Slotting Allowances: An Empirical Investigation1

K. Sudhir

Yale School of Management 135 Prospect St, PO Box 208200

New Haven, CT 06520 Email: sk389@mail.som.yale.edu

Phone: 203-432-3289 Fax: 203-432-3003

Vithala R. Rao

Johnson Graduate School of Management Cornell University

Sage Hall Ithaca, NY 14853

E-Mail: vrr2@cornell.edu Phone: 607-255-3987 Fax: 607-254-4590

March 2004

1 We thank Dmitri Kuksov, J.P. Dube, Edward McLaughlin and participants at the Marketing Workshops at the University of Colorado, Washington University and Yale University for their comments.

Slotting Allowances: An Empirical Investigation

Abstract

Slotting allowances are lump-sum payments by manufacturers to retailers for stocking new

products. The economic rationale for slotting allowances is controversial. Several theoretical

arguments have been provided for the use of slotting allowances. Some are based on efficiency

considerations: (1) efficient allocation of scarce retail shelf space, (2) equitable allocation of new

product failure risk between manufacturers and retailers and (3) signaling device for

manufacturers to communicate private information about potential success to the retailer. Others

have argued tha t slotting allowances are an anti-competitive device that (1) manufacturers use to

reduce retail competition (2) large manufacturers use to exclude small manufacturers. Others

suggest that slotting allowances are the result of retailers exercising power. However, empirical

research has been virtually non-existent due to the difficulty in obtaining data about these

transactions.

Using data on all new products that were offered to one retailer for a period of six months, we

empirically investigate support for the alternative rationales for slotting allowances. Our analysis

indicates that broadly there is more support for the efficiency theories than for the anti-

competitive theories. We find evidence that slotting allowances (1) serve to efficiently allocate

scarce retail shelf space; (2) help balance the risk of new product failure between manufacturers

and retailers; (3) help only large manufacturers communicate their private information about

potential success of new products and (4) serve to widen retail distribution for manufacturers by

mitigating retail competition.

1

1. Introduction

Slotting allowances are lump-sum payments by manufacturers to retailers for stocking

new products. Over the last two decades, they have gained increasing prominence and have

emerged to be a very major share of new product development costs.1 According to Deloitte and

Touche (1990), slotting allowances account for over 16% of a new product’s introductory costs,

while R&D and market analysis expenditures account for about 14%. In terms of dollar amounts,

slotting allowance for an SKU varies from $ 75-$300 per store (FTC 2001, p.11), from $3,000-

$40,000 for a regional chain (Fields and Fulmer 2000; Desiraju 2001), and from $1.4 to 2 million

for a national introduction (Thompson 2000; Vosburgh 2001).

Given the substantial amounts of money involved, one might expect that the strategic

rationales for slotting allowances would be well understood. But slotting allowances are

extremely controversial and there is very little consensus either among practitioners, regulators

or researchers as to the true role of slotting allowances in facilitating new product introductions

(Bloom et al. 2000 and Gundlach et al. 2002). While some theorists and practitioners have

suggested that slotting allowances are anti-competitive, others have argued that slotting

allowances serve to enhance efficiency of market outcomes.

A measure of the controversial role of slotting allowances is in the contrasting positions

taken by two regulatory authorities. The Bureau of Alcohol, Tobacco and Firearms (BATF)

banned slotting allowances in the alcohol trade in 1995. In contrast, the Federal Trade

Commission (FTC) which regulates the grocery industry refuses to provide guidelines given that

slotting allowances can have both efficiency and anti-competitive effects and therefore further

investigation is needed (FTC 2001). The U.S. Senate’s Committee for Small Business and

Entrepreneurship funded a full-scale public investigation of slotting allowances by the FTC, but

the report was inconclusive, citing difficulties in obtaining information from either

manufacturers or retailers about slotting allowances. See Gundlach and Bloom (1998) for a

discussion on the special characteristics of the alcohol market that led to the BATF decision.

There are three major efficiency arguments that are cited in favor of slotting allowances

(1) that they serve to efficiently allocate scarce retailer shelf space to the most valuable

(profitable) new products (2) that they serve to allocate risk of new product failure in a balanced

1 The date of origin of slotting allowances is ambiguous. According to Supermarket News, slotting allowances per se did not exist before 1984 (August 27 1984). PROMO: The Magazine for Promotional Marketing, (January, 1989) notes the prevalence of slotting allowances starting 1982 in a report based on surveys of manufacturers and retailers.

2

manner between manufacturers and retailers and (3) that they serve to communicate private

information that manufacturers may have about the potential success of the new product to the

retailer. The main anti-competitive explanations for slotting allowances are: (1) they are a

mechanism by which large manufacturers with deep pockets exclude small manufacturers with

fewer resources; (2) they are a means to mitigate retail competition and (3) it is due to the

exercise of retail power. The retail power argument suggests that in many local markets, high

retail concentration has concentrated shelf-space ownership in the hands of few retailers,

enabling them to demand slotting allowances.

Despite the abundance of theoretical rationales for slotting allowances, the extant

empirical research on this topic is limited with inconclusive findings. Two survey-based studies

(Bloom et al. 2000 and Gundlach et al. 2002) show limited consensus between manufacturers

and retailer on the reasons for the use of slotting allowances. The problem may be due to the use

of the survey method; Bloom et al. (2000) acknowledge the limitations of survey method when

they state: “ …although an anonymous survey targeted towards informed industry participants

was used, manager perceptions, for a variety of reasons, may not accurately reflect reality and

their opinions about slotting fees. Respondents may have had limited information and knowledge

or their perceptions may have been distorted by self- interest.” This problem does not seem to

disappear even when the respondents (retailers and manufacturers) were instructed to focus on

the characteristics of specific transactions in which the participants were involved and slotting

allowances are offered; see Rao and Mahi (2003). While the survey-based studies can be useful

in gaining insights on how the two channel members disagree, they give limited insights into the

rationales for observed market outcomes involving slotting allowances.

Sullivan (1997) uses correlational analysis on objective time series data at the market

level (e.g., number of new products supplied by manufacturers, number of SKUs kept at

retailers, quantities sold by retailers and prices charged by manufacturers) and argues that the use

of slotting allowances results in an equilibrium in which the number of products offered by the

manufacturer equals the number of products demanded by the retailer. But the aggregate level

data are not capable of distinguishing between competing predictions present in the theoretical

literature.

Given the state of empirical research in this area, a study that uses observational data on

new product introductions and the associated terms of trade (including slotting allowances) can

3

be insightful in resolving the several contradictions due to the divergence of manufacturer and

retailer opinions. In this paper, we use a unique dataset with actual information on all new

product offers presented to a retailer during a period of six months. The dataset has a

combination of objective information associated with the new product offer and retailer

evaluations about the manufacturer and the product offered. The focal retailer is a medium sized

chain with about 100 stores. It is very unlikely that any manufacturer introducing a product in

this region would bypass this retailer. Further, its headquarters is based in a market that is widely

used as a test-market due to its representative consumer profile. Hence we believe that studying

the offers presented to this retailer in detail can give us results that permit a cautious

generalization about the rationales for the use of slotting allowances.

Given the difficulty in collecting empirical data on slotting allowances, we believe our

detailed dataset on over 1000 products offers a rare opportunity for an empirical investigation of

the rationales for slotting allowances.2 The analysis of a large number of offers to one retailer,

while restricting generalizability helps us address econometric difficulties that would be hard to

address if we pooled (a small number of) observations across multiple retailers. By restricting

our analysis to multiple transactions involving one retailer, we are able to control for the retailer

end of the transaction-dyad and see how slotting allowances vary across different manufacturers.

This strategy is critical in developing appropriate tests of game theoretic predictions. While

rejecting signaling theories based on their analysis of data across multiple manufacturers and

retailers, Rao and Mahi caution (p. 265), “… slotting allowances paid for each transaction in our

data may in fact be entirely consistent with a signaling story. … Hence each data point may be a

consequence of particular levels of information asymmetry that are specific to that transaction.

Only dyadic data would allow us to assess whether signaling is in fact occurring.” It is plausible

that some of the counter-intuitive empirical results of Rao and Mahi (2003) arise due to this

inability to control at least one end of the transaction-dyad.

We provide a brief overview of the efficiency and anti-competitive rationales in Section

2. In Section 3, we present the data that we use and give a brief descriptive analysis. Section 4

develops the empirical tests of the different rationales and describes the empirical results. Section

5 concludes.

2 The 2001 FTC report discusses the difficulty in obtaining data on slotting allowances and in obtaining cooperation from large manufacturers and retailers. Congress has provided a budget of $900,000 to the FTC for an empirical study of slotting allowances, considering the difficulties in obtaining data on this issue (FTC, 2001).

4

2. Rationales for Slotting Allowances

The literature offers two classes of rationales for slotting allowances: (1) Efficiency-

enhancing and (2) Anti-competitive. We discuss these rationales and the extant empirical support

or lack thereof for them in existing research. As we shall see, there is very little consensus about

the roles of the different rationales in existing research.

A. Efficiency-Enhancing Rationales

A1. Slotting Allowances Help Efficient Allocation of Scarce Shelf Space

A typical supermarket carries about 35,000 SKUs (Food Marketing Institute, 2003a),

while the number of new products introduced by manufacturers range from about 10,000 to

16,000 SKUs (Food Marketing Institute, 2003b). Hence introducing a new product invariably

requires retailers to drop an existing product. Shelf space is therefore clearly a scarce resource

with high opportunity cost. Further, private labels have also grown in market share over the last

two decades. This has further increased the opportunity cost for shelf space, since private labels

tend to have higher retail margins than national brands (Coppa, 2003). Sullivan (1997) claims

support for the shelf-space argument for the use of slotting allowance.

But, results from surveys of manufacturers and retailers have been mixed. While

manufacturers and retailers agree that the number of new products introduced is a key force in

the widespread use of slotting fees (Wilkie et al. 2002; Bloom et al. 2000), Bloom et al. (2000)

find that retailers do not believe that slotting fees are related to the opportunity cost of shelf-

space. Wilkie et al. (2002) also find no support for the growth of private label brands as a

rationale for the growth of slotting allowances. Basically, in these survey-based studies, it

appears that retailers do not want to accept their role in the emergence of slotting allowances, but

want to blame manufacturers. Therefore, an empirical analysis based on observed outcomes

(rather than based on opinions, which may be self-serving) as we propose to do in this data, can

help in assessing the true role of opportunity cost of shelf-space as a rationale for slotting

allowances.

The opportunity cost of shelf-space rationale is distinct from an operating cost-based

rationale that has been offered in the literature. One operating cost rationale is that retailers use

slotting allowances to cover new product introduction costs (e.g., cost of entering changes in the

inventory and accounting systems, setting “slots” in the warehouse etc) The current use of

slotting allowances goes far beyond the original purpose of recovering the true cost incurred by

5

the retailer for introducing the product. Manufacturers do not believe operating costs to be a

major rationale for slotting allowances, though retailers claim otherwise. We do not believe the

operating costs argument is reasonable, because if it were, slotting allowances should be the

similar across most new products.

In fact, Rao and Mahi (2003) test the operating cost rationa le and find the “surprising”

result that slotting allowances are negatively related to retailer operating costs. However, the

surprise is easy to explain (as the authors themselves do), when it is recognized that their study

analyzes data across multiple retailers. Retailers who consider that they are more efficient tend to

have lower operating costs in the industry and are also likely to be the larger retailers who have

greater opportunity costs of shelf-space. Hence the negative relationship between slotting

allowances and operating costs may in fact imply a positive relationship between slotting

allowances and opportunity costs.

A2. Slotting Allowances Balance New Product Failure Risk between Manufacturer and

Retailer

New product failure rates are alarmingly high and estimated at about 80-90% (FTC,

2001). Given the upfront operating costs incurred by the retailer when introducing a new

product, the high rate of failure poses a risk to the retailer. In addition, retailers are left with

unsold inventory tha t they have already paid for. In addition, they lose potential revenue from

more profitable products. FTC Commissioner, Deborah K. Owen (Antitrust and Trade

Regulation Report, 1994 p. 703; 60 Federal Register 20409) states that slotting allowances

“provide a form of insurance for the retailer…[that] reduce, and perhaps eliminate [its] risk– or

at least transfer some of it to the producer– by charging a fee that essentially provides

indemnification from the loss of profits that would arise if the new product fails to sell well.”

The notion that new-product failure risk is an important explanation for the use of slotting

allowances has been supported in survey opinions of both manufacturers and retailers (Bloom et

al. 2000).

Some retailers charge failure fees and require buyback guarantees from manufacturers to

refund the cost of unsold inventory to ameliorate these risks. However such requirements are not

necessarily honored, especially by smaller manufacturers. Wilkie et al. (2002) quote a large

wholesaler from their survey, “When a product fails, many small manufacturers are not around to

clean up the residue, which is then discounted to salvage dealers/auctioneers” (p.282). This

6

suggests that risk is greater with respect to small manufacturers than with respect to large

manufacturers who are more likely to be longer-term players. The greater risk might imply that

slotting allowances will be demanded more from smaller firms. The challenge in empirically

testing this theory would be to distinguish the risk based explanation from a retail power-based

explanation, which suggests that smaller manufacturers would have to pay more slotting

allowances due to their relatively low power with respect to the retailer.

Related to this issue are the different levels of risk aversion between large and small

manufacturers in taking the risk of when paying slotting allowances. Large manufacturers with

deep pockets can withstand a new product failure more easily than smaller manufacturers. Given

this, small manufacturers will be far more risk averse to pay upfront slotting allowances and may

require a higher degree of confidence in success of their products than larger manufacturers

before “betting the bank.”3.

A3. Slotting Allowances Communicate Manufacturers’ Private Information about New

Product Success to Retailer

In most new product introductions, it can be reasonably argued that manufacturers may

have private information about the potential success of a new product. Lariviere and

Padmanabhan (1997) and Desai (2000) argue that slotting allowances can serve as signals by

which manufacturers can credibly communicate positive private information to the retailer. The

argument essentially is that only manufacturers who have positive private information about the

potential success of the product will pay the slotting allowances, while others with negative

private information will not pay the slotting allowances.

Survey based research has found little support for the signaling theory. According to

Bloom et al. (2000), neither manufacturers nor retailers believe that “slotting fees size is a good

indicator of the likely success of a new product” (p. 101). Rao and Mahi (2003) also find little

support for the signaling theory but are equivocal when they state that the inability to control for

dyadic information is a possible reason for their negative finding.

An important point to note about the signaling theories is to distinguish between

observable aspects of product success and unobservable aspects of product success. Practitioners

tend to say that slotting allowances do not affect their buying decisions; retailers accept excellent

3 The problems of smaller manufacturers in obtaining finances for paying slotting allowances are described by a FTC workshop participant quite well: “[B]anks do not finance marketing in any way, shape or form. They finance machinery, automobiles. They don’t even like to finance your office building.” (FTC 2001, p. 23)

7

products with high brand equity without slotting allowances. However this misses the point

about signaling. Products rated highly by the retailer are likely to be accepted even without

slotting allowances; products that are rated poorly by the retailer are likely to be rejected whether

slotting allowances are offered or not. It is when there is a high degree of uncertainty for the

retailer, that slotting allowances can add value to the retailer by communicating private

information. Hence any empirical strategy needs to account for the level of uncertainty that the

retailer has about the likelihood of product success.

B. Anti-Competitive Rationales

B1. Manufacturers Seek to Mitigate Retail Competition to Increase Profits

Shaffer (1991) provides an interesting anti-competitive rationale for the use of slotting

allowances. He argues that slotting allowances serve to mitigate retail competition and increases

retail prices and retail profits. This is because competitive manufacturers raise their wholesale

prices in the presence of slotting allowances and the higher wholesale prices commits retailers to

less aggressive retail pricing.

Bloom et al. (2000) find that both manufacturers and retailers agree that slotting

allowances raise retail prices. They use this to claim support for Shaffer’s theory, because it

predicts that slotting allowances will raise retail prices. However, such a test is by no means

proof of Shaffer’s theory because almost all equilibrium models of slotting allowances predict

that slotting allowances are accompanied by higher retail prices. For example, the signaling

models of Lariviere and Padmanabhan (1997) and Desai (2000) also lead to similar predictions.

Hence there is no empirical evidence for or against Shaffer’s rationale in the extant literature.

B2. Large Manufacturers Seek to Exclude Small Manufacturers (Competitive Foreclosure)

Small manufacturers complain that slotting allowances are often used by larger

manufacturers to exclude them from markets. In fact, the Senate Committee for Small Business

and Entrepreneurship held its hearings because of its interest in this issue. Ferrell (2001) argues

that dominant manufacturers may pursue a policy of exclusion using slotting allowances in order

to keep wholesale prices higher by keeping out smaller manufacturers. Bloom et al. (2000) find

that manufacturers believe this argument to be true, while retailers do not.

B3. Exercise of Retail Power

The rise in retail power has been widely suggested as a rationale for the use of slotting

allowances. Due to the wave of mergers in the supermarket industry and to the rise of strong

8

national players such as Wal-Mart, buying power is highly concentrated in the hands of chain-

level buyers. The top five firms’ share of sales has increased from 20% in 1993 to 42% of sales

in 2000 (Swenson 2000). Further, the four firm retail concentration in the top 100 markets

averaged 72% in 1998 (Kaufman 2000).

Chu (1992) argues that retail power is a primary reason for slotting allowances. He

develops a screening model by which retailers charge slotting allowances and extract all of the

manufacturer’s profits due to its power. Bloom et al. find that manufacturers believe that this is a

primary reason for the increased use of slotting allowances; retailers rate it much lower but still

believe it has an important impact. Rao and Mahi (2003) also find support the retail power

argument. Nevertheless, Messinger and Narasimhan (1995) and Farris and Ailawadi (1992) find

little evidence that power has shifted in the grocery channel based on their empirical analysis of

retailer profitability over time. However, it is possible that the balance of power has shifted since

1995, given that retailers such as Wal-Mart have grown more powerful in the last decade.

3. Data and Descriptive Analysis

Data

We use data on all new product offers by manufacturers to a large supermarket chain

over a six month period from June 1986 to February 1987. Since we have data on all new

product offers and not just those that are accepted, we do not have any endogenous selection

problems. The supermarket chain has approximately 100 stores and covers a large trading area in

the northeastern United States. The headquarters’ region of this chain is frequently used by

manufacturers for test marketing, because of its representative consumer profile. Further, it is

highly unlikely that any food manufacturer would bypass this retailer in the introduction of a

new product. Hence even though our data applies to one company, we believe the

representativeness of the retailer permits us to cautiously generalize to other large regional

retailers.

We have two types of primary data from the retail chain: (a) manufacturer-supplied

materials such as product physical characteristics (e.g., case cubic dimensions), financial

information (e.g., gross margin), promotional support (e.g., extent of advertising), market

research (whether test market was conducted or not), terms of trade (slotting allowances) and (b)

a one-page questionnaire completed by the retail buyer assessing his/her judgments of

9

product/manufacturer attributes (e.g., likelihood of product success based on past experience

with manufacturer/industry reputation). It is important to note that we do not have data on the

magnitude of slotting allowance; we only know whether slotting allowances are provided or not

with a particular product offer.

Descriptive Analysis

During the six month period of data collection, 2186 products were introduced by

manufacturers and considered by the retail buyers. Of these, only 1021 observations were usable

due to missing data problems. Statistical tests based on differences between means and

frequencies (two sample t-tests and 2ϒ tests) were made on each of the variables to determine

the degree of difference between the total sample of 2186 products and the sub-sample of 1021

products. We therefore conclude that the selected sub-sample is not systematically different

from the population of 2186 products. We further note that the analysis includes all products that

are offered to the retailer; not just the ones that are accepted by the retailer.

The 1021 products offered by manufacturers covered 21 different categories as shown in

Table 1. The three largest categories were (1) frozen foods, (2) canned products such as fruits,

vegetables, juice and drinks and (3) dairy and refrigerated foods. Of the 1021 products, 143

products (14%) received slotting allowances. Ignoring categories with small number of product

introductions, the frequency of slotting allowances was particularly high in canned products such

as fruits, vegetables, juice and drinks (24%), household supplies (20%) and health and beauty

aids (20%).

***Insert Table 1 here***

Table 2 shows the means for a number of variables that we study classified on the basis

of whether slotting allowances were offered or not. Products with slotting allowances have

considerably higher average opportunity cost and more private labels suggesting preliminary

support for the opportunity cost hypothesis. The number of competing stores for products that

have slotting allowances are substantially greater than for ones that have no slotting allowances

suggesting support for the retail competition mitigation hypothesis. It is harder to observe

support for the signaling theories from these average numbers, because one needs to account for

interactions between variables. Table 3 shows the correlation between variables of interest.

***Insert Tables 2 and 3 here***

10

We divide firms introducing products into large and small manufacturers to study

differences in slotting allowance offers between the two groups. All of the firms that are

classified as large firms are members of the Grocery Manufacturers Association (GMA), the

umbrella association of leading grocery product manufacturers. The list of all large

manufacturers in our data is provided in Table 4. These 26 large manufacturers introduced 281 of

the 1021 products that we have in our data. On average, the frequency of slotting allowances is

higher for large manufacturers compared to small manufacturers (17% versus 13%).

***Insert Table 4 here***

4. Empirical Tests of Rationales

Broadly, our empirical strategy is as follows: For each of the rationales, we generate

predictions about the relationships between slotting allowances and appropriate variables that are

available in our data. We then estimate logistic regressions with slotting allowances as the

dependent variable and the variables of interest as the explanatory variables. If the estimated

relationship is consistent with the predictions, we treat this as empirical evidence in favor of the

rationale. In cases, where multiple rationales would lead to similar predictions, we refine the

tests using other variables that help distinguish between the rationales.

A. Efficiency Enhancing Rationales

A1. Allocation of Scarce Shelf Space

To test the rationale that slotting allowances serve to efficiently allocate scarce shelf

space, we propose to test whether observed slotting allowances are positively related to the

opportunity cost of shelf-space. We measure opportunity cost of shelf space with two measures:

(i) the presence/absence of private labels in the category, 4 because the higher margins for private

labels in the category can increase the opportunity costs and (ii) the product of shelf space and

gross margins. The second measure is a proxy for profit margins associated with the shelf space.

Ideally, this measure should include the speed with which inventories are turned over.5

Unfortunately, we do not have data on inventory turnover rates. One theoretical issue with this

4 We also use the actual number of private labels in the category and the results are similar. 5 Donald Sussman, principal purchasing executive of Stop and Shop stated in a 2001 FTC workshop that “We have different slotting guidelines in each of our categories depending on the size of the item, the amount of shelf space it takes, the turnover of the item and the category growth of the item” (FTC, 2001, p.15).

11

measure is that unit profits can change in response to the use of slotting allowances. We

therefore use two variations of this measure, one that uses the item-specific profit margin and the

other the category-specific margin. The category-specific profit margin based measure is not

ideal, because not all products in a category are competitors (there can be sub-categories within

which products are likely to be better substitutes) and therefore it will only be weakly correlated

with the true opportunity cost. On the positive side, this measure provides a conservative test of

the opportunity cost hypothesis.

We perform a logistic regression with whether slotting allowances are offered or not as

the dependent variable and the opportunity cost measure (with one of two variations of unit-

specific and category-specific) and private label as the explanatory variables.6 The results are

presented in Table 5. Both the opportunity cost measures and private label have a positive impact

on slotting allowances and are highly significant (p < 0.01) in both regressions. Our analysis

provides fairly strong evidence in favor of the argument that opportunity cost of shelf-space is a

driver of slotting allowances in contrast to the mixed findings of Bloom et al. (2000) discussed

earlier. Further, unlike the opinions expressed by manufacturers and retailers in Bloom et al, we

find that the presence of private labels has the expected positive impact on slotting allowances.

***Insert Table 5 here***

A2. Slotting Allowances Balance New Product Failure Risk between Manufacturer and

Retailer

To understand whether slotting allowances balance new product failure risk between

manufacturers and retailers, a measure of new product risk is necessary. In our data, we have one

variable which measures the retailer’s rating of the manufacturer in producing a successful

product in the category based on either the retailer’s past experience with the manufacturer,

general industry reputation or both. This variable is on a 0-10 scale, where 0 indicates no chance

of success and 10 indicates very high likelihood of success.

One would expect that the uncertainty is greatest around the mid-point of the scale at 5.

When the retailer rates a manufacturer as zero, the retailer is almost certain that the product will

6 To simplify exposition and to graphically illustrate many of our results, it helps to include only the relevant variables for the specific rationales we test in the logistic regression. One might suspect whether these results will continue to hold if all of the variables enter the regression simultaneously. We will show in the robustness section that indeed the coefficients remain fairly similar and significant when we include all regressors simultaneously.

12

be unsuccessful (and therefore unlikely to accept the product, whether it is accompanied by

slotting allowances or not). Similarly, there is very little uncertainty at a rating of 10.

It is important to recognize that the retailer does not provide high ratings to even well-

known large manufacturers for all products they introduce. Similarly small manufacturers do not

necessarily get low ratings. We illustrate this by providing sample ratings for products of some

firms in Table 6. Large manufacturers such as Del Monte, General Foods, Kraft, Proctor and

Gamble and Quaker Oats have ratings that vary from 2 to 8. Seneca Foods, a small local

manufacturer, gets relatively high ratings for the three products it introduced.

***Insert Table 6 here***

We report the results of a logistic regression with slotting allowance as the dependent

variable and rating as the explanatory variable in column 1 of Table 7. Since we expect an

inverted U shaped relationship, we add a squared rating term in the regression. The regression

results are more easily interpreted using a graph showing how the estimated probability of

slotting allowances is related to the rating for the product.7 The graph is shown in Figure 1a. The

inverted U shaped curve for the probability of slotting allowances peaks at around 5.4, around

the region of maximum uncertainty. This suggests that balancing risk is indeed a rationale for

slotting allowances.

***Insert Table 7 here***

We analyze the risk differential between large and small firms by adding a dummy

variable for the size of the manufacturer. These regression results are reported in Column 2 of

Table 7.8 The results show a distinct difference between large and small manufacturers. Figure

1b that shows how the probabilities change for large versus small firms is more revealing. Large

manufacturers offer more slotting allowances more frequently; however the peak for slotting

allowances for large manufacturers occur at a rating of 4.2, which is much lower than that for

small manufacturers at 5.8. At low values of ratings, large firms offer more slotting allowances

(positive coefficient for large firms), but as the rating goes over 6, the probability of slotting

7 The estimated probabilities track the raw probabilities of slotting allowances observed at different levels of manufacturer rating quite well. 8 We included Large*Rating2as well, but this was not significant. So in all of the remaining results, we do not include an interaction with the squared term. Our results here are consistent with results from separate regressions using Rating and Rating2 for small and large manufacturers as well.

13

allowances for large firms becomes lower than that of small firms (due to the negative

coefficient on Large*Rating).

***Insert Figure 1 here***

Why do small manufacturers increase their likelihood of slotting allowances at higher

ratings than large manufacturers? This could be due to the greater risk perceived by the retailer,

because it can be harder to recover costs from small manufacturers if the product fails; or it could

be due to the greater risk aversion of the small manufacturer who does not offer slotting

allowances at lower levels of confidence (lower ratings) in the potential success of the product or

the dollar value of slotting allowance offered is much smaller..

In contrast to these risk balancing explanations, it is also possible that the retailer

exercises greater power over small manufacturers and extracts slotting allowances even when the

expected chance of success is higher. However, the finding that slotting allowance falls at high

levels of rating even for small manufacturers, suggest that our results are not as consistent with a

retail power explanation as with a risk-balancing explanation (or a possible signaling

explanation, which we discuss next). For instance, according to Chu’s (1992) model, the retailer

should be able to extract more slotting allowances when the product is more likely to be

successful. The decline in slotting allowances as ratings increase is inconsistent with the retailer

power argument. We will revisit this issue in section on “Exercise of Retail Power.”

A3. Slotting Allowances Communicate Manufacturers’ Private Information about New

Product Success to Retailer

Do slotting allowances serve as a signal to communicate positive private information that

a manufacturer has? In order to test this, it is necessary to find variables which correlate with

positive/negative private information and then see how the probability of slotting allowances

changes in response to this variable. We consider two variables: provision of test market data or

not and (ii) advertising for the new product by the manufacturer.

Test Markets

Our data tells us whether manufacturers provide test market data to the retailer. Many

people believe that test markets may serve as substitutes for slotting allowances, because they

can help manufacturers to communicate the potential success of the product without using

14

slotting allowances as a signal. For example, a large wholesaler is quoted in Wilkie et al. (2002)

as saying: “Slotting fees are the result of no test markets. It is a charge for testing the item’s

sales potential for a manufacturer. The higher the risk of failure, the higher the charge” (p. 282).

An alternative viewpoint is that test markets are unlikely to be a credible means by which

manufacturers can communicate any private information about the potential success of a new

product. Chu (1992) suggests several reasons for this: (1) manufacturers may selectively report

only positive test market studies out of multiple studies done (2) they may choose test market

locations where their brand equity is strong in order to make the results appear more favorable

than they truly are (3) manufacturers adjust marketing mix in response to test markets in order to

improve and therefore may believe the potential for success, but this is not truly verifiable.

Whether test markets serve as substitutes for slotting allowances or whether slotting

allowances are needed to complement test markets to credibly communicate positive private

information is an empirical matter. We report the results of the logistic regression including the

test market variable in Table 8. The first column merely looks at how test markets affect slotting

allowances by large and small firms. The positive coefficient on test market suggests that when

small firms show test market information, there is not much credibility and therefore they have to

complement test market information with slotting allowances in order to communicate their

private information. The magnitude of the negative effect of Large*Test Market interaction term

is greater than the positive effect of Test Market, but it is not significantly different from zero.

Either test markets do not have an impact on large firms, or there may be opposing effects at low

and high values of rating, which might be averaged out to zero.

***Insert Table 8 here***

By including interactions of test market and rating, we are able to understand why the net

effect of test markets on large firms is zero. The graphs in Figure 2 help interpret the results.

Only at high levels of manufacturer rating (> 5, which may be related to credibility) do large

firms substitute test markets for slotting allowances. At low levels of credibility, large firms also

complement test market with slotting allowances, just like small retailers. To summarize, Chu

(1992) suspects correctly that test markets are not necessarily credible to retailers, due to their

potential for being biased in favor of the manufacturers who present the results in the manner

most favorable to them. Hence, manufacturers seem to complement test market information with

15

slotting allowances to credibly signal the private information they have in the test market data.

However, for larger firms who retailers rate highly, test markets indeed serve as a substitute for

slotting as conventional wisdom suggests.

***Insert Figure 2 here***

Advertising

Another approach to check support for the signaling theory is to test a prediction

developed by Desai (2000), who argues that manufacturers can use slotting allowances and

advertising to signal potential new product success. However manufacturers would prefer to

signal using advertising, if advertising is effective in raising demand. Thus slotting allowances

and advertising will be substitutes if advertising is effective. It appears reasonable to assume that

large national brand manufacturers are more likely to be effective in their advertising than small

manufacturers. Hence, we hypothesize that slotting allowances and advertising will be better

substitutes for large manufacturers than for small manufacturers. Further, since the level of

advertising is just a commitment by a manufacturer, there is some question of its credibility as a

signal because a manufacturer may not deliver on this commitment. Here again one might argue

that commitments by larger firms are more credible due to their repeated interactions and

potential long-term negative impact on their reputations due to reneging on their commitments.

Smaller firms may be less credible because they are potentially more likely to renege on their

commitments. This problem of moral hazard also reduces the ability of small manufacturers to

effectively use advertising as a signal. We use interaction terms to examine these subtle

predictions.

The results of the logistic regression with advertising variables are reported in Table 9.

First, we use the square root of Gross Rating Points (GRP) in the regression to account for the

decreasing returns to advertising. From column 1, we indeed find that for large firms, slotting

allowances serve as a substitute as predicted by Desai (2000). However, for small firms slotting

allowances and advertising serve as complements. To further understand whether the effects

changed as a function of manufacturer reputation, we include interactions with manufacturer

rating as we did with the test market case. This is reported in column 2. Given the nonlinearities

in the rating relationship, the interactions are hard to interpret graphically. Hence in column 3,

we simplify the model by treating advertising as simply a dummy variable so that we can use the

16

graphical approach that we used earlier to interpret the test market results. Since only 17% of

products have any promised advertising, this dichotomization seems reasonable to facilitate

interpretation. Comparing the results in Columns 2 and 3, we see that the signs of the coefficients

remain the same, except for changes in magnitude due to the categorization of the advertising

variable.

***Insert Table 9 here***

The results are shown graphically in Figure 3. At low levels of ratings, large firms

substitute advertising for slotting allowances. Slotting allowances are effective substitutes to

advertising for large manufacturers in the region where slotting allowances are used as a signal

(i.e., when ratings are low). At higher ratings, there is no significant relationship between slotting

allowances and advertising. A plausible explanation is that at high levels of ratings, slotting

allowances are not necessary as a signal and advertising if any is used to build demand, rather

than to signal high demand. Hence it is reasonable that there is no significant relationship

between slotting allowances and advertising at high ratings.

***Insert Figure 3 here***

In contrast, small firms use slotting allowances as a complement to advertising when their

ratings are low. In this region it is conceivable that small retailers have little credibility and there

is a higher potential for moral hazard (i.e., manufacturers may not deliver on their commitment

to advertise), so small manufacturers cannot simply invest all their resources into advertising that

can build demand, because retailers will not accept the product. Instead small manufacturers

appear to split their marketing budgets across slotting allowances and advertising, because they

want to both build demand and increase the likelihood of retailer acceptance.

As the ratings become larger (in the region where risk-averse small manufacturers offer

slotting allowances normally and where retailers have greater faith in the product and the

potential effectiveness of advertising), slotting allowances indeed become substitutes. Thus we

find support for Desai’s (2000) argument that slotting allowances and advertising will be

substitutes when advertising is effective (or as discussed earlier, when moral hazard concerns are

lower).

B. Anti-Competitive Rationales

17

B1. Manufacturers Seek to Mitigate Retail Competition to Increase Profits

A direct test of whether manufacturers seek to mitigate retail competition can be done by

looking at the relationship between the likelihood of offering slotting allowances and the number

of competing retail stores that have accepted the new product. If the goal of manufacturers is to

mitigate retail competition as in Shaffer (1991), we should expect slotting allowances to increase

as the number of competing stores increase.

Alternatively, we could argue for an opposite relationship between slotting allowances

and competing stores that is consistent with a pro-competitive efficiency rationale. When other

retailers accept the new product, it can be informative to the focal retailer that the new product is

likely to be successful and therefore the need for slotting allowances as a signal is reduced.

Further, the retailer can be at a competitive disadvantage by not carrying the product stocked by

his competitors. Retailers may therefore want to carry the product and we should expect use of

slotting allowances to fall as the number of competing stores increase.

Whether the anticompetitive or pro-competitive effect will dominate is an empirical

question. The results are presented in Table 10. Since it is well known that the number of

competitors have a concave effect on profits, we used the square root of number of competing

stores that have accepted the product in the regression. Based on the first column, we see that

slotting allowances increase only for small manufacturers, the net effect is close to zero for large

manufacturers when competing stores accept the product. Thus it appears that the results are

consistent with Shaffer’s theory for small manufacturers, but not for large manufacturers. This is

a puzzling result.

Investigating how the relationship changes with retailer perceptions of likely product

success, offers additional insights. The results are reported in column 2. Since it is easier to

interpret graphs using categorical variables, we also created a dummy variable as to whether any

competing store has accepted the product. These results are in column 3. The results in columns

2 and 3 are similar, except for scale differences due to the categorization of the competing store

variable. Figure 4 shows the graphs of how the presence of competing retailers differentially

affects large and small retailers.

***Insert Table 10 here***

***Insert Figure 4 here***

18

We find that acceptance by competing stores increases the probability of slotting

allowances for large manufacturers, when their ratings are low. But acceptance at competing

stores has no impact on slotting allowances, when large manufacturers have high ratings. As we

expected from the earlier results in column 1, acceptance by competing stores increases the

probability of slotting allowances for small manufacturers at all levels of ratings. Overall these

results appear to support Shaffer’s theory that slotting allowances are a means to mitigate retail

competition. The information effect seems to be overwhelmed by the need to mitigate retail

competition.

However, the question remains as to why the effect vanishes when large manufacturers

have high ratings. A plausible argument is the following: the expected retail profit is low when

the product is produced by the small manufacturers (due to the greater risk), and by a large

manufacturer who has low ratings. If expected profits are low to begin with, the presence of

retail competition can further suppress retailer profitability. Hence the retailer may not wish to

take up distribution of a product, unless there are some additional incentives such as slotting

allowances. According to this reasoning, while slotting allowances serves to mitigate retail

competition, the effect is not driven by the desire to be anti-competitive, but is a means by which

manufacturers obtain retailer participation to ensure broader distribution for the product (an

efficiency argument). For products from large manufacturers with high ratings that are expected

to be successful, the retailer participation constraint is not binding and hence there is no impact

of competing stores on the likelihood of slotting allowances. Quite interestingly, Desai (2000)

anticipates this empirical finding by theoretically demonstrating that slotting allowances are a

means by which manufacturers seek to mitigate the effect of retail competition in order to ensure

retailer participation.

In summary, our results suggest that Shaffer’s theoretical intuition that slotting

allowances can serve to mitigate retail competition is consistent with our empirical results.

However, we suggest that the mitigation of retail competition is not driven by anti-competitive

objectives (as suggested by Shaffer) as much as it is driven by the objective of manufacturers to

obtain broader retail distribution.

B2. Large Manufacturers Seek to Exclude Small Manufacturers (Competitive Foreclosure)

It is a common complaint of small manufacturers that slotting allowances are used by

larger manufacturers to exclude them from markets. According to Ferrell (2001), dominant

19

manufacturers who have exclusive rights at one retailer, will be more willing to offer slotting

allowances at another retail firmstore. Most observers however believe that exclusive dealing is

rare in grocery markets. A comprehensive analysis of whether large manufacturers pursue

exclusion strategies is beyond the scope of this paper, given the lack of data. A weaker test

would be to see if manufacturers who have been accepted at competing retailer are more likely to

offer slotting allowances to a new retailer to keep out other manufacturers. As we discussed

earlier, larger manufacturers with high ratings do not offer more slotting allowances when they

have been accepted in other stores. This does not provide support for the competitive foreclosure

story.

We however wish to add the caveat that the exclusion argument is usually made for

existing products that have already established dominant market share and not for new products

that are still to establish market presence. Nevertheless, it is worthwhile to empirically document

that at the new product introduction stage, exclusion of small manufacturers does not appear to

be the motivation for slotting allowances.

B3. Exercise of Retail Power

We discussed in the section on risk balancing (A2) that our finding that small

manufacturers offered slotting allowances at high levels of ratings (relative to large

manufacturers) is more consistent with a risk-balancing argument than an exercise of retailer

power. We now discuss this issue again in the context of our results on the impact of acceptance

by competing stores on the slotting allowance offers.

Iyer and Villas-Boas (2003) argue that retailer power over manufacturers falls when

manufacturers have alternative outlets for distribution. Our empirical results suggest that the

likelihood of slotting allowances increases for small manufacturers when they have alternative

outlets, a result that is inconsistent with the retailer power rationale. As we argued earlier, our

result is consistent with these manufacturers trying to obtain wider distribution.

Nevertheless, we believe that additional empirical tests about the exercise of retail power

are needed to investigate how slotting allowances change as retailer characteristics change before

we can conclusively rule out the retailer power explanation. This would require more

comprehensive data from a cross-section of retailers, so that we can study whether slotting

allowances across retailers increase as a function of retailer power. However, such an exercise

20

has to be done very carefully to avoid the confounding between greater power and greater

opportunity cost. For example, a large retailer may be considered powerful due to its size, but the

opportunity cost of its shelf space will be greater as well.

Robustness Checks

Endogeneity of Manufacturer Ratings

Our empirical analysis has focused on how the likelihood of slotting allowances changes

with the retailer’s ratings of small and large manufacturers in the presence of test market

information, advertising and competing stores. The manufacturer rating variable measures the

manufacturer’s likelihood of success on the new product based on past experience with the

manufacturer and industry reputation. So it was reasonable to assume that these ratings would be

given independent of the characteristics of the transaction in question: whether test market

information was shown, whether advertising gross rating points are promised or whether

competing stores have accepted the product. Nevertheless, since buyers rate the manufacturer

after the terms of trade are revealed, it is possible that the terms of trade affect manufacturer

ratings.

We check for this potential endogeneity of manufacturer ratings. The regression results

are reported in Table 11. We find an interesting difference between small and large

manufacturers. Test markets, advertising and competing store acceptance all affect the ratings of

small manufacturers, but the net effect on large manufacturers is close to zero. In hindsight, these

results are not surprising, because the retailer has more past experience with large manufacturers

and their industry reputations are relatively strong, suggesting that the effect of the current

transaction’s trade terms on ratings are minimal. In contrast, the retailer’s experience is likely to

be lower with small manufacturers on average and their industry reputations are not as well

defined. Hence from a Bayesian updating perspective it makes sense that the ratings of small

manufacturers are affected more by the current transaction’s trade terms.

***Insert Table 11 here***

This endogeneity however suggests that we need to work in terms of the retailer’s apriori

manufacturer ratings of the manufacturers rather than the posterior ratings after the retailer’s

knowledge of the terms of trade offered by the manufacturer. Using the regression equation in

Table 11, we compute the apriori ratings and did the entire analysis as before with the apriori

21

ratings. Though the coefficient estimates do change somewhat due to the changes in the rating

variable used, none of the qualitative insights we obtained earlier are changed.9 The graphs are

similar to Figures 1-4 as well.

Simultaneous Inclusion of All Terms of Trade

In the analysis so far, we include test markets, advertising and competing stores piece-

meal into each logistic regression. It is reasonable to question whether the effects would persist if

we included all the variables simultaneously. We report the results from such a regression in

Table 12. The first column contains the results with all variables included. In the remaining three

columns we report the previous results from Tables 8, 9 and 10 for comparison. From the

comparison, the estimates in the first column are qualitatively similar to the estimates in the last

three columns, except for the “Large” and “Large*Rating” variables which are affected by the

inclusion of other correlated variables (e.g. all other interaction variables involving Large).

Essentially, our results are robust to the inclusion of all effects simultaneously.

***Insert Table 12 here***

5. Conclusion

This paper provides the first empirical investigation of multiple rationales presented in

the literature for the use of slotting allowances in new product introductions. For this purpose,

we use a unique dataset that consists of all new products that were offered to a retailer during a

period of 6 months (some of which had slotting allowances and others that did not). Our

empirical analysis suggests that slotting allowances are efficiency-enhancing and not anti-

competitive. Specifically, we conclude that:

(1) Slotting allowances serve to efficiently allocate scarce retail shelf space.

(2) Slotting allowances help to balance risk by shifting the downstream risk of retailers upward

towards manufacturers.

(3) Slotting allowances are offered in a manner consistent with predictions of signaling theories.

(4) Though we find that slotting allowances serve to mitigate retail competition, our results

suggest that the impetus for this is to ensure broader retail distribution rather than the anti-

competitive rationale attributed in the extant literature.

(5) We do not find support for the manufacturer exclusion and retailer power rationales.

9 The detailed results are available from the authors upon request.

22

Our study suggests two major takeaways for any future empirical analysis of slotting

allowances. First, the effect of critical variables (test markets, advertising, competing stores) on

offers of slotting allowances are a function of the perceived likely success of the product. As we

discovered, these effects are highly nonlinear and therefore require to be carefully controlled.

Second, the role of slotting allowances is quite different for large versus small

manufacturers. In extant opinion-based survey research, this distinction has not been taken into

account. Even though Bloom et al. (2000) and Wilkie et al. (2002) collect information about the

size of manufacturers, wholesalers and retailers, they report only average estimates of survey

responses across all respondents. Rao and Mahi (2003) also do not make this distinction in their

analysis. Our results show that due to the disproportionate number of small manufacturers in any

sample, the average results will be weighted towards the effects of small manufacturers. In our

data, if we did not make the distinction between small versus larger manufacturers, we find that

the results are by and large consistent with the results that we report for small manufacturers;

essentially the results are hijacked by the effects of small manufacturers who dominate the

sample. Their impact on the econometric results is disproportionate relative to their economic

impact. The several cross-over interaction effects we discover for large and small manufacturers

over high and low values of manufacturer ratings (see Figures 1-4), clearly demonstrates the

importance of these two ideas in the empirical analysis of slotting allowances.

Limitations and Future Research

Admittedly, we analyze data from just one retailer. While we have suggested that it is

possible to cautiously generalize to other large regional retail chains, it is important to investigate

whether our results continue to hold at other retailers. Just as we found that behavior of small and

large manufacturers differ in many subtle and important ways, it is quite possible that behavior

of retailers would differ depending on their strengths. Therefore, behavior might be different at

national level retailers or smaller independent grocery stores and therefore this would require

further study.

Another important caveat is that the data are from the period 1986-87. It is a fact that the

intensity of slotting allowances has been rising over the last 15 years. It is possible that rationales

underlying slotting allowances could have changed over this period. It would require newer data

to investigate whether these rationales continue to hold. In that sense, our study provides a

23

historical snapshot. However, we note that many of these rationales are not entirely new and

have been discussed in many articles in the trade press during the early nineties.

Future research should investigate how the magnitude of slotting allowances changes in

response to the variables studied. Fortunately, the empirical framework can be quite similar to

what we use in this study. We need to replace the logistic regression, with a tobit-regression if

we have data on the magnitude of slotting allowances.

In conclusion, our study is the first to provide a systematic empirical approach to test

alternative rationales for the use of slotting allowances. While our results permit us to conclude

that efficiency rationales are more at work than anti-competitive rationales, much more work

needs to be done using more recent data from a larger sample of retailers before this is

conclusively accepted. Nevertheless, our findings supporting the efficiency enhancing aspects of

slotting allowances show that the FTC was correct in its reluctance to ban the practice of slotting

allowances in the grocery sector.

24

References Bloom, P.N., Gundlach, G. T. and J.P. Cannon (2000), “Slotting Allowances and Fees: Schools

of Thought and the Views of Practicing Managers,” Journal of Marketing, 64(April), 92-

108.

Cappo, J. (2003), “How Retailer Power Changes Marketing,” Advertising Age, July 21, 2003.

Chu, W. (1992), “Demand Signaling and Screening in Channels of Distribution,” Marketing

Science, 11(4), 324-47.

Deloitte and Touche (1990), “Managing the Process of Introducing and Deleting Products in the

Grocery and Drug Industry,” Washington D.C.: Grocery Manufacturers of America.

Desai, P. (2000), “Multiple Messages to Retain Retailers: Signaling New Product Demand,”

Marketing Science, 19(4), 381-89.

Desiraju, R. (2001), “New Product Introductions, Slotting Allowances and Retailer Discretion,”

Journal of Retailing, 77(3), 335-58.

Farris, P.W. and K.L. Ailawadi (1992), “Retail Power: Monster or Mouse?” Journal of Retailing,

92(Winter), 351-369.

Federal Trade Commission (2001), “Report on the Federal Trade Commission Workshop on

Slotting Allowances and Other Marketing Practices in the Grocery Industry,” Washington

D.C.: U.S. Government Printing Office.

Fields, R. and M. Fullmer (2000), “Markets’ Shelf Fees Put Squeeze on Small Firms,” Los

Angeles Times, (January 29), Home Edition, Part A, I.

Food Marketing Institute (2003a), “Food Marketing Industry Speaks: The State of the Food

Retail Industry,” Report.

Food Marketing Institute (2003b), “New Products and Services,” (accessed March 10, 2003),

[available at http://www.fmi.org/facts_figs/newproductsandservices.pdf].

Gundlach, G. T. and P.N. Bloom (1998), “Slotting Allowances and the Retail Sale of Alcoholic

Beverages,” Journal of Public Policy and Marketing, 17(2), 173-84.

Kaufman, P.R., (2000), “Consolidation in Food Retailing: Prospects for Consumers and Grocery

Suppliers,” Agricultural Outlook, Economic Research Services USDA, (August), 18-22.

Kelly, K. (1991), “The Antitrust Analysis of Grocery Slotting Allowances: The Pro-competitive

Case,” Journal of Public Policy and Marketing, 10 (1), 187-98.

25

Lariviere, M.A. and V. Padmanabhan (1997), “Slotting Allowances and New Product

Introductions,” Marketing Science, 16(2), 112-28.

Messinger, P.R. and C. Narasimhan (1995), “Has Power Shifted in the Grocery Channel,”

Marketing Science, 14(2), 189-223.

Rao, A. and H. Mahi (2003), “The Price of Launching a New Product: Empirical Evidence on

Factors Affecting the Relative Magnitude of Slotting Allowances,” Marketing Science,

22(2), 246-68.

Rao, V.R., and E. W. McLaughlin (1989), “Modeling the Decision to Add New Products by

Channel Intermediaries,” Journal of Marketing, 53 (1), 80-88.

Shaffer, G. (1991), “Slotting Allowances and Retail Price Maintenance: A Comparison of

Facilitating Practices,” RAND Journal of Economics, 22(1), 120-35.

Sullivan, M. W. (1997), “Slotting Allowances and the Market for New Products,” Journal of

Law and Economics, 40(2), 461-93.

Swenso, L (2000), “Testimony before the Judiciary Subcommittee on Antitrust,” (U.S. Senate),

statement of Leland Swenson, President National Farmers Union, (September 28).

Thompson, S. (2000), “Grocers’ Rising ‘Slotting’ Fees Drive Unusual Ad Campaigns,” Crain’s

Chicago Business, 23(45), 44-45.

Vosburgh, R. (2001), “Report Digs Deeper into Produce Slotting Fees,” Supermarket News,

(February 12), 29.

Wilkie, W. L., D.M. Descrochers and G.T. Gundlach (2002), “Marketing Research and Public

Policy: The Case of Slotting Fees,” Journal of Public Policy and Marketing, 21(2), 275-88.

26

Table 1: Product Details by Category

Category Total

Products Slot (%)

Baby Foods 7 43 Baking Ingredients 35 14 Beverages 85 16 Breads, Cakes, Cookies 31 3 Breakfast Cereals 19 5 Candy/Gum 52 8 Canned Fruits/Vegetables/Juice/Drinks 142 24 Canned Meat/Fish 5 60 Dairy and Refrigerated Foods 124 2 Dessert Powders/Sugars/Syrups/Spreads 7 0 Frozen Foods 219 17 Health and Beauty Aids 3 67 Household Supplies 65 20 Diet Foods 3 33 Macaroni, Potatoes and Rice 36 14 Paper Products 7 14 Pet Products 57 12 Sauces, Spices, Condiments, Oils, Dressings 62 5 Snacks, Crackers, Nuts 47 13 Soups 12 0 Tobacco Products 3 0

27

Table 2: Means by Slotting Allowance

Table 3: Correlations

Table 4: List of Large Firms in Data (The Large firms are all members of Grocery Manufacturers Association)

Slotting Allowance Not Offered Offered Opportunity Cost 1.30 1.72 Private Label 2.34 2.60 Vendor Reputation 5.77 5.67 Advertising GRP 2.20 5.58 Test Market 0.17 0.35 Competing Stores 1.99 3.60 Large Manufacturers (%) 83 17 Small Manufacturers (%) 87 13

Opport. Cost

Private Label

Rep-utation

GRP Test Market

# Comp Stores

Large Mfr

Slot

0.11 0.06 -0.02 0.13 0.16 0.13 0.05

Opport. Cost

0.00 0.04 0.05 0.11 0.04 0.04

Private Label

-0.04 0.02 -0.04 0.26 -0.04

Reputation

0.04 0.13 0.12 0.19

GRP

0.31 0.13 0.04

Test Market

0.03 0.18

#Comp Stores

0.05

Campbell Land O'Lakes Carnation Lever Brothers Clorox M&M/Mars Coca Cola McCormick Colgate Palmolive Nabisco Dannon Nestle Del Monte Pillsbury General Foods Poland Springs General Mills Procter and Gamble H.J. Heinz Quaker Oats Hershey Ragu Foods Inc Kellogg's Scott Paper Kraft Tropicana

28

Table 5: Opportunity Costs (***: p < 0.01; ** p < 0.05; * p <0.1)

Item-Specific Margin in

Opportunity Cost

Category Margin in

Opportunity Cost

Variable Estimate

(s.e.) Estimate

(s.e.)

Intercept -2.56***

(0.25) -2.65***

(0.26)

Opportunity Cost 0.12***

(0.04) 0.48***

(0.19)

Competing Private Label 0.66***

(0.25) 0.60** (0.25)

ρ2 0.02 0.02

Table 6: Rating of Products from Selected Manufacturers

Vendor Rating Del Monte General Foods Kraft P&G

Quaker Oats Seneca Foods

8 Shredded Sharp Cheddar

Variety Pack Granola Bars

7

Mexican Style Stewed Tomatoes

Birdseye Baby Broccoli Spears

Cremerie French Onion

Liquid Cheer 128, 96, 64 oz

Ken-L Pupperonis Bacon

Cranberry Juice Cocktail

6 Cut Green Beans

Kool-aid Koolers

Kraft Extra Thick American Singles White Cloud

Aunt Jemina Family Waffles

64 Oz Cranberry-Apple Juice Cocktail

5

Sierra Trail Mix-Pineapple Nuggets

Stove Tip Chicken Shipper

Grated Parmesan Display Unit

Pampers Ultra-Small

Ken-L Hearty Chunks Chicken

Frozen White Grape Juice

4

Hawaiian Punch Lite-Tropical Fruit

Carroll Shellby's Texas Chili

Puffs Dispenser Pack

Quaker Oh's Honey Graham

3

Duncan Hines Brownies

2

Stove Top Flexible Serving Chicken

29

Table 7: Risk Balancing (***: p < 0.01; ** p < 0.05; * p <0.1)

Variable Estimate

(s.e.) Estimate

(s.e.)

Intercept -8.22***

(1.65) -10.17***

(1.82)

Manufacturer Rating 2.55***

(0.60) 3.01***

(0.63)

Manufacturer Rating2 -0.24***

(0.05) -0.26***

(0.05)

Large 4.89***

(1.12)

Large * Rating -0.79***

(0.19)

ρ2 0.04 0.06

Table 8: Role of Test Markets (***: p < 0.01; ** p < 0.05; * p <0.1)

Variable Estimate

(s.e.) Estimate

(s.e.) Intercept -10.43***

(1.90) -10.92***

(1.93) Manufacturer Rating 3.06***

(0.65) 3.49***

(0.69) Manufacturer Rating 2 -0.27***

(0.06) -0.33***

(0.06) Large 4.97***

(1.15) 3.03***

(1.39) Large * Rating -0.72***

(0.20) -0.37

(0.24) Test Market 1.52***

(0.25) -2.76* (1.55)

Large * Test Market -1.80*** (0.44)

5.69** (2.62)

Test Market * Rating

0.73*** (0.26)

Large *Test Market * Rating

-1.31*** (0.46)

ρ2 0.1 0.12

30

Table 9: Role of Advertising

(***: p < 0.01; ** p < 0.05; * p <0.1)

GRP GRP GRP

Dummy Variable Estimate

(s.e.) Estimate

(s.e.) Estimate

(s.e.) Intercept -9.51***

(1.82) -9.82***

(1.95) -10.61***

(1.99) Manufacturer Rating 2.73***

(0.63) 2.68***

(0.66) 2.97***

(0.67) Manufacturer Rating 2 -0.24***

(0.05) -0.22***

(0.06) -0.24***

(0.06) Large 4.93***

(1.11) 6.56***

(1.23) 7.21***

(1.35) Large * Rating -0.75***

(0.19) -1.03***

(0.21) -1.16***

(0.24) GRP 0.04***

(0.01) 0.45***

(0.12) 6.64***

(1.29) Large * GRP -0.07***

(0.03) -0.79***

(0.23) -11.02***

(2.73) GRP * Rating -0.07***

(0.02) -1.04***

(0.30) Large *GRP * Rating 0.12***

(0.04) 1.74***

(0.47)

ρ2 0.08 0.11 0.09

31

Table 10: Role of Store Competition (***: p < 0.01; ** p < 0.05; * p <0.1)

#Competing Stores #Competing Stores

Competing Store

Dummy Variable Estimate

(s.e.) Estimate

(s.e.) Estimate

(s.e.) Intercept -10.21***

(1.85) -10.16***

(1.83) -9.85***

(1.83) Manufacturer Rating 2.90***

(0.63) 3.19***

(0.65) 3.07***

(0.65) Manufacturer Rating 2 -0.25***

(0.05) -0.31***

(0.06) -0.29***

(0.06) Large 5.09***

(1.13) 3.54***

(1.37) 2.96** (1.47)

Large * Rating -0.75*** (0.20)

-0.47** (0.24)

-0.38 (0.26)

Competing Stores 0.41*** (0.08)

-0.92** (0.47)

-2.06 (1.36)

Large * Competing Stores -0.37*** (0.16)

1.27 (0.97)

4.14* (2.32)

Competing Stores * Rating 0.23*** (0.08)

0.53** (0.24)

Large * Competing Stores * Rating

-0.28* (0.16)

-0.84** (0.40)

ρ2 0.09 0.11 0.09

32

Table 11: Effect of Terms of Trade on Manufacturer Reputation (***: p < 0.01; ** p < 0.05; * p <0.1) Variable Estimate

(s.e.) Intercept 5.29***

(0.07) Large 0.98***

(0.16) Test Market 0.53***

(0.16) GRP Dummy 0.50***

(0.16) Competing Store Dummy

0.53*** (0.11)

Slotting Allowance -0.35*** (0.13)

Large * Test Market -0.64*** (0.25)

Large * GRP -0.57*** (0.25)

Large * Competing Stores

-0.43*** (0.21)

ρ2 0.09

33

Table 12: All Effects Simultaneously Included

(***: p < 0.01; ** p < 0.05; * p <0.1)

All

Variables Test

Market Advertising Compe ting

Stores

Variable Estimate

(s.e.) Estimate

(s.e.) Estimate

(s.e.) Estimate

(s.e.)

Intercept -13.03***

(2.40)

-10.92***

(1.93) -10.61***

(1.99) -9.85***

(1.83)

Manufacturer Rating 4.30***

(0.86) 3.49***

(0.69) 2.97***

(0.67) 3.07***

(0.65)

Manufacturer Rating 2 -0.42***

(0.08) -0.33***

(0.06) -0.24***

(0.06) -0.29***

(0.06)

Large 1.65

(2.63) 3.03***

(1.39) 7.21***

(1.35) 2.96** (1.47)

Large * Rating -0.11

(0.46) -0.37

(0.24) -1.16***

(0.24) -0.38

(0.26)

Test Market -3.48***

(1.58) -2.76* (1.55)

Large * Test Market 14.77***

(3.85) 5.69** (2.62)

Test Market * Rating 0.88***

(0.27) 0.73***

(0.26)

Large *Test Market * Rating

-3.05*** (0.71)

-1.31*** (0.46)

GRP 7.95***

(1.82) 6.64***

(1.29)

Large * GRP -15.32***

(3.33) -11.02***

(2.73)

GRP * Rating -1.36***

(0.32) -1.04***

(0.30)

Large *GRP * Rating 2.59***

(0.57) 1.74***

(0.47)

Competing Stores -1.82

(1.49) -2.06

(1.36)

Large * Competing Stores 5.57* (3.27)

4.14* (2.32)

Competing Stores * Rating 0.47* (0.26)

0.53** (0.24)

Large * Competing Stores * Rating

-1.02* (0.56)

-0.84** (0.40)

ρ2 0.18 0.12 0.09 0.09

34

Figure 1a: Risk Balancing

Figure 1b: Risk Balancing: Large Versus Small

Figure 2a: Role of Test Markets: Large Manufacturers

Figure 2b: Role of Test Markets: Small Manufacturers

0.000

0.100

0.200

0.300

0.400

0.500

0.600

1 2 3 4 5 6 7 8 9 10

Rating

P(S

lott

ing

Allo

wan

ces)

Large Large*TestMkt

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

1 2 3 4 5 6 7 8 9 10

Rating

P(S

lott

ing

Allo

wan

ces)

Small Large

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

1 2 3 4 5 6 7 8 9 10

Rating

P(S

lott

ing

Allo

wan

ces)

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

0.450

0.500

1 2 3 4 5 6 7 8 9 10

Rating

P(S

lott

ing

Allo

wan

ces)

Small Small*TestMkt

35

Figure 3a: Role of Advertising: Large Manufacturers

Figure 3b: Role of Advertising: Small Manufacturers

Figure 4a: Role of Competing Stores: Large Manufacturers

Figure 4b: Role of Competing Stores: Small Manufacturers

0.000

0.100

0.200

0.300

0.400

0.500

0.600

1 2 3 4 5 6 7 8 9 10

Large Large*Ad

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

0.450

0.500

1 2 3 4 5 6 7 8 9 10

Small Small*Ad

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

1 2 3 4 5 6 7 8 9 10

Rating

P(S

lott

ing

Allo

wan

ces)

Small Small*CS

0.000

0.100

0.200

0.300

0.400

0.500

0.600

1 2 3 4 5 6 7 8 9 10

R a t i n g

Large Large*CS