 1997] and Caplin [1993]
2 See for example Mankiw and Romer [1991] Sheshinski and Weiss [1993]Andersen [1994] Ball and Mankiw [1994] Romer [1996] and studies citedtherein
3 These studies include Sheshinski Tishler and Weiss [1981] Rotemberg[1982] Lieberman and Zilberfarb [1985] Carlton [1986 1989] Cecchetti [1986]Danziger [1987] Ball Mankiw and Romer [1988] Gordon [1990] Lach and Tsid-don [1992 1996] Blinder [1994] Eden [1994] Amano and Macklem [1995] Balland Mankiw [1995] Kashyap [1995] Warner [1995] Warner and Barsky [1995]and Slade [1996a 1996b]
4 We also discuss other components of menu costs including the costs ofmaking corporate level managerial price change decisions and provide some evi-dence on their approximate magnitude although we do not include these guresin the measures of menu cost we report
QUARTERLY JOURNAL OF ECONOMICS792
price law average $105887 annually per store In relative termsthese menu costs comprise 070 percent of revenues 352 percentof net margins and $052 per price change on average
Our second major contribution in this paper is providing evi-dence that these menu costs can form a barrier to price changeactivity at these chains offering direct support for the relation-ship between menu costs and store-level individual price rigidityWe present three types of evidence that these menu costs form abarrier to price change activity at these rms First we contrastthe price change activity of a chain that operates in a state withan item pricing law with the rst four chains that operate instates not subject to such laws Item pricing laws require that aseparate price tag be placed on each individual item sold (in addi-tion to the shelf price tag) We show that the average menu costper price change for the chain subject to the item pricing law is$133 over two and a half times the corresponding gure for theother four chains ($052) These larger menu costs lead to verydifferent levels of price change activity by these chains Specifi-cally the four supermarket chains that are not subject to itempricing laws on average change prices on 156 percent of the prod-ucts they carry each week In contrast the chain that is subjectto the item pricing law (and therefore faces higher menu costs)changes prices on only 63 percent of the products it carrieswhich is less than half the average of the other four chains
Second within the chain facing the item pricing law thereare 400 products that are exempt from this law and thereby facelower menu costs For these products the chain each weekchanges the prices of 21 percent of the products on average whichis over three times more frequently than for products subject tothe item pricing law Third we provide evidence from the super-market chains that the menu costs they incur form a barrier tocertain cost-based price adjustments Specically we show thatthe chains not subject to item pricing laws each week experiencecost increases on about 800ndash1000 products they sell Yet they ad-just prices of only about 70ndash80 percent of these products Theremaining 20ndash30 percent of the prices are not adjusted immedi-ately because the existing menu costs make the necessary priceadjustment unprotable Considering all three of these ndingstogether we conclude that menu costs can indeed affect pricechange activity at the level of the individual rmmdashoffering directevidence that these menu costs are relevant to marginal pricechange decisions Finally on a related macroeconomic issue we
THE MAGNITUDE OF MENU COSTS 793
provide empirical evidence which suggests that the price changeprocess in these supermarket chains has a strong time-dependent element
Relating our ndings to the existing theoretical models weconclude that the magnitude of the menu costs we nd is largeenough to be capable of having macroeconomic signicance Firstrecall that according to the studies of Akerlof and Yellen [1985]Mankiw [1985] Parkin [1986] and Caplin and Leahy [1997] evensmall menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Second when considered in the context of thetheoretical menu cost models of Blanchard and Kiyotaki [1987]and Ball and Romer [1990] we nd that the menu cost gureswe report are ldquonontrivialrdquo and their relative magnitudes cross theminimum theoretical threshold needed to form a barrier to priceadjustments
Although in this paper we provide direct measurements ofthe marginal costs associated with changing prices it should bementioned that there are still many aspects of menu costs we areunable to measure In particular we do not provide measure-ments of the marginal benets associated with changing priceswhich can be signicant [Lieberman and Zilberfarb 1985 She-shinski and Weiss 1977] At the local level this industry is ex-tremely competitive [Calatone et al 1989 Progressive GrocerNovember 1992 p 50 Chevalier 1995] In such a competitive in-dustry the benets of frequently changing prices can be high asunmatched price cuts or consumer perceptions of higher pricescan lead to signicant losses in sales5 This helps explain whydespite the magnitude of the menu costs we found we still ob-serve frequent weekly price change activity by the chains westudy For example stores change an average of 15ndash16 percent oftheir prices each week Also they seem to adjust the prices of70ndash80 percent of the products for which they experience cost in-creases6 Thus although we have evidence that the menu costswe report in this study clearly matter in the sense that they cre-
5 Also changing prices frequently can make it more difcult for customersto compare prices of branded items across supermarkets because of higher searchcosts [Carlton 1986] which is valuable for creating differentiation between retailoutlets [Bergen Dutta and Shugan 1996]
6 This level of price change activity is similar to that found in other studiesof U S supermarket prices [Dutta Bergen and Levy 1995] although the pricereaction to cost changes can be signicantly more rigid depending on the natureof the cost changes the retailer faces [Levy Dutta and Bergen 1996]
QUARTERLY JOURNAL OF ECONOMICS794
ate some barriers to price change activity at these retail super-market outlets they are not large enough to prevent a signicantshare of prices to adjust
The paper is organized as follows In Section II we describethe data In Section III we discuss the price change process insupermarket chains and report absolute measures of menu costsIn Section IV we assess the effect of item pricing laws on menucosts In Section V we discuss the signicance of the menu costsWe end with conclusions and suggestions for future research
II DATA DESCRIPTION
The data come from a company that sells electronic shelf la-bel (ESL) systems These systems allow retailers to change theshelf prices electronically from a central computer (where pricechanges are actually done) via a wireless communication systemand thus reduce the physical costs and lead times currently asso-ciated with changing shelf prices In order to sell the product thecompany needed to validate what the existing costs of changingshelf prices were in supermarket chains ie the existing levelsof menu costs This company received access from corporate head-quarters at each of the ve chains in our sample to go to represen-tative stores and carefully record the exact steps involved in theprice change process These studies considered the entire pricechange process in each chain For this detailed work-ow sche-matics of each task in the price change process was developedObservations of the process were conducted in multiple stores ofthe chains (at least two representative stores for each chain) toverify its accuracy Information received from chainsrsquo pricing sys-tems in-store observations in-store counts and in-store timemeasurements (with a stopwatch) were used to determine thevolume of work performed in each step of the tasks weekly fre-quency of each step performed and the exact amount of time re-quired to perform one unit of the work After computing the totalhours per task this information was reconciled with the knowntotal hours spent each week This allowed for task level compari-sons for the existing and test process Each study required hun-dreds of man-hours to create The studies were conducted duringthe years 1991ndash1992
Although we believe the menu costs reported in this paperare representative of menu costs in the U S supermarket indus-try we should mention that they may be biased upward because
THE MAGNITUDE OF MENU COSTS 795
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
micro-based explanation for monetary nonneutrality Secondeven small menu costs may be sufcient to generate substantialaggregate nominal rigidity and large business cycles1 Conse-quently menu costs have received considerable attention in thetheoretical macroeconomics literature as many predictions gener-ated by traditional Keynesian and more recent new Keynesianmodels crucially depend on the existence of some form of pricerigidity2
Despite the theoretical importance of menu costs howeverlittle is known about their actual magnitude as the above quota-tion from Blinder succinctly reects Because of the practical dif-culty of measuring menu costs directly a common feature of theexisting empirical studies of menu costs is that they all provideindirect evidence3 Yet many authors including Blinder [1994]Kashyap [1995] and Slade [1996a] have emphasized the impor-tance of assessing the empirical relevance of menu costs at thelevel of individual rms For example according to Slade [1996ap 19] ldquoGiven the large number of theoretical papers that evalu-ate the implications of [price] adjustment costs obtaining directevidence that such costs are present seems crucialrdquo
Our primary contribution in this paper is providing directmeasures of menu costs at ve large U S retail supermarketchains Using a unique store-level data set we show that chang-ing prices in these establishments is a complex process requiringdozens of steps and a nontrivial amount of resources The menucosts reported in this study are made up of (1) the labor cost ofchanging shelf prices (2) the costs of printing and delivering newprice tags (3) the costs of mistakes made during the price changeprocess and (4) the cost of in-store supervision of the pricechange process4 We nd that the measurable components ofmenu costs for the four chains that are not subject to an item
1 See Akerlof and Yellen [1985] Mankiw [1985] Parkin [1986] Blanchardand Kiyotaki [1987] Caplin and Leahy [1991 1997] and Caplin [1993]
2 See for example Mankiw and Romer [1991] Sheshinski and Weiss [1993]Andersen [1994] Ball and Mankiw [1994] Romer [1996] and studies citedtherein
3 These studies include Sheshinski Tishler and Weiss [1981] Rotemberg[1982] Lieberman and Zilberfarb [1985] Carlton [1986 1989] Cecchetti [1986]Danziger [1987] Ball Mankiw and Romer [1988] Gordon [1990] Lach and Tsid-don [1992 1996] Blinder [1994] Eden [1994] Amano and Macklem [1995] Balland Mankiw [1995] Kashyap [1995] Warner [1995] Warner and Barsky [1995]and Slade [1996a 1996b]
4 We also discuss other components of menu costs including the costs ofmaking corporate level managerial price change decisions and provide some evi-dence on their approximate magnitude although we do not include these guresin the measures of menu cost we report
QUARTERLY JOURNAL OF ECONOMICS792
price law average $105887 annually per store In relative termsthese menu costs comprise 070 percent of revenues 352 percentof net margins and $052 per price change on average
Our second major contribution in this paper is providing evi-dence that these menu costs can form a barrier to price changeactivity at these chains offering direct support for the relation-ship between menu costs and store-level individual price rigidityWe present three types of evidence that these menu costs form abarrier to price change activity at these rms First we contrastthe price change activity of a chain that operates in a state withan item pricing law with the rst four chains that operate instates not subject to such laws Item pricing laws require that aseparate price tag be placed on each individual item sold (in addi-tion to the shelf price tag) We show that the average menu costper price change for the chain subject to the item pricing law is$133 over two and a half times the corresponding gure for theother four chains ($052) These larger menu costs lead to verydifferent levels of price change activity by these chains Specifi-cally the four supermarket chains that are not subject to itempricing laws on average change prices on 156 percent of the prod-ucts they carry each week In contrast the chain that is subjectto the item pricing law (and therefore faces higher menu costs)changes prices on only 63 percent of the products it carrieswhich is less than half the average of the other four chains
Second within the chain facing the item pricing law thereare 400 products that are exempt from this law and thereby facelower menu costs For these products the chain each weekchanges the prices of 21 percent of the products on average whichis over three times more frequently than for products subject tothe item pricing law Third we provide evidence from the super-market chains that the menu costs they incur form a barrier tocertain cost-based price adjustments Specically we show thatthe chains not subject to item pricing laws each week experiencecost increases on about 800ndash1000 products they sell Yet they ad-just prices of only about 70ndash80 percent of these products Theremaining 20ndash30 percent of the prices are not adjusted immedi-ately because the existing menu costs make the necessary priceadjustment unprotable Considering all three of these ndingstogether we conclude that menu costs can indeed affect pricechange activity at the level of the individual rmmdashoffering directevidence that these menu costs are relevant to marginal pricechange decisions Finally on a related macroeconomic issue we
THE MAGNITUDE OF MENU COSTS 793
provide empirical evidence which suggests that the price changeprocess in these supermarket chains has a strong time-dependent element
Relating our ndings to the existing theoretical models weconclude that the magnitude of the menu costs we nd is largeenough to be capable of having macroeconomic signicance Firstrecall that according to the studies of Akerlof and Yellen [1985]Mankiw [1985] Parkin [1986] and Caplin and Leahy [1997] evensmall menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Second when considered in the context of thetheoretical menu cost models of Blanchard and Kiyotaki [1987]and Ball and Romer [1990] we nd that the menu cost gureswe report are ldquonontrivialrdquo and their relative magnitudes cross theminimum theoretical threshold needed to form a barrier to priceadjustments
Although in this paper we provide direct measurements ofthe marginal costs associated with changing prices it should bementioned that there are still many aspects of menu costs we areunable to measure In particular we do not provide measure-ments of the marginal benets associated with changing priceswhich can be signicant [Lieberman and Zilberfarb 1985 She-shinski and Weiss 1977] At the local level this industry is ex-tremely competitive [Calatone et al 1989 Progressive GrocerNovember 1992 p 50 Chevalier 1995] In such a competitive in-dustry the benets of frequently changing prices can be high asunmatched price cuts or consumer perceptions of higher pricescan lead to signicant losses in sales5 This helps explain whydespite the magnitude of the menu costs we found we still ob-serve frequent weekly price change activity by the chains westudy For example stores change an average of 15ndash16 percent oftheir prices each week Also they seem to adjust the prices of70ndash80 percent of the products for which they experience cost in-creases6 Thus although we have evidence that the menu costswe report in this study clearly matter in the sense that they cre-
5 Also changing prices frequently can make it more difcult for customersto compare prices of branded items across supermarkets because of higher searchcosts [Carlton 1986] which is valuable for creating differentiation between retailoutlets [Bergen Dutta and Shugan 1996]
6 This level of price change activity is similar to that found in other studiesof U S supermarket prices [Dutta Bergen and Levy 1995] although the pricereaction to cost changes can be signicantly more rigid depending on the natureof the cost changes the retailer faces [Levy Dutta and Bergen 1996]
QUARTERLY JOURNAL OF ECONOMICS794
ate some barriers to price change activity at these retail super-market outlets they are not large enough to prevent a signicantshare of prices to adjust
The paper is organized as follows In Section II we describethe data In Section III we discuss the price change process insupermarket chains and report absolute measures of menu costsIn Section IV we assess the effect of item pricing laws on menucosts In Section V we discuss the signicance of the menu costsWe end with conclusions and suggestions for future research
II DATA DESCRIPTION
The data come from a company that sells electronic shelf la-bel (ESL) systems These systems allow retailers to change theshelf prices electronically from a central computer (where pricechanges are actually done) via a wireless communication systemand thus reduce the physical costs and lead times currently asso-ciated with changing shelf prices In order to sell the product thecompany needed to validate what the existing costs of changingshelf prices were in supermarket chains ie the existing levelsof menu costs This company received access from corporate head-quarters at each of the ve chains in our sample to go to represen-tative stores and carefully record the exact steps involved in theprice change process These studies considered the entire pricechange process in each chain For this detailed work-ow sche-matics of each task in the price change process was developedObservations of the process were conducted in multiple stores ofthe chains (at least two representative stores for each chain) toverify its accuracy Information received from chainsrsquo pricing sys-tems in-store observations in-store counts and in-store timemeasurements (with a stopwatch) were used to determine thevolume of work performed in each step of the tasks weekly fre-quency of each step performed and the exact amount of time re-quired to perform one unit of the work After computing the totalhours per task this information was reconciled with the knowntotal hours spent each week This allowed for task level compari-sons for the existing and test process Each study required hun-dreds of man-hours to create The studies were conducted duringthe years 1991ndash1992
Although we believe the menu costs reported in this paperare representative of menu costs in the U S supermarket indus-try we should mention that they may be biased upward because
THE MAGNITUDE OF MENU COSTS 795
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
price law average $105887 annually per store In relative termsthese menu costs comprise 070 percent of revenues 352 percentof net margins and $052 per price change on average
Our second major contribution in this paper is providing evi-dence that these menu costs can form a barrier to price changeactivity at these chains offering direct support for the relation-ship between menu costs and store-level individual price rigidityWe present three types of evidence that these menu costs form abarrier to price change activity at these rms First we contrastthe price change activity of a chain that operates in a state withan item pricing law with the rst four chains that operate instates not subject to such laws Item pricing laws require that aseparate price tag be placed on each individual item sold (in addi-tion to the shelf price tag) We show that the average menu costper price change for the chain subject to the item pricing law is$133 over two and a half times the corresponding gure for theother four chains ($052) These larger menu costs lead to verydifferent levels of price change activity by these chains Specifi-cally the four supermarket chains that are not subject to itempricing laws on average change prices on 156 percent of the prod-ucts they carry each week In contrast the chain that is subjectto the item pricing law (and therefore faces higher menu costs)changes prices on only 63 percent of the products it carrieswhich is less than half the average of the other four chains
Second within the chain facing the item pricing law thereare 400 products that are exempt from this law and thereby facelower menu costs For these products the chain each weekchanges the prices of 21 percent of the products on average whichis over three times more frequently than for products subject tothe item pricing law Third we provide evidence from the super-market chains that the menu costs they incur form a barrier tocertain cost-based price adjustments Specically we show thatthe chains not subject to item pricing laws each week experiencecost increases on about 800ndash1000 products they sell Yet they ad-just prices of only about 70ndash80 percent of these products Theremaining 20ndash30 percent of the prices are not adjusted immedi-ately because the existing menu costs make the necessary priceadjustment unprotable Considering all three of these ndingstogether we conclude that menu costs can indeed affect pricechange activity at the level of the individual rmmdashoffering directevidence that these menu costs are relevant to marginal pricechange decisions Finally on a related macroeconomic issue we
THE MAGNITUDE OF MENU COSTS 793
provide empirical evidence which suggests that the price changeprocess in these supermarket chains has a strong time-dependent element
Relating our ndings to the existing theoretical models weconclude that the magnitude of the menu costs we nd is largeenough to be capable of having macroeconomic signicance Firstrecall that according to the studies of Akerlof and Yellen [1985]Mankiw [1985] Parkin [1986] and Caplin and Leahy [1997] evensmall menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Second when considered in the context of thetheoretical menu cost models of Blanchard and Kiyotaki [1987]and Ball and Romer [1990] we nd that the menu cost gureswe report are ldquonontrivialrdquo and their relative magnitudes cross theminimum theoretical threshold needed to form a barrier to priceadjustments
Although in this paper we provide direct measurements ofthe marginal costs associated with changing prices it should bementioned that there are still many aspects of menu costs we areunable to measure In particular we do not provide measure-ments of the marginal benets associated with changing priceswhich can be signicant [Lieberman and Zilberfarb 1985 She-shinski and Weiss 1977] At the local level this industry is ex-tremely competitive [Calatone et al 1989 Progressive GrocerNovember 1992 p 50 Chevalier 1995] In such a competitive in-dustry the benets of frequently changing prices can be high asunmatched price cuts or consumer perceptions of higher pricescan lead to signicant losses in sales5 This helps explain whydespite the magnitude of the menu costs we found we still ob-serve frequent weekly price change activity by the chains westudy For example stores change an average of 15ndash16 percent oftheir prices each week Also they seem to adjust the prices of70ndash80 percent of the products for which they experience cost in-creases6 Thus although we have evidence that the menu costswe report in this study clearly matter in the sense that they cre-
5 Also changing prices frequently can make it more difcult for customersto compare prices of branded items across supermarkets because of higher searchcosts [Carlton 1986] which is valuable for creating differentiation between retailoutlets [Bergen Dutta and Shugan 1996]
6 This level of price change activity is similar to that found in other studiesof U S supermarket prices [Dutta Bergen and Levy 1995] although the pricereaction to cost changes can be signicantly more rigid depending on the natureof the cost changes the retailer faces [Levy Dutta and Bergen 1996]
QUARTERLY JOURNAL OF ECONOMICS794
ate some barriers to price change activity at these retail super-market outlets they are not large enough to prevent a signicantshare of prices to adjust
The paper is organized as follows In Section II we describethe data In Section III we discuss the price change process insupermarket chains and report absolute measures of menu costsIn Section IV we assess the effect of item pricing laws on menucosts In Section V we discuss the signicance of the menu costsWe end with conclusions and suggestions for future research
II DATA DESCRIPTION
The data come from a company that sells electronic shelf la-bel (ESL) systems These systems allow retailers to change theshelf prices electronically from a central computer (where pricechanges are actually done) via a wireless communication systemand thus reduce the physical costs and lead times currently asso-ciated with changing shelf prices In order to sell the product thecompany needed to validate what the existing costs of changingshelf prices were in supermarket chains ie the existing levelsof menu costs This company received access from corporate head-quarters at each of the ve chains in our sample to go to represen-tative stores and carefully record the exact steps involved in theprice change process These studies considered the entire pricechange process in each chain For this detailed work-ow sche-matics of each task in the price change process was developedObservations of the process were conducted in multiple stores ofthe chains (at least two representative stores for each chain) toverify its accuracy Information received from chainsrsquo pricing sys-tems in-store observations in-store counts and in-store timemeasurements (with a stopwatch) were used to determine thevolume of work performed in each step of the tasks weekly fre-quency of each step performed and the exact amount of time re-quired to perform one unit of the work After computing the totalhours per task this information was reconciled with the knowntotal hours spent each week This allowed for task level compari-sons for the existing and test process Each study required hun-dreds of man-hours to create The studies were conducted duringthe years 1991ndash1992
Although we believe the menu costs reported in this paperare representative of menu costs in the U S supermarket indus-try we should mention that they may be biased upward because
THE MAGNITUDE OF MENU COSTS 795
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
provide empirical evidence which suggests that the price changeprocess in these supermarket chains has a strong time-dependent element
Relating our ndings to the existing theoretical models weconclude that the magnitude of the menu costs we nd is largeenough to be capable of having macroeconomic signicance Firstrecall that according to the studies of Akerlof and Yellen [1985]Mankiw [1985] Parkin [1986] and Caplin and Leahy [1997] evensmall menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Second when considered in the context of thetheoretical menu cost models of Blanchard and Kiyotaki [1987]and Ball and Romer [1990] we nd that the menu cost gureswe report are ldquonontrivialrdquo and their relative magnitudes cross theminimum theoretical threshold needed to form a barrier to priceadjustments
Although in this paper we provide direct measurements ofthe marginal costs associated with changing prices it should bementioned that there are still many aspects of menu costs we areunable to measure In particular we do not provide measure-ments of the marginal benets associated with changing priceswhich can be signicant [Lieberman and Zilberfarb 1985 She-shinski and Weiss 1977] At the local level this industry is ex-tremely competitive [Calatone et al 1989 Progressive GrocerNovember 1992 p 50 Chevalier 1995] In such a competitive in-dustry the benets of frequently changing prices can be high asunmatched price cuts or consumer perceptions of higher pricescan lead to signicant losses in sales5 This helps explain whydespite the magnitude of the menu costs we found we still ob-serve frequent weekly price change activity by the chains westudy For example stores change an average of 15ndash16 percent oftheir prices each week Also they seem to adjust the prices of70ndash80 percent of the products for which they experience cost in-creases6 Thus although we have evidence that the menu costswe report in this study clearly matter in the sense that they cre-
5 Also changing prices frequently can make it more difcult for customersto compare prices of branded items across supermarkets because of higher searchcosts [Carlton 1986] which is valuable for creating differentiation between retailoutlets [Bergen Dutta and Shugan 1996]
6 This level of price change activity is similar to that found in other studiesof U S supermarket prices [Dutta Bergen and Levy 1995] although the pricereaction to cost changes can be signicantly more rigid depending on the natureof the cost changes the retailer faces [Levy Dutta and Bergen 1996]
QUARTERLY JOURNAL OF ECONOMICS794
ate some barriers to price change activity at these retail super-market outlets they are not large enough to prevent a signicantshare of prices to adjust
The paper is organized as follows In Section II we describethe data In Section III we discuss the price change process insupermarket chains and report absolute measures of menu costsIn Section IV we assess the effect of item pricing laws on menucosts In Section V we discuss the signicance of the menu costsWe end with conclusions and suggestions for future research
II DATA DESCRIPTION
The data come from a company that sells electronic shelf la-bel (ESL) systems These systems allow retailers to change theshelf prices electronically from a central computer (where pricechanges are actually done) via a wireless communication systemand thus reduce the physical costs and lead times currently asso-ciated with changing shelf prices In order to sell the product thecompany needed to validate what the existing costs of changingshelf prices were in supermarket chains ie the existing levelsof menu costs This company received access from corporate head-quarters at each of the ve chains in our sample to go to represen-tative stores and carefully record the exact steps involved in theprice change process These studies considered the entire pricechange process in each chain For this detailed work-ow sche-matics of each task in the price change process was developedObservations of the process were conducted in multiple stores ofthe chains (at least two representative stores for each chain) toverify its accuracy Information received from chainsrsquo pricing sys-tems in-store observations in-store counts and in-store timemeasurements (with a stopwatch) were used to determine thevolume of work performed in each step of the tasks weekly fre-quency of each step performed and the exact amount of time re-quired to perform one unit of the work After computing the totalhours per task this information was reconciled with the knowntotal hours spent each week This allowed for task level compari-sons for the existing and test process Each study required hun-dreds of man-hours to create The studies were conducted duringthe years 1991ndash1992
Although we believe the menu costs reported in this paperare representative of menu costs in the U S supermarket indus-try we should mention that they may be biased upward because
THE MAGNITUDE OF MENU COSTS 795
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
ate some barriers to price change activity at these retail super-market outlets they are not large enough to prevent a signicantshare of prices to adjust
The paper is organized as follows In Section II we describethe data In Section III we discuss the price change process insupermarket chains and report absolute measures of menu costsIn Section IV we assess the effect of item pricing laws on menucosts In Section V we discuss the signicance of the menu costsWe end with conclusions and suggestions for future research
II DATA DESCRIPTION
The data come from a company that sells electronic shelf la-bel (ESL) systems These systems allow retailers to change theshelf prices electronically from a central computer (where pricechanges are actually done) via a wireless communication systemand thus reduce the physical costs and lead times currently asso-ciated with changing shelf prices In order to sell the product thecompany needed to validate what the existing costs of changingshelf prices were in supermarket chains ie the existing levelsof menu costs This company received access from corporate head-quarters at each of the ve chains in our sample to go to represen-tative stores and carefully record the exact steps involved in theprice change process These studies considered the entire pricechange process in each chain For this detailed work-ow sche-matics of each task in the price change process was developedObservations of the process were conducted in multiple stores ofthe chains (at least two representative stores for each chain) toverify its accuracy Information received from chainsrsquo pricing sys-tems in-store observations in-store counts and in-store timemeasurements (with a stopwatch) were used to determine thevolume of work performed in each step of the tasks weekly fre-quency of each step performed and the exact amount of time re-quired to perform one unit of the work After computing the totalhours per task this information was reconciled with the knowntotal hours spent each week This allowed for task level compari-sons for the existing and test process Each study required hun-dreds of man-hours to create The studies were conducted duringthe years 1991ndash1992
Although we believe the menu costs reported in this paperare representative of menu costs in the U S supermarket indus-try we should mention that they may be biased upward because
THE MAGNITUDE OF MENU COSTS 795
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
the rm had an incentive to overestimate the magnitude of themenu costs in order to sell the ESL system We think howeverthat the menu cost measures we report in this paper are not sub-ject to signicant biases of this sort for a number of reasonsFirst the ESL people measured and documented all price changeactivities jointly with the supermarket employees using the wagegures provided by the supermarket management Second timeand motion measurements of the type used for measuring themenu costs we report here are routinely done by supermarketchains themselves in order to assess the efciency of their pricechange processes The supermarket managers compared theirgures to the ESL company gures and found them to be similarFurther these gures were presented to upper management ofthese chains and were found to be representative of their coststructures In fact the validity of the menu cost measures con-structed by the ESL company was never disputed If there wasany disagreement between the ESL company and the supermar-ket chains it was about the size of the savings the ESL systemwould provide not about the accuracy of the menu cost measure-ments7 Further we looked at these reports and searched for g-ures that could be biased upward There were a few such as lossof goodwill costs and inventory holding costs and to be on theconservative side we did not include them in our measures ofmenu costs Thus we only report gures for which we could seeno upward bias Finally note that the menu cost gures we reportare clearly biased downward because we were unable to measurein dollar terms several components of menu costs and thus theyare not included in our gures (see subsection III5 for details)
Table I displays some general information about the super-market chains we study their pricing strategy and informationabout the frequency of weekly price changes the stores under-take The chains involved in this study are all large U S super-market chains from different regions in the United Statesranging from the Northeast to the West Coast and operating anaverage of 400 stores each At the request of these retailers wewill keep the companies in this study anonymous but they areall large multistore chains that seem reasonably representativeof large supermarket chains currently selling in the United
7 Indeed four out of the ve chains included in our sample have purchasedESL systems three of the four actually purchased multiple systems (between twoand twenty systems) and Chain E is considering buying 50 more
QUARTERLY JOURNAL OF ECONOMICS796
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TABLE IGENERAL INFORMATION ON EACH SUPERMARKET CHAIN AND THEIR PRICE
CHANGE ACTIVITY
Chain Chain Chain Chain Average of Chain E (itemA B C D chains AndashD pricing law)
General pricingstrategya HL HL EDLP EDLP HL
Number of pricechanges perstore per week 4278 4316 3846 3223 3916 1578
of productsfor which priceschange in anaverage weekb 1711 1726 1538 1289 1566 631
a HL (HighLow) and EDLP (Every Day Low Price) refer to the general pricing strategy followed by theretail chain Under the EDLP strategy the retailer rsquos prices are low for extended periods of time and there-fore it will offer fewer promotional sales or discounts Under the HL pricing strategy in contrast the retail-errsquos prices are higher and the retailer tends to offer more frequent discounts through sales and promotionsSee the text for more details
b The share of products for which prices change on an average week is the ratio of number of pricechanges per store per week to 25000 The latter is the average number of products carried per store eachweek
States These chains are similar in the variety selection andquantity of the products they carry Supermarket chains of thistype make up $310146666000 in total annual sales which is863 percent of total supermarket chain sales in 1992 [Supermar-ket Business 1993] so the chains in our sample are representativeof a major class of the retail grocery trade
According to the second row in Table I the number of weeklyprice changes in Chains AndashD ranges from 3223 to 4316 for anaverage of 3916 per store8 The variation in the number of weeklyprice changes across the chains is due in large part to their choiceof pricing strategy Chains A and B follow a highlow (HL) pricestrategy while Chains C and D follow an everyday-low-pricestrategy (EDLP) Under the EDLP strategy the retailerrsquos pricesare low for an extended period of time and therefore it will offerfewer promotional sales or discounts Under the HL pricing strat-egy in contrast the retailerrsquos prices are higher and the retailertends to offer more frequent discounts through sales and promo-tions The pricing strategy therefore will have an effect on thefrequency of price changes observed In particular we would ex-
8 Since Chain E is subject to an item pricing law it is discussed separatelyin Section IV
THE MAGNITUDE OF MENU COSTS 797
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
pect EDLP stores to have less frequent price changes in compari-son to the HL stores Indeed according to Table I Chains A andB tend to have a higher number of weekly price changes (4278and 4316 respectively) than Chains C and D (3846 and 3223respectively)
Supermarkets of the size we study tend to carry around25000 different items on a regular basis9 The last row in thetable presents the share of the 25000 products for which the su-permarket chains change their prices in a period of one weekThe average share for the four chains is 1566 percent
III ABSOLUTE MEASURES OF MENU COSTS
There are four components of menu costs that we are able tomeasure in dollar terms10 These are (1) the cost of labor requiredto change the shelf price tags (2) the cost of printing and deliv-ering new price tags (3) the cost of mistakes made during theprocess of changing prices and (4) the cost of in-store supervisiontime spent on implementing price changes
III1 Costs of the Labor Required to Change Shelf Prices
Figure I displays an overview of the steps involved in chang-ing prices at Chain A There are three main components in thelabor costs incurred by the chains (a) labor cost of tag changepreparation (b) labor cost of the tag change itself and (c) laborcost of verifying whether the price changes are done correctlywhich include tag change verication in-store resolution of pricemistakes and zone and corporate resolution of price mistakes11
9 The supermarkets often have about 40000 UPC codes in their computerdatabase records but internal studies undertaken by the ESL company indicatethat the supermarkets usually carry no more than 25000 products at any giventime The extra UPC codes are for seasonal or promotional sizes and packages ofproducts and for discontinued products
10 In this paper we only report measures of the marginal cost of changingprices The costs of putting a price tag for the rst time and other costs thatwould be included in the average cost are not included in the gures we reportOnly when discussing Chain E which is the chain subject to item pricing lawsdid we face the issue of cost of pricing (at the rst time) versus the cost of changinga price Since there was no clear way of separating the two types of costs thereported cost gure ($44168 in the second paragraph of Section IV) was excludedaltogether from our calculations Also see footnote 13
11 For a detailed description of the entire price change process with ow-charts documenting the specic steps undertaken in this process and the exacttime period spent on each step see Levy Dutta Bergen and Venable [1997] Anappendix reporting computational details are contained in a working paper ver-sion of this paper which is available upon request
QUARTERLY JOURNAL OF ECONOMICS798
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
FIGURE IOverview of the Price Change Process
POS denotes Point of Sale and refers to the cash register or the database itis connected to or both For information on the various steps undertaken in eachstage of the price change process and the amount of the labor time spent on mosttime-consuming steps see Table II
Standard price tag changes which include the steps outlined onthe left-hand side of Figure I make up the majority of pricechanges in these chains Some price changes also require addi-tional price signs to be placed at different locations in the storesuch as on an end of aisle display or near the shelf12 These sign
12 These are usually related to the product being on promotion feature ad-vertising display or in-store sale such as ldquomanagerrsquos specialrdquo ldquotodayrsquos specialrdquoetc The steps involved in making sign changes are similar to price changes Themain differences are that (i) the time required for each of the steps in sign changes
THE MAGNITUDE OF MENU COSTS 799
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
changes add to the menu costs and they are outlined on theright-hand side of Figure I the sign change preparation actualsign changes and sign change verication boxes The bottom twoboxes in Figure I are additional menu costs related to the extrasteps taken in these stores to make sure the tag and sign changeshave been done correctly13
Table II lists each stage of the price change process the totalamount of time spent on each stage each week on average andthe number of tasks performed In addition the table identiessome of the main tasks performed in each stage as well as thefour most time-consuming tasks in each stage To compute thetotal labor time used in changing prices on a weekly basis wecombine the data collected through in-store time and motion ob-servations with information on the volume of products for whichprices are changed These weekly hours are multiplied by thewage rates (adjusted for fringe benets) of the employees used inthe price change process to get the total costs of labor required tochange prices
For the four supermarket chains in our study the total an-nual labor cost of changing the shelf price tags ranges from$40027 to $61414 per store for an average of $52084 (see therst row in Table III making up about 49 percent of the totalmenu costs on average The labor cost of changing the price signsrange from $16411 to $27955 per store for an average of$22183 (see the second row in Table III) making up almost 21percent of the menu costs on average Thus for the four super-market chains the total annual menu costs associated with thelabor required to change prices (shelf price tags and price signscombined) range from $62210 to $81703 per store for an annualaverage of $74267 This is the single largest component of themenu costs we report in this study making up about 701 percentof the total menu costs for these chains on average This shouldnot be surprising given that the most signicant portion of retail
are longer because they are in less standard locations and (ii) there are fewersign changes than standard shelf price tag changes
13 The menu cost measures we report do not include the cost of changingprices in cases where items are moved from shelf to shelf or where shelf space isreallocated by increasing the shelf space for some products at the expense of oth-ers However they do include the cost of pricing new products when they are rstintroduced While this could bias the menu cost measures upward since it reallycaptures the cost of pricing rather than the cost of changing price the size of thisbias is marginal due to the small number of new products For example accordingto ESL company executives the number of new products introduced at thesechains each week ranges from 20 to 100 approximately In comparison to the num-ber of products for which prices are changed each week (3223ndash4316) the biasis negligible
QUARTERLY JOURNAL OF ECONOMICS800
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TA
BL
EII
ST
AG
ES
OF
PR
ICE
CH
AN
GE
PR
OC
ES
ST
IME
SP
EN
TO
NE
AC
HS
TA
GE
EA
CH
WE
EK
AN
DT
HE
MA
INT
AS
KS
PE
RF
OR
ME
DA
TE
AC
HS
TA
GE
AT
CH
AIN
A
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
She
lfpr
ice
tag
216
824
11R
ecei
veta
gss
ort
bygr
ocer
ypr
oduc
eS
ort
byde
part
men
t29
39
chan
ge4
63
gene
ralm
erch
andi
sed
istr
ibu
teto
Dis
trib
ute
tode
part
men
ts6
89pr
epar
atio
nde
part
men
tsfo
rn
ight
crew
sor
tby
Sor
tby
effe
ctiv
eda
te10
63
effe
ctiv
eda
tes
ort
byai
sle
sepa
rate
Sor
tan
dse
para
teby
aisl
e39
18
byai
sle
Pri
ceta
g14
249
28
32S
elec
tan
dlo
cate
aisl
ess
ort
byL
ocat
eit
em48
33
chan
ge30
44
subc
omm
odit
ies
sele
ctan
dlo
cate
Com
pare
item
UP
CC
ode
159
7pr
oces
ssu
bcom
mod
itie
sse
lect
and
read
Rem
ove
old
pric
eta
g6
71ta
gsl
ocat
eit
ems
com
pare
UP
Cin
foIn
stal
lnew
pric
eta
g8
72(c
ode
quan
tity
siz
epr
ice)
not
eal
lm
ism
atch
esr
emov
eol
dta
gpu
tne
wta
gre
peat
the
proc
ess
for
all
prod
ucts
atal
lloc
atio
ns
Pri
ceta
g14
273
73
26S
ort
byai
sle
and
subc
omm
odit
ygo
toR
ead
item
from
repo
rt10
57
chan
ge30
49
aisl
ere
adit
emfr
omre
port
loc
ate
Loc
ate
item
onsh
elf
533
7ve
ri
cati
onit
emon
the
shel
flo
cate
pric
eta
gL
ocat
eth
epr
ice
tag
589
com
pare
pric
esc
ompa
reef
fect
ive
Com
pare
pric
es14
88
date
sch
eck
off
onre
port
not
em
ism
atch
es(p
rint
edh
andw
ritt
eno
rD
SD
tags
)m
ark
repo
rtto
orde
rm
issi
ngta
gs
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Pri
cesi
gn1
422
910
Rec
eive
sign
sso
rtby
groc
ery
Sor
tby
depa
rtm
ent
480
7ch
ange
030
pr
oduc
ege
ner
alm
erch
andi
se
Dis
trib
ute
tode
part
men
ts8
01pr
epar
atio
ndi
stri
bute
tode
part
men
tsfo
rn
ight
Sor
tby
effe
ctiv
eda
te8
18cr
ews
ort
byef
fect
ive
date
sor
tby
Sor
tan
dse
para
teby
aisl
e14
32
aisl
ese
para
teby
aisl
e
Han
dmad
e75
171
620
Go
toen
dof
aisl
eno
teit
ems
onN
ote
item
son
disp
lay
and
pric
e18
11
pric
esi
gn16
06
disp
lay
and
pric
ego
toai
sle
whe
reG
oto
aisl
ew
ith
disp
lay
item
s13
61
chan
gedi
spla
yit
emis
shel
ved
loca
teit
em
Loc
ate
item
onth
esh
elf
181
1pr
oces
sco
mpa
reth
eta
gan
ddi
spla
ypr
ice
Pre
pare
new
sign
san
ddi
scar
dol
d26
44
not
em
ism
atch
rem
ove
sign
wit
hw
rong
pric
epr
epar
ene
wsi
gnan
ddi
scar
dol
din
stal
lnew
sign
rep
eat
Pre
prin
ted
323
856
20G
etsi
gns
goto
aisl
elo
cate
exis
tin
gL
ocat
eex
isti
ngsi
gns
102
3pr
ice
sign
692
si
gnc
hec
kef
fect
ive
date
com
pare
Loc
ate
othe
rol
dsi
gns
940
chan
geta
gan
dsi
gnpr
ice
rem
ove
old
sign
Inst
alln
ewsi
gns
453
2pr
oces
s(t
ear
inha
lf)
inst
all
new
sign
C
ompa
read
and
shel
fpr
ice
tag
914
com
pare
tag
and
sign
pric
elo
cate
item
wit
hn
ewsi
gnn
ote
item
sno
tfo
und
repe
atg
etco
pyof
ad
com
pare
adan
dsh
elf
pric
esn
ote
mis
mat
ches
rep
eat
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
Pri
cesi
gn16
194
016
Get
wee
kly
adve
rtis
emen
tin
sert
go
Not
eit
ems
ondi
play
and
pric
e42
27
chan
ge3
46
toit
emdi
spla
ysc
heck
whe
ther
they
Com
pare
adan
ddi
spla
ypr
ices
580
veri
ca
tion
are
adve
rtis
edc
ompa
repr
ices
L
ocat
eit
emon
the
shel
f21
82
corr
ect
mis
mat
ches
go
toth
eai
sle
Com
pare
tag
and
disp
lay
pric
e5
37w
her
epr
oduc
tis
shel
ved
loca
teit
em
com
pare
tag
and
disp
lay
pric
es
rem
ove
wro
ngsi
gnp
repa
rene
wsi
gni
nsta
llne
wsi
gnr
epea
t
In-s
tore
261
818
29L
ook
up
onsy
stem
ch
eck
Loo
kup
onsy
stem
120
3re
solu
tion
of5
59
auth
oriz
atio
nta
gsi
gnd
isca
rdw
rong
Loc
ate
tag
orsi
gnor
both
226
9pr
oble
ms
tag
sign
n
dm
ake
and
inst
all
Inst
allt
agor
sign
orbo
th11
34
occu
rrin
gin
corr
ect
tag
sign
em
ail
toC
SC
ZS
C
Mak
eco
rrec
tion
sas
need
ed14
09
the
pric
em
ake
corr
ecti
ons
(Spe
ci
csde
pen
dch
ange
onth
ety
peof
prob
lem
eg
m
issi
ng
proc
ess
item
sta
gsm
ism
atch
betw
een
shel
fan
dsi
gnpr
ices
orbe
twee
nU
PC
info
and
shel
fta
g)
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TA
BL
EII
CO
NT
INU
ED
Tim
esp
ent
onea
chN
um
ber
ofM
ost
tim
e-co
nsum
ing
task
san
dth
eir
stag
e(i
nse
con
ds)
and
task
sin
shar
ein
the
tota
ltim
esp
ent
onth
est
age
Sta
geit
ssh
are
inth
eto
tal
stag
eM
ain
task
spe
rfor
med
inea
chst
age
(in
perc
ents
)
Zone
and
820
520
Det
erm
ine
whe
ther
itis
ast
ore
erro
rE
mai
lfr
omS
SCan
dco
rrec
tion
s43
88
corp
orat
e0
17
com
mun
icat
eto
ZSC
via
emai
lC
onso
lida
tefr
omzo
ne43
88
reso
luti
onof
cons
olid
ate
from
allz
ones
C
omm
uni
cate
toS
SC
via
emai
l7
50pr
oble
ms
com
mun
icat
eto
SS
Can
dpr
ice
Info
rmSS
Cab
out
the
corr
ecti
ons
293
occu
rrin
gin
inte
grit
yvi
aem
ail
dete
rmin
eth
epr
ice
wh
ethe
rit
iszo
ne
erro
rde
term
ine
chan
geth
ere
quir
edco
rrec
tion
com
mu
nica
tepr
oces
sto
CS
Can
dZS
Cr
esol
veth
epr
oble
m
Pri
ce9
007
612
Cu
stom
ern
otes
pric
em
ista
ke
Cus
tom
erte
lls
cash
ier
tag
pric
e17
41
disc
repa
ncy
192
ca
shie
rve
rie
sth
em
ista
kean
dC
ashi
erof
fers
one
item
free
784
and
scan
offe
rsth
elo
wer
pric
e(o
ron
eit
emC
ashi
erl
lspr
ice
disc
repa
ncy
form
130
6gu
aran
tee
free
ifth
elo
wer
pric
eis
not
acce
pted
SS
Cre
sear
ches
and
corr
ects
521
0re
fun
dpr
oces
sby
the
cust
omer
)ca
shie
rco
mpl
etes
pric
edi
scre
panc
yfo
rmS
SC
rese
arch
esan
dco
rrec
tsth
em
ista
ke(o
nth
esh
elf
orsc
anne
rda
taba
seor
both
)
CS
C
ZS
Ca
nd
SS
Cst
and
for
Cor
pora
teS
can
Coo
rdin
ator
Zon
eS
can
Coo
rdin
ator
an
dS
tore
Sca
nC
oord
inat
orr
espe
ctiv
ely
For
am
ore
deta
iled
dis
cuss
ion
ofth
epr
ice
chan
gep
roce
sss
ee[L
evy
Ber
gen
D
utt
aan
dV
enab
le19
97]
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TAB
LE
III
ES
TIM
AT
ES
OF
TH
EA
NN
UA
LM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
S)
Ch
ain
Cha
inC
hain
Cha
inA
vera
geof
Cha
inE
Men
uco
stco
mpo
nen
tA
BC
Dch
ains
AndashD
(ite
mpr
icin
gla
w)
Lab
orco
stof
pric
ech
ange
s61
414
531
4940
027
537
4852
084
529
44(4
92
)L
abor
cost
ofsi
gnch
ange
sa16
411
221
8322
183
279
5522
183
221
83(2
09
)C
osts
ofpr
inti
ngan
d4
110
100
183
048
687
96
014
764
4de
live
ring
pric
eta
gs(5
7
)
Mis
take
cost
sb19
135
205
9320
692
201
4020
140
207
99(1
90
)In
-sto
resu
perv
isio
nco
stsc
424
16
692
546
65
466
546
65
466
(52
)
Tota
lann
ual
men
uco
st10
531
111
263
591
416
114
188
105
887
109
036
per
stor
e(1
00
)
aT
he
labo
rco
sts
ofsi
gnch
ange
sw
ere
not
repo
rted
for
Ch
ain
sB
C
an
dE
an
dso
we
use
inst
ead
the
aver
age
ofC
hai
ns
Aan
dD
b
Th
em
ista
ke
cost
sw
ere
not
rep
orte
dfo
rC
hai
nD
an
dso
we
use
inst
ead
the
aver
age
mis
take
cost
sof
Ch
ain
sA
B
an
dC
c
Th
ein
-sto
resu
perv
isio
nco
sts
wer
en
otre
por
ted
for
Ch
ain
sC
D
and
Ea
nd
sow
eu
sein
stea
dth
eav
erag
eof
Ch
ain
sA
and
B
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
grocery operating expenses is labor costs [Hoch Dreze andPurk 1994]14
III2 Costs of Printing and Delivering New Price Tags
There are direct costs associated with printing and deliv-ering the price and sign tags The order must be recorded andprocessed at the chain sent to the printer recorded and pro-cessed at the printer printed packaged and then delivered toeach store The cost per tag is usually quite low $0017 per tagat Chain A which includes stock costs of $00118 per tag impres-sion and data center operator costs of $00037 per tag and mailroom handling costs of $00010 per tag There are however manyprice changes undertaken each week In total the costs of print-ing and delivering the price and sign tags range from $3048 to$10018 averaging $6014 per store per year (see Table III)These costs comprise less than 6 percent of the total menu costswe report
III3 Costs of Mistakes Made in the Process of Changing Prices
Despite the labor put into checking to make sure that theprice changes are done correctly there are still many price mis-takes that are not caught until customers discover them through-out the week and these mistakes impose costs on the chainClearly the costs associated with these errors must be consideredwhen deciding whether to change prices or not and therefore area relevant dimension of menu costs These mistakes can occur sooften that they were a feature of a Dateline segment on U Ssupermarket chains [NBC April 1992] and an article in MoneyMagazine [April 1993] Both the Money Magazine article andGoodstein [1994] report that on average 10 percent of the prod-ucts they examined had price mistakes A more recent study bythe Federal Trade Commission [1996] reports a total error rate ofabout 5 percent
The menu costs associated with these mistakes include lost
14 Our measure of labor cost may overstate the true costs of changing pricesif supermarkets hoard labor to save hiring and ring costs However this is notlikely to be the case for several reasons First the labor costs of changing priceswe report are based on actual measurements of the minimum amount of time andlabor required to accomplish the task rather than on the number of employeeshired to change prices at the store Second the adjustment in the amount of laboris usually done through hours worked which makes cost of hiring and ring lessrelevant And third the workers on the oor are routinely moved from task totask according to the need These tasks include stocking cleaning price changingcustomer service etc The workers employed by supermarket chains are alwaysbusy and so the opportunity cost of changing price is not zero Therefore laborhoarding is not likely to be an important factor in our measurements
QUARTERLY JOURNAL OF ECONOMICS806
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
cashier time to correct the errors scan guarantee refunds andinventory mistakes associated with incorrect price tags15 Lostcashier time is measured here in terms of wage payments Scanguarantee refunds are additional price reductions beyond the er-ror or additional items given away for free because of the mis-take Finally the inventory mistakes costs we report include onlythe cost of stockouts which occur when shelf price is lower thanintended
The mistake costs were available at Chains A C and D andthey range from $19135 to $20692 for an average of $20140 perstore annually (see Table III) These costs are the second largestcomponent of menu costs in our study comprising about 19 per-cent of the total
III4 Costs of In-Store Supervision Time Spent on ImplementingPrice Changes
Managers at the store level spend time overseeing imple-menting and troubleshooting the price change process OnlyChains A and B had a measure of the menu costs associated withthis in-store supervision time and both of these came from a self-reported number of hours spent on changing prices by the manag-ers at these chains The hours spent on changing prices were thesame across the chains approximately ve hours per week Thusthe menu costs for in-store supervision time came to $4241 and$6692 per store annually for Chains A and B respectively (seeTable III) These costs comprise less than 6 percent of the totalmenu costs we report in this paper However notice that thesemeasures do not include the cost of the management time spenton price change decisions made at corporate headquarters whichis discussed next
III5 Components of Menu Costs We Are Unable to Measure inDollar Terms
Our data set does not contain the exact dollar measures ofsome menu cost components One such component is the cost ofcorporate management time spent on price change decisions16 In
15 Other likely costs of price mistakes that we do not consider explicitlybecause we are unable to measure them in dollar terms are legal problems (whenthe scanner price is higher than the shelf price) loss in customer goodwill anddecreased protability (when the scanner price is lower than the shelf price)
16 It has been suggested that the cost of managerial decisions is one of themost important components of menu cost See for example Ball and Mankiw[1994] Kashyap [1995] and Meltzer [1995] Since the ESL system was not de-
THE MAGNITUDE OF MENU COSTS 807
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
the supermarket chains we study prices are generally set at cor-porate headquarters in a weekly meeting where the manager incharge of setting prices looks at a variety of information including(a) any manufacturer wholesale price changes promotions andother related issues (b) past sales for this product and (c) com-petitorsrsquo prices Based on this information and on discussionswith other managers the price-setting manager decides whetherto change prices and if so by how much
To estimate the magnitude of these managerial componentsof the menu costs consider the following In an average chainthere is at least one executive of merchandising who devotesmost of hisher time to pricing decisions In addition there areup to three senior managers who would deal with pricing andto whom category managers report There are also ten-to-twelvecategory managers who are responsible for setting prices on allthe products within their category An additional two-to-fourpeople spend full time handling the implementation of pricechanges across retail outlets coordinating the printing and deliv-ery of price tags and handling pricing problems in the systemAnother ve-to-seven workers gather the data on competitorsrsquoprices and analyse both the competitorsrsquo data and the storersquos ownscanner data to put it in a form useable by managers makingpricing decisions Thus in total there are about 21ndash27 peopleworking at the corporate headquarters on price change decisionsAssuming $150000 as the average annual salary of the executiveof merchandising and senior managers $100000 for categorymanagers and $50000 for the rest the chainwide managerialcost falls in the neighborhood of $23ndash$29 million a year whichseems a substantial amount However note that these pricing de-cision are made for the entire chain and therefore the costs perstore are signicantly less especially for the larger chains Forexample using $29 million as the upper bound the additionalannual menu cost per store for Chains AndashD which on averageoperate about 400 stores each averages about $7250 which ismuch lower than expected and is due to the centralization of theprice change decisions in the chains we study17
signed to save the costs of corporate headquarter managerial time spent on pricechange decisions the ESL company did not measure this component of menucosts The estimates reported in this section are based on the information receivedfrom the ESL company executives
17 A decentralization of the price change process say by allowing the store-level managers to make price change decisions can change these gures tremen-dously For example consider the following thought experiment conducted using
QUARTERLY JOURNAL OF ECONOMICS808
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
Our menu cost measures also do not include the cost ofchanging prices of direct store delivery (DSD) products Theseproducts are almost completely handled by manufacturers in-cluding stocking monitoring inventories and setting and chang-ing prices18 We have data on the weekly frequency of pricechanges of DSD products in Chain E In an average week thereare 174 price changes of DSD products which is about 10 percentof the supermarketrsquos total weekly price changes Using the costof changing the price of a regular product $133 (see Section IVfor details) the annual cost of changing prices of the DSD prod-ucts in this chain roughly equals $1203419 Our menu cost mea-sures do not include these gures since we do not have similardata for the other chains20
III6 Total Menu Costs
The total annual menu costs reported for each chain perstore are listed in the last row of Table III As the table indicatesthe menu costs range from $91416 to $114188 for an average of$105887 per store per year21 Note that these menu cost mea-
the average annual per store gures for Chains AndashD Suppose that the supermar-ket chains decentralize the price change process by hiring for each store only one-quarter of the price change managing team while at the same time they com-pletely eliminate the corporate level team The resulting annual menu cost perstore would be $830887 ( 5 $105887 1 $725000) which would dwarf any of themeasurable menu cost gures we report in this study Even if the average storeonly hired just the merchandising manager at the annual cost of $150000 itwould still incur annual managerial cost of about $255887 ( 5 $105887 1$150000) In other words such a trivial decentralization would double the store-level menu costs This would indeed make the cost of price change decisions themost important component of menu cost
18 DSD products usually are high-volume fast-moving or perishable prod-ucts such as milk soda eggs bread dairy snacks etc In the chains we studyabout 10ndash20 percent of the products are of a DSD type but that share may reachas much as 40 percent of the products carried [Direct Store Delivery Work Groupet al 1995]
19 The process of changing prices of DSD and of regular products is similarBut with DSD products there are additional costs of the time spent on driving tothe store parking and setting up the price change process at each store
20 Our measures of menu cost also do not incorporate the cost of informingconsumers about price changes which can take a variety of forms such as newspa-per ads and inserts TV and radio ads in-store promotion signs etc Finally wehave no data on lost customer goodwill and damaged reputation caused by dis-crepancies between price tag and cash register [Okun 1981 Carlton and Perloff1994 Haddock and McChesney 1994]
21 The variation in the menu costs across the chains is mostly due to wagerate and labor-efciency variations Also note that since the supermarket chainsin our sample are similar in the size of their stores carry similar sets of productsand follow similar processes of price change and price change decisions it is rea-sonable to believe that the chains incur similar types of costs Therefore as notedunderneath Table III these menu cost measures are computed by replacing theunreported gures by the averages of the available values from the other chains
THE MAGNITUDE OF MENU COSTS 809
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
sures include only the components we could accurately measurein dollar terms which are discussed in subsections III1ndashIII4
IV ITEM PRICING LAWS AND MENU COSTS
In this section we provide evidence on the menu costs for anadditional chain (Chain E) which unlike the other four chainsoperates in a state with an item pricing law The most importantaspect of item pricing laws is that they require a price tag on eachindividual item sold not just on the shelf which is what the otherfour chains in our sample do For example the item pricing lawin Connecticut requires that the grocers ldquoshall mark or cause tobe marked each consumer commodity which bears a UniversalProduct Code with its retail pricerdquo 22 From a menu cost perspec-tive the requirement of posting prices on every item (in additionto the shelf price tag) introduces additional costs in the processof changing prices
Although most of the steps involved in changing prices arethe same for both types of chains stores that are subject to itempricing laws have to undertake additional steps to obey this law23
The labor cost component of menu costs in Chain E totals $75127a year of which $49710 is the cost of the labor time spent onchanging the prices of individual items $3234 is the cost of thelabor time spent on shelf tag replacements and $22183 is thecost of the labor time spent on sign changes (see Table III)24 Anadditional $44168 is spent annually putting prices on new itemsas they are brought to the shelves Although we do not includethis last gure in our menu cost measures because we do not haveinformation on how much of it is due to price change activity andhow much due to just pricingmdashthese costs are clearly a directconsequence of the item pricing law The printing and deliverycost of price tags and price signs are $7644 and the mistake
22 Public Act No 75ndash391 An Act Concerning Pricing of Consumer Merchan-dise [Public Acts of the State of Connecticut 1975 Vol 1 p 390]
23 These steps which are in addition to the regular steps undertaken toreplace price tags on the shelves include (1) obtaining item price tags (2) settingup workstations (3) locating the product (4) removing an item from the shelf (5)removing old price tag (6) setting marking gun (7) applying new price tag and(8) returning the item back to the shelf
24 Since we do not have a measure of the cost of labor time spent on signchanges and the cost of in-store managerial supervision time for this chain weuse the average of the chains that report them (A and D) Note also that thehourly wage rate of $907 paid by Chain E is lower than the hourly wage of about$1400ndash$2000 paid by Chains AndashD
QUARTERLY JOURNAL OF ECONOMICS810
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
25 Further note that the chains with smaller menu costs are likely tochange their prices more frequently which suggests that the gures of the totalannual menu costs understate the differences between Chain E and the otherchains Indeed despite the large difference in the menu cost per price change thetotal annual menu costs per store for Chain E ($109036) are more comparable tothose of Chains AndashD ($105887)
26 In calculations that follow we use industry averages because (1) thechains did not share this proprietary information with us and (2) we were re-quired to keep the identity of these chains condential as some of these numbersare detailed enough to enable some readers to identify the chains under study
costs are $20799 These gures along with the amount of $5466for the cost of in-store managerial supervision time yield totalannual menu costs of $109036 per store
Although the magnitude of the total annual menu costs seemsimilar to the other four chains note that the average weeklyfrequency of price changes in Chain E is only 1578 which is only403 ( 5 15783916) percent of the price changes made by theother four chains Thus the average menu cost per price changefor Chain E is $133 (see Table IV last row) In contrast the corre-sponding gures for Chains AndashD average $052 Hence the menucosts per price change incurred by the supermarket chain facingitem pricing laws are more than two and a half times the amountincurred by chains that are not subject to this law25 Overallthese ndings suggest that legal restrictions of the type of itempricing laws can have a signicant impact on the menu costs in-curred by sellers
V SIGNIFICANCE OF THE MENU COSTS
The next natural question to ask is how important are thesecosts To address this question we (1) provide several relativemeasures of the menu costs (2) discuss the effect of the menucosts on supermarketsrsquo price change activity and (3) discuss mac-roeconomic implications by relating our ndings to the existingtheoretical models of menu costs In each case we provide addi-tional evidence to help assess the importance of these menu costs
V1 Relative Measures of the Menu Costs
In order to assess the relative magnitude of the menu costsreported in Section IV we now present the menu cost gures rela-tive to the average store-level revenues and net prot margins26
In addition we present the menu cost gures per price changeThe annual average revenues of a large U S supermarket
chain of the type and size included in our sample is $15052716
THE MAGNITUDE OF MENU COSTS 811
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TA
BL
EIV
RE
LA
TIV
EM
EA
SU
RE
SO
FM
EN
UC
OS
TS
PE
RS
TO
RE
FO
RE
AC
HC
HA
IN(I
N19
91ndash1
992
DO
LL
AR
SO
RIN
PE
RC
EN
T)
Rel
ativ
em
easu
reof
Cha
inC
hain
Cha
inC
hai
nA
vera
geof
Cha
inE
men
uco
sts
AB
CD
chai
ns
AndashD
(ite
mpr
icin
gla
w)
Tota
lann
ual
men
uco
st($
)10
531
111
263
591
416
114
188
105
887
109
036
MC
rev
enu
esa
()
070
075
061
076
070
072
MC
ope
rati
ng
expe
nses
b(
)3
113
322
703
373
133
22M
Cg
ross
mar
gin
c(
)2
802
992
433
032
812
90M
Cn
etm
argi
nd
()
350
374
304
379
352
362
MC
per
prod
uct
carr
iede
($)
421
450
366
457
423
436
MC
per
item
sold
f($
)0
0119
001
270
0103
001
290
0119
001
23M
Cpe
rpr
ice
chan
geg
($)
047
050
046
068
052
133
Th
en
otes
belo
wpr
ovid
eth
e
gure
su
sed
inco
mp
uta
tion
sM
Cst
ands
for
tota
lan
nu
alm
enu
cost
See
text
for
mor
ede
tail
sa
Th
ean
nu
alre
ven
ues
are
$15
052
716
per
stor
eon
aver
age
[Su
perm
ark
etB
usi
nes
s19
93p
52
]b
Th
ean
nu
alop
erat
ing
expe
nse
sar
e$3
386
861
per
stor
eon
aver
age
base
don
225
perc
ent
ofre
ven
ues
[Hoc
hD
reze
an
dP
urk
1994
]c
Th
ean
nu
algr
oss
mar
gin
is$3
763
179
per
stor
eon
aver
age
base
don
25pe
rcen
tof
reve
nu
es[H
och
Dre
zea
nd
Pu
rk19
94S
upe
rma
rket
Bu
sin
ess
1993
]d
Th
ean
nu
aln
etm
argi
nis
$301
054
per
stor
eon
aver
age
base
don
2pe
rcen
tof
reve
nu
es[M
ontg
omer
y19
94]
eM
Cp
erpr
odu
ctca
rrie
dis
com
pute
das
ara
tio
MC
toth
eav
erag
en
um
ber
ofp
rodu
cts
carr
ied
per
stor
e(2
500
0)
fM
Cp
erit
emso
ldis
com
pute
das
the
rati
oof
MC
(re
ven
ue
aver
age
pric
ep
erit
emso
ld)
Th
eav
erag
epr
ice
per
item
sold
is$1
70
See
not
ea
abov
efo
rre
ven
ue
info
rmat
ion
Not
eth
ed
iffe
ren
cebe
twee
nn
um
ber
ofp
rod
uct
sca
rrie
dan
dn
um
ber
ofit
ems
sold
As
anex
ampl
ea
Tar
tar
Con
trol
Cre
st8
ozw
ould
beco
nsi
dere
da
prod
uct
carr
ied
and
300
un
its
ofth
emso
ldp
erye
arw
ould
beco
nsi
dere
dn
um
ber
ofit
ems
sold
g
MC
per
pric
ech
ange
isco
mpu
ted
as(M
C5
2)(
nu
mbe
rof
pric
ech
ange
sw
eek
)w
her
en
um
ber
ofpr
ice
chan
ges
wee
kis
take
nfr
omT
able
I
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
per store27 According to the second row of Table IV the ratio ofmenu costs to revenues for Chains AndashD ranges between 061ndash076percent averaging 070 percent28 We relate the size of thesegures to the existing theoretical menu cost models in sub-section V3
Net prot margin which measures the revenues minus allcosts is approximately 1ndash3 percent of revenues for these chains[Montgomery 1994] so we use 2 percent as a working averageThe reason for the low prot rate in this industry is the intensecompetition [Calatone et al 1989] especially at the regional level[Chevalier 1995] which has been increasing since the early 1990s[Progressive Grocer November 1992 p 50] It follows that theaverage store protability for these chains equals $301054 peryear Therefore the ratio of total menu cost to net margin forChains AndashD ranges between 304ndash379 percent for an average of352 percent Thus the menu costs we report in this study repre-sent a signicant share of supermarketsrsquo prots
Another way to look at the menu cost gures we report is toexpress them relative to the frequency of price changes In thelast row of Table IV we present the menu cost gures per pricechange for each chain As the table indicates the cost of changinga price in Chains AndashD ranges between $046ndash$068 for an averageof $052 For Chain E the gure is substantially larger $133 perprice change29 These gures are lower than the estimated cost of
27 This is the average of two different estimates $12945432 and$17160000 The source of the rst gure is internal record of a supermarketchain of the type and size studied here The second gure comes from Supermar-ket Business [1993 p 52]
28 We also measured the menu costs in terms of controllable operating ex-penses (which are the portion of costs that the retailer has direct control over andtherefore are often the costs that the retailer is most focused on managing) andgross margins (which measure the supermarket revenues minus direct cost of theproducts it sells) We nd that the menu costs comprise 313 percent of controlla-ble operating expenses and 281 percent of gross margins at these chains on aver-age Finally if we measure the menu costs relative to the number of productscarried per store (about 25000) then we nd that the menu costs per productaverage $423 for Chains AndashD See Table IV and its footnotes for details
29 We can also try to express the menu cost gures relative to the numberof individual items sold by these chains each year (Note the difference betweennumber of products carried and number of items sold As an example a TartarControl Crest 8oz would be considered a product carried and 300 units of themsold per year would be considered number of items sold) Using $170 as the aver-age price of all the products sold we nd that the menu cost per item sold for thefour chains is in the range of $00103ndash$00129 for an average of $00119 (see TableIV second to last row) To see the relative magnitude of these gures note thataccording to Berkowitz Kerin and Rudelius [1986 p 319] the goal of the super-market chains of the type we study ldquo is to make 1 penny of prot on each dollarof salesrdquo Thus menu cost per item sold when adjusted for the average price
THE MAGNITUDE OF MENU COSTS 813
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
$200ndash$300 per price change reported by Slade [1996a] Thereare several possible reasons for this (1) our menu cost gures arebased on actual measurements of the resources that go into theprice change process whereas she estimates menu costs econo-metrically as model coefcients using a mix of store-level priceand aggregate cost data (2) we cover 25000 products that thesupermarkets carry rather than a single product category (3) ourmenu cost gures do not include all components of menu costsand (4) there could be differences in wage rates that may be im-portant given the signicance of the labor cost component inmenu costs
V2 The Effect of Menu Costs on the Price Change Activity of theSupermarkets
In this section we present evidence which suggests that thesemenu costs may be forming a barrier to price change activity Tobegin with consider the effect of the item pricing law on the pricechange activity of Chain E The data on the weekly frequency ofprice changes in each chain are displayed in the last two rows ofTable I According to these gures the average weekly frequencyof price changes in Chain E is only 1578 Thus on average ChainE changes the prices of only 631 percent of its products eachweek In contrast the average weekly frequency of price changesat Chains AndashD ranges from 3223 to 4316 for the four-chain aver-age of 3916 price changes weekly So Chains AndashD change priceson 1289 to 1726 percent of their products each week yielding afour-chain average of 1566 percent Thus Chain E which facesthe item pricing law changes prices only about one-third timesas frequently as do Chains AndashD30
Moreover within Chain E there are 400 products that areexempt from the item pricing law and thereby face lower menucosts We nd that there are an average of 83 weekly pricechanges for these products yielding 21 percent of these productschanging prices So within Chain E they change prices over threetimes as frequently for products with lower menu costs than for
roughly equals one-half of their target net prot per item which seemssubstantial
30 However note that our comparison of the two types of chains may not becompletely ceteris paribus because these stores are located in different states andtherefore there may be other chain-specic factors (in addition to item pricinglaws) that distinguish these stores We do not have any specic information onthese differences We do know however that the stores in these chains are similarin size and carry similar sets of products
QUARTERLY JOURNAL OF ECONOMICS814
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
the products with higher menu costs Note that this nding is onthe price change activity within the same chain offering almost anatural experiment on the impact of menu costs on the chainrsquospricing behavior Hence we provide evidence both across chainsas well as across products within a chain that as the costs ofchanging price go up the frequency of price changes goes downThus the menu costs we nd are signicant enough to affect thechainrsquos price change practice
We also have additional evidence from Chains A B and Dabout these menu costs being a barrier to certain price changesand it is summarized in Table V According to the Price Manage-ment Department of Chain A the chain experiences cost in-creases on 860 products in an average week to which it wouldlike to react with a price increase31 However on average 22 per-cent of these price increases are not implemented immediatelybecause the cost of changing prices for these products is too highto make it economically worthwhile Figures of the same magni-tude were reported by other chains For example Chain D experi-ences cost increases for 961 products in an average week Out ofthese the chain adjusts prices of only 633 products The re-maining 34 percent of the prices are left unadjusted because ofthe menu costs Similarly Chain B nds it economically ineffi-cient to adjust prices of 508 ie 30 percent of its products eachweek because of menu costs This provides additional evidencethat the menu costs incurred by these chains are preventing priceadjustments to some costs changes leading to price rigidities ofthe products involved32 Note that although we do not have simi-lar data for Chain E we would expect signicantly lower num-bers on this measure at that chain These gures also suggest anupper bound on the benet of complete price exibility under thecurrent pricing practice of weekly price adjustments Howeverif new technologies (eg ESL systems) and new pricing prac-tices (eg a decentralization of the price change decisions) are
31 Our data contain information only on the frequency of cost increasesAlthough it is more than likely that the supermarkets are experiencing cost de-creases as well our data set contained no information on such decreases
32 Menu costs may even be playing a role in the observed movement towardpricing strategies that rely on fewer price changes such as EDLP [Blattberg andNeslin 1989 Lattin and Ortmeyer 1991 Marketing News April 13 1992 p 8]According to Progressive Grocer [November 1992 p 50] ldquoA growing number ofoperators say they have switched from high-low pricing [to EDLP] They cite theinefciencies of making frequent price changes rdquo Similarly Hoch Dreze andPurk [1994 p 16] state that EDLP lowers operating costs by lowering ldquo in-store labor costs because of less frequent changeovers in special displaysrdquo
THE MAGNITUDE OF MENU COSTS 815
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TABLE VWEEKLY FREQUENCY OF COST-BASED PRICE ADJUSTMENTS NOT IMPLEMENTED
BECAUSE OF MENU COSTS
Chain Chain ChainA B D
Number of products for 860 1693 961which costs increase in anaverage week
Number of price 671 1185 633adjustments implemented (78) (70) (66)
Number of price 189 508 328adjustments not (22) (30) (34)implemented because ofmenu costs
Chains C and E did not report these data
adopted allowing a fundamental change in the way the pricingmechanism is currently set the managers could consider morefrequent price change activity (eg multiple times each week) asmenu costs go down
In this paper we measure the menu costs at the chainsrsquo cur-rent level of price change activity It may be worthwhile to specu-late on the shape of the cost function relating menu costs to thefrequency of price changes by assessing how these costs wouldchange if the price change activity were to increase or to decreaseIt seems likely that many of the costs we report in this paper will(approximately) change linearly with additional price changeswithin the current range of price changes the chains are under-taking (plus or minus perhaps a thousand per week) For ex-ample the labor costs associated with changing a shelf price tagcosts of verifying price changes and the costs of mistakes oc-curring during this process all increase linearly with the fre-quency of price changes This is because most of the time-consuming steps involved in the price change process must berepeated each time an additional shelf price tag is changed (seeTable II) Further we were unable to nd many tasks that gener-ated signicant returns to scale It is unclear what cost savingsare available if the price change activity drops substantially be-
QUARTERLY JOURNAL OF ECONOMICS816
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
low the levels it is at now Clearly if price changes were nearzero or done less frequently than weekly (say monthly) therecould be major differences in these costs
What is less clear is how these costs would change at moreextreme levels of price change activity With less frequent pricechanges the menu costs could probably be reduced signicantlyalthough the cost of deciding what prices to set initially and thecost of putting the initial shelf price tags would still remain as alower bound of the menu cost However for achieving more priceexibility by changing prices more frequently (ie multiple timeseach week) substantial changes must be undertaken At presentthe supermarket chains are set up to make the majority of theirprice changes on a weekly basis with the appropriate systems inplace to make this process most efcient For example in orderto save costs and to have higher quality price tags (eg withcolor more details greater clarity glossy etc) the chains oftensend new price tag requests to a printing shop which takes threedays to print and deliver the labels to each store Then giventhe large number of price changes taking place and the goals ofminimizing customer disruptions and labor costs (such as over-time pay) and coordinating with advertised price promotions theprices are changed in the store over a two-to-three-day period (seeTable VI) The combination of these schedule and built-in lags isacceptable under the current practice of changing prices weeklybut would have to be adjusted dramatically to allow for signifi-cantly more frequent price changes on a regular basis
Perhaps even more imposing are the costs of changing themanagerial costs associated with the current weekly system ofprice changes The process of pricing at this organizational levelis not a self-contained activity taking place in isolation from otheroperations of the supermarket management Rather it is a partof a larger system of business decisions and operations The cur-rent process of operations (such as data collection and theiranalysis management meetings coordination of the decisionsacross functional areas etc) is set up to function on a weeklybasis33 Further these management accounting and logisticssystems are currently built around a tremendous amount of in-
33 For example the data are analyzed and presented to managers on Mon-day and Tuesday with meetings set for Thursday and Friday to make pricingdecisions for the next week in coordination with all other business functions ofthe chain
THE MAGNITUDE OF MENU COSTS 817
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
TABLE VIWEEKLY SCHEDULE OF PRICE CHANGES BY TYPE OF MERCHANDISE IN CHAIN A
Number of products for Time of the week whenType of merchandise which prices change the prices are changed
General merchandise 72 Saturday nightadvertised (17)Grocery 2100 Sunday night
(491)Market (produce) 171 Sunday night
(40)General 1853 Monday nightmerchandise (433)Grocery advertised 82 Tuesday night
(19)
Total number of 4278weekly price changes (100)
formation to be analyzed for thousands of products34 Accommo-dating more frequent price changes on a regular basis wouldrequire a radical adjustment of many managerial substructuresand units at both corporate and store levels which could involvecostly investment in restructuring and retraining the existing la-bor force as well as in new physical and human capital Thus thecost function relating the menu costs to the frequency of pricechanges would be approximately linear from zero to roughlyabout 5000 price changes per week With higher frequency ofprice changes under the current weekly pricing practice themenu costs are likely to increase dramatically perhaps by asmuch as ten-to-twenty times according to the ESL executives35
34 For example under the current system a chainwide category manageroperating at the corporate headquarters needs to study and assess numerouspieces of detailed information such as competitorsrsquo price and sale information fromthe last week the past weekrsquos sales at the own chain manufacturersupplier pro-motions wholesale price reductions coop allowances new product offerings shelfspace decisions coordination with newspaper advertising logistics at the ware-houses as well as at each store store differences in terms of customer characteris-tics and competitive environments productshelf selection and layout etc
35 If the supermarket chains indeed restructure the entire price change pro-cess and begin to change prices much more frequently (say two-to-three times aweek) on a regular basis then the shape of this cost function could change in anunpredictable way
QUARTERLY JOURNAL OF ECONOMICS818
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
V3 Relating Our Findings to the Existing Theoretical Models ofMenu Costs
In order to assess the magnitude of these menu cost guresin the context of the existing theoretical menu cost literatureconsider the following calculation experiments done within theframework of two theoretical menu cost models Blanchard andKiyotaki [1987] study a general equilibrium menu cost modelwith monopolistic competition and with unit elasticity of aggre-gate demand with respect to real money balances According totheir calculations menu costs of the magnitude of 008 percent ofrevenues which they consider ldquovery smallrdquo may be sufcient toprevent adjustment of prices Ball and Romer [1990] conductsimilar calculations in the framework of a model with imperfectcompetition and menu costs They nd that for plausible markupand labor elasticity parameter values menu costs needed to pre-vent price adjustment are 070 percent of revenues which as theysuggest are nonnegligible For smaller menu costs eg 004 per-cent of the revenue which Ball and Romer consider ldquotrivialrdquo im-plausibly large values of markup and labor elasticity parametersare needed to prevent price adjustment to monetary shocks
According to the gures in Table IV the reported menu costto revenues ratio for Chains AndashD averages 070 percent rangingfrom 061 percent to 076 percent These are signicantly largerthan the ldquotrivialrdquo 004 or 008 percent gures mentioned abovesuggesting that the menu cost gures we nd here are nontrivialFurther under the model and parameter values considered byBlanchard and Kiyotaki [1987] the menu cost gures we nd arehigher than the theoretical minimum needed to form a barrier toprice adjustments Under the model and parameter values con-sidered by Ball and Romer [1990] the reported menu costreve-nue ratio for all but one chain reaches or crosses the theoreticalthreshold of 070 needed to form a barrier to price adjustmentsThe existence of numerous unmeasured menu cost componentsdiscussed in subsection III5 also raises the possibility that theactual menu costs incurred by these chains are signicantlyhigher than that threshold This suggests that the menu costs wereport may be large enough to form a barrier to nominal priceadjustments when interpreted in the context of these models
An issue of interest in this literature is time-dependent ver-sus state-dependent pricing rules (see for example Caplin andLeahy [1991]) The evidence from our data set on the timing of
THE MAGNITUDE OF MENU COSTS 819
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
price changes suggests that the price change decisions in thesesupermarkets have a strong time-dependent price-setting ele-ment36 For example according to Table VI the prices in Chain Aare changed weekly according to the following schedule the pricechanges of general merchandise that is advertised is done Satur-day nights grocery and produce prices are changed Sundaynights the rest of the general merchandise prices are changed onMonday nights and the prices of advertised grocery items arechanged on Tuesday nights Thus as the gures in Table VI indi-cate over 96 percent of the price changes are done during Sundayand Monday nights However this does not imply that state-dependent pricing rules are unimportant Even if price changesacross product categories follow a prescheduled weekly time ta-ble the prices of which products to change is likely to be a state-dependent decision For example it could depend on changes insupply and demand conditions such as competitorsrsquo price changedecisions Indeed Levy Dutta and Bergen [1996] use retailstore-level orange juice price and cost data to demonstrate thatthe extent of price response to cost shocks that is the extent ofprice rigidity may depend on the nature of supply shocks37
We do not want to overstate the usefulness of our data fordirectly addressing the issue of monetary nonneutrality On onehand authors such as Caplin and Spulber [1987] have demon-strated that under certain conditions individual price rigiditymay not be sufcient for aggregate price rigidity but unfortu-nately our data do not speak directly to this issue38 On the otherhand authors such as Akerlof and Yellen [1985] Mankiw [1985]Parkin [1986] and Caplin and Leahy [1997] have shown that even
36 Danziger [1983] Caballero [1989] and Ball and Mankiw [1994] suggestthat time-dependent price adjustment of the type documented here can be optimalif the cost of gathering information about the state exceeds the cost of making theprice adjustment itself
37 We also considered the possibility of convexities in the cost of changingprices Our data do not suggest many convexities since the labor time spent inthe price tag change process the cost of printing and delivering price tags andin-store supervision time do not change with the size of a price change The onlymeasurable component of menu costs that could be convex is the cost of mistakesmade the larger the price change the higher the probability of a customer notic-ing the mistake and the larger the amount required to be refunded as compensa-tion Among the unmeasured components of menu cost the cost of corporatemanagerial time may increase with the size of a price change because larger pricechanges may require more serious consideration
38 Also see Balke and Wynne [1996] and Bryan and Cecchetti [1996] Sincewe do not have measures of how menu costs change over time our data also havelittle to say about time variation in menu costs or about the relationship betweenmenu costs and ination which during the period of the data collection (1991ndash1992) averaged about 3 percent annually
QUARTERLY JOURNAL OF ECONOMICS820
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
small menu costs can be relevant since they may be sufcient togenerate substantial aggregate nominal rigidity and thus largebusiness cycles Therefore as Blinder [1994] Kashyap [1995]and Slade [1996a] emphasize it is important to search for directevidence that such costs are indeed present at the micro level Bydirectly identifying documenting and measuring the magnitudeof menu costs at the store level we are taking an important stepin that direction
VI CONCLUSION AND FUTURE RESEARCH
Our main contribution is that we provide direct microeco-nomic evidence on the actual magnitude of menu costs for fourlarge U S retail supermarket chains The annual menu costs perstore at these chains average $105887 comprising 070 percentof revenues 352 percent of net prot margins and $052 perprice change on average
Further we provide evidence which suggests that thesemenu costs may be forming a barrier to price changes Specifi-cally we show that (1) supermarket chains not subject to itempricing law change prices two and a half times more frequentlythan the chain that is subject to the law (2) within the chain thatis subject to the item pricing law they change prices over threetimes as frequently for products that are exempt from this lawthan for the products which are subject to this law and (3) wend that these chains do not adjust prices of up to one-third ofthe products for which they face cost increases because of menucost considerations
When we relate these ndings to the existing theoreticalmodels of menu costs we conclude that the magnitude of menucosts we nd is large enough to be capable of having macroeco-nomic signicance Specically when considered in the context ofthe theoretical menu cost models of Blanchard and Kiyotaki[1987] and Ball and Romer [1990] we nd that the menu costgures we report are ldquonontrivialrdquo and their relative magnitudescross the minimum theoretical threshold needed to form a barrierto price adjustments
Despite the high relative magnitude of the marginal cost ofchanging price that we found the data still indicate frequentweekly price changes This is because of the high marginal bene-t of changing price which is due to the erce competition foundin the retail supermarket industry That marginal benets may
THE MAGNITUDE OF MENU COSTS 821
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
outweigh the marginal costs of changing prices in this industryhowever does not rule out the possibility that menu costs of themagnitude we nd here can create substantial nominal rigidityin other industries or markets In less competitive industries andmarkets menu costs of the type and magnitude we documenthere are likely to have a bigger impact on the frequency of priceadjustment and consequently on the degree of price rigidity
At a minimum the direct dollar measures of menu costs wereport here can be used as a starting point for future research onmeasuring their magnitude in other industries and establish-ments We anticipate that these menu costs will be similar inother markets which rely on posted prices (such as drugstoresdepartment stores etc) because the steps involved in the pricechange process are likely to be similar What may differ are thefrequency of price changes and the labor wage rates at thesestores For example since supermarket chains change thousandsof prices each week the total annual menu costs incurred bythese chains per store are likely to be high in comparison to otherretail formats such as chain drugstores or department storesThere are however a variety of industries for which the stepsinvolved in changing prices would be signicantly different fromthose reported in our study For example business-to-businesssales which often rely on a sales force will require changes in thelist price sheets changes in the instructions to the sales forcewhich may include education and discussion with the salespeoplein the company and so forth These business-to-business pricesalso often have more complex pricing schemes including quantitydiscounts bundling and individually negotiated prices As an-other example the composition of the cost of changing the news-stand prices of magazines [Cecchetti 1986] or the prices ofproducts sold through catalogs [Kashyap 1995] are different frommany of the menu cost components we discuss here Further thending that a centralization of pricing decisions makes the store-level managerial component of menu cost relatively small sug-gests that the managerial menu costs likely are highest insettings where price change decisions are decentralized Thus fu-ture empirical work should look at menu cost and its compositionin a variety of other industries markets and products with bothcentralized as well as decentralized price change decisions in or-der to see whether the magnitude of these costs can be general-ized and benchmarks can be established Further there aretechnological changes taking place in this and other industries
QUARTERLY JOURNAL OF ECONOMICS822
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
that promise to alter the structure of menu costs which deserveattention At the theoretical level our ndings suggest that it maybe worthwhile to explore models which incorporate the idea ofitem pricing laws as an additional component of menu costs
EMORY UNIVERSITY
UNIVERSITY OF MINNESOTA
UNIVERSITY OF SOUTHERN CALIFORNIA
ROBERT W BAIRD amp COMPANY
REFERENCES
Akerlof George A and Janet L Yellen ldquoA Near-Rational Model of the BusinessCycle with Wage and Price Inertiardquo Quarterly Journal of Economics C(1985) 823ndash38
Amano Robert A and R Tiff Macklem ldquoMenu Costs Relative Prices and Ina-tion Evidence for Canadardquo Working Paper Research Department Bank ofCanada 1995
Andersen Torben M Price Rigidity Causes and Macroeconomic Implications(Oxford Clarendon Press 1994)
Balke Nathan S and Mark A Wynne ldquoSupply Shocks and the Distribution ofPrice Changesrdquo Federal Reserve Bank of Dallas Economic Review (1996)10ndash18
Ball Laurence and N Gregory Mankiw ldquoA Sticky-Price Manifestordquo Carnegie-Rochester Conference Series on Public Policy (1994) 127ndash52
Ball Laurence and N Gregory Mankiw ldquoRelative Price Changes as AggregateSupply Shocksrdquo Quarterly Journal of Economics CX (1995) 161ndash93
Ball Laurence N Gregory Mankiw and David Romer ldquoThe New Keynesian Eco-nomics and the Output-Ination Trade-offrdquo Brookings Papers on EconomicActivity 1 (1988) 1ndash65
Ball Laurence and David Romer ldquoReal Rigidities and Nonneutrality of MoneyrdquoReview of Economic Studies LVII (1990) 183ndash203
Bergen Mark Shantanu Dutta and Steven M Shugan ldquoUsing Branded Vari-antsrdquo Journal of Marketing Research XXXIII (1996) 9ndash19
Berkowitz Eric N Roger A Kerin and William Rudelius Marketing (St LouisMO Times MirrorMosby College Publishing 1986)
Blanchard Olivier J and Nobuhiro Kiyotaki ldquoMonopolistic Competition and theEffects of Aggregate Demandrdquo American Economic Review LXXVII (1987)647ndash66
Blattberg Robert C and Scott A Neslin Sales Promotion Concepts Methodsand Strategies (Englewood Cliffs NJ Prentice Hall 1989)
Blinder Alan S ldquoWhy Are Prices Sticky Preliminary Results from an InterviewStudyrdquo American Economic Review LXXXI (1991) 89ndash96
mdashmdash ldquoOn Sticky Prices Academic Theories Meet the Real Worldrdquo in MonetaryPolicy N Gregory Mankiw ed (Chicago IL University of Chicago Press forthe NBER 1994)
Bryan Michael F and Stephen G Cecchetti ldquoInation and the Distribution ofPrice Changesrdquo NBER Working Paper No 5793 1996
Caballero Ricardo J ldquoTime-Dependent Rules Aggregate Stickiness and Infor-mal Externalitiesrdquo Columbia University Discussion Paper No 428 1989
Calatone Roger J Cornelia Droge David Litvack and C Anthony DiBenedettoldquoFlanking in a Price Warrdquo Interfaces XIX (1989) 1ndash12
Caplin Andrew ldquoIndividual Inertia and Aggregate Dynamicsrdquo in Optimal Pric-ing Ination and the Cost of Price Adjustment Eytan Sheshinski and YoramWeiss eds (Cambridge MA The MIT Press 1993)
Caplin Andrew S and John Leahy ldquoState Dependent Pricing and the Dynamicsof Money and Outputrdquo Quarterly Journal of Economics CVI (1991) 683ndash708
THE MAGNITUDE OF MENU COSTS 823
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
Caplin Andrew and John Leahy ldquoAggregation and Optimisation with State-Dependent Pricingrdquo Econometrica LXV (1997) 601ndash05
Caplin Andrew S and Daniel F Spulber ldquoMenu Costs and the Neutrality ofMoneyrdquo Quarterly Journal of Economics CII (1987) 703ndash25
Carlton Dennis W ldquoThe Rigidity of Pricesrdquo American Economic Review LXXVI(1986) 637ndash58
mdashmdash ldquoThe Theory and the Facts of How Markets Clear Is Industrial OrganizationValuable for Understanding Macroeconomicsrdquo in Handbook of Industrial Or-ganization Volume 1 Richard Schmalensee and Robert D Willig eds (Am-sterdam North-Holland 1989)
Carlton Dennis W and Jeffrey M Perloff Modern Industrial Organization (NewYork NY Harper Collins 1994)
Cecchetti Stephen G ldquoThe Frequency of Price Adjustment A Study of the News-stand Prices of Magazinesrdquo Journal of Econometrics XXXI (1986) 255ndash74
Chevalier Judith A ldquoCapital Structure and Product Market Competition Em-pirical Evidence from the Supermarket Industryrdquo American Economic Re-view LXXXV (1995) 415ndash35
Danziger Leif ldquoPrice Adjustments with Stochastic Inationrdquo International Eco-nomic Review XXIV (1983) 699ndash707
mdashmdash ldquoInation Fixed Cost of Price Adjustments and Measurement of RelativePrice Variabilityrdquo American Economic Review LXXVII (1987) 704ndash13
Dutta Shantanu Mark Bergen and Daniel Levy ldquoPrice Flexibility Evidencefrom Scanner Datardquo Working Paper University of Chicago and Emory Uni-versity 1995
Eden Benjamin ldquoTime Rigidities in the Adjustment of Prices to Monetary ShocksAn Analysis of Micro Datardquo Discussion Paper No 9416 Bank of Israel 1994
Federal Trade Commission ldquoPrice Check A Report on the Accuracy of CheckoutScannersrdquo (October 1996)
Goodstein Ronald C ldquoUPC Scanner Pricing Systems Are They Accuraterdquo Jour-nal of Marketing LVIII (1994) 20ndash30
Gordon Robert J ldquoWhat is New-Keynesian Economicsrdquo Journal of EconomicLiterature XXVIII (1990) 1115ndash71
Haddock David D and Fred S McChesney ldquoWhy Do Firms Contrive ShortagesThe Economics of Intentional Mispricingrdquo Economic Inquiry XXXII (1994)562ndash81
Hoch Stephen J Xavier Dreze and Mary E Purk ldquoEDLP Hi-Lo and MarginArithmeticrdquo Journal of Marketing LVIII (1994) 16ndash27
Direct Store Delivery Work Group et al ldquoDirect Store Delivery An Efcient Con-sumer Response Best Practices Reportrdquo Joint Industry Project on EfcientConsumer Response Industry Relations Department Grocery Manufactur-ers of America 1995
Kashyap Anil K ldquoSticky Prices New Evidence from Retail Catalogsrdquo QuarterlyJournal of Economics CX (1995) 245ndash74
Lach Saul and Daniel Tsiddon ldquoThe Behavior of Prices and Ination An Empiri-cal Analysis of Disaggregated Datardquo Journal of Political Economy C (1992)349ndash89
Lach Saul and Daniel Tsiddon ldquoStaggering and Synchronization in Price-Setting Evidence from Multiproduct Firmsrdquo American Economic Review1996 LXXXVI (1996) 1175ndash96
Lattin James M and Gwen Ortmeyer ldquoA Theoretical Rationale for EverydayLow Pricing by Grocery Retailersrdquo Working Paper Graduate School of Busi-ness Stanford University 1991
Levy Daniel Shantanu Dutta and Mark Bergen ldquoHeterogeneity in Price Rigidityand Shock Persistencerdquo Working Paper University of Chicago and EmoryUniversity 1996
Levy Daniel Shantanu Dutta Mark Bergen and Robert Venable ldquoPrice Adjust-ment at Multiproduct Retailersrdquo Working Paper Emory University 1997
Liebermann Yehoshua and Ben-Zion Zilberfarb ldquoPrice Adjustment Strategy un-der Conditions of High Ination An Empirical Examinationrdquo Journal of Eco-nomics and Business XXXVII (1985) 253ndash65
Mankiw N Gregory ldquoSmall Menu Costs and Large Business Cycles A Macroeco-nomic Model of Monopolyrdquo Quarterly Journal of Economics C (1985) 529ndash39
QUARTERLY JOURNAL OF ECONOMICS824
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825
Mankiw N Gregory and David Romer New Keynesian Economics (CambridgeMA MIT Press 1991)
Meltzer Allan H ldquoInformation Sticky Prices and Macroeconomic FoundationsrdquoFederal Reserve Bank of St Louis Review LXXVII (1995) 101ndash18
Money Magazine ldquoDonrsquot get cheated by Supermarket Scannersrdquo an article byVanessa OrsquoConnell XXII (1993) 132ndash38
Montgomery Alan L ldquoThe Impact of Micro-Marketing on Pricing StrategiesrdquoPhD thesis Graduate School of Business University of Chicago 1994
Okun Arthur M Prices and Quantities A Macroeconomic Analysis (WashingtonDC The Brookings Institution 1981)
Parkin Michael ldquoThe Output-Ination Trade-off When Prices Are Costly toChangerdquo Journal of Political Economy XCIV (1986) 200ndash24
Romer David Advanced Macroeconomics (New York NY McGraw-Hill 1996)Rotemberg Julio J ldquoSticky Prices in the United Statesrdquo Journal of Political
Economy XC (1982) 1187ndash211Sheshinski Eytan A Tishler and Yoram Weiss ldquoInation Costs of Price Adjust-
ments and the Amplitude of Real Price Changes An Empirical Analysisrdquo inDevelopment in an Inationary World J Flanders and Assaf Razin eds (NewYork NY Academic Press 1981)
Sheshinski Eytan and Yoram Weiss ldquoInation and Costs of Price AdjustmentsrdquoReview of Economic Studies XXXXIV (1977) 287ndash303
Sheshinski Eytan and Yoram Weiss Optimal Pricing Ination and the Cost ofPrice Adjustment (Cambridge MA The MIT Press 1993)
Slade Margaret E ldquoOptimal Pricing with Costly Adjustment Evidence fromRetail-Grocery Pricesrdquo The University of British Columbia submitted1996a
mdashmdash ldquoSticky Prices in a Dynamic Oligopoly An Investigation of (sS) ThresholdsrdquoUniversity of British Columbia manuscript 1996b
Supermarket Business ldquoConsumer Expenditures Studyrdquo XLVIII (1993) 52Warner Elizabeth J ldquoPricing in the Retail Industry A Case Studyrdquo manuscript
Hamilton College 1995Warner Elizabeth J and Robert Barsky ldquoThe Timing and Magnitude of Retail
Store Markdowns Evidence from Weekends and Holidaysrdquo Quarterly Jour-nal of Economics CX (1995) 321ndash52
THE MAGNITUDE OF MENU COSTS 825