Contracting in the wine supply chainwith bilateral moral hazard residualclaimancy and multi-tasking

Bodo Steiner

University College Cork Ireland

Received June 2009 final version accepted March 2011

Review coordinated by Thomas Heckelei

Abstract

This paper takes a quasi-case-study approach to stylised wine industry facts to assesspredictions about the optimal sharing rule from a principalndashagent model with residualclaimancy An optimal sharing contract is developed between a grape grower and awinery when a risk-averse grower allocates efforts among multiple activities thatdiffer in measurability while double-sided moral hazard is assumed to be presentSeveral comparative static results regarding the Pareto optimal share are in linewith certain production practices and properties of observed contracts that arefound in markets where residual claimancy is used namely in Australia CaliforniaNew Zealand and France

Keywords incentive contract residual claimancy wine double-sided moral hazardmulti-tasking supply chain

JEL classification L22 M31 D23

1 Introduction

The choice of appropriate performance indicators aligning an agentrsquos actionswith a principalrsquos objectives is one of the central problems that organisationsface in implementing effective incentive contracts particularly when agentsperform multiple tasks (Holmstrom and Milgrom 1991) In such an environ-ment of multi-tasking issues of internal organisational design of the firm mayarise due to effort substitution in particular when tasks differ in performancemeasurability (eg Holmstrom and Milgrom 1991 Slade 1996) Consideringagricultural production producers frequently need to allocate their effortsacross a variety of tasks that differ in the measurability of their impact onthe final goodrsquos quality When measurability differs between multiple tasks

Corresponding author Department of Food Business amp Development 2nd Floor OrsquoRahilly

Building University College Cork Cork Ireland E-mail bsteineruccie

European Review of Agricultural Economics Vol 39 (3) (2012) pp 369ndash395doi101093eraejbr054Advance Access Publication 19 October 2011

Oxford University Press and Foundation for the European Review of Agricultural Economics 2011 all rightsreserved For permissions please email journalspermissionsoupcom

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

so that the resulting measurement bias can lead to distortions of efforts thetheoretical multi-tasking literature has established that the optimal contractoffers weaker overall incentives (eg Holmstrom and Milgrom 19911994) Ackerberg and Botticini (2002) econometrically explore the role ofmulti-tasking in the context of the wine industry and thereby add empiricalevidence that multi-tasking can be an important factor in contract choice

This paper expands the analysis of multi-tasking to a subset of contracts inthe wine industry in order to shed further light on organisational design issueswhen bilateral moral hazard is present and residual claimancy is employedSimilar to the approach of Hennessy and Lawrence (1999) and Hueth andMelkonyan (2002 2004) the paper takes a quasi-case-study approach toanalyse contract specifications and their relationship with performancemeasurement and control in the wine industry Stylised facts are employedfrom previous industry studies to explore the applicability of model predic-tions to a subset of firms in the wine industry By exploring the extent towhich this paperrsquos theoretical framework fits these stylised facts fromprevious wine industry studies and by extending previous theoretical predic-tions beyond a risk-neutral and single-task setting this paper makes both anempirical and a theoretical contribution

The paper develops a sharing incentive contract where residual claimancyis used through bottle-price indexing to align principal and agent and whereboth agent (grape grower) and principal (winery) are assumed to performmultiple tasks It is assumed that a grape grower contributes to final winequality in terms of production efforts and that the winery contributes interms of processing and marketing efforts Since it is further assumed thatefforts are mutually imperfectly observed and their impact on final bottlequality can only be imperfectly measured opportunism and monitoring aremodelled on both sides

While opportunism of grape growers could be associated with their inputapplications opportunism of wineries could be associated with marketinggrapes when marketing effectiveness cannot be fully observed by a grapegrower or it could relate to grape quality assessment efforts incurred by thewinery As Fraser (2003) suggests wineries have incentives to underestimategrape quality since it lowers the price they have to pay to supplying growers

Given these information problems the model accounts for monitoring andresidual claimancy There are further theoretical and empirical reasons toaccount for monitoring starting with the observation that the relationshipbetween monitoring and residual claims is linked to the very rationale forthe existence and organisation of the classical firm (Alchian and Demsetz1972) Further Holmstromrsquos (1979) sufficient statistic result implies thatoutcome conditioning (eg via bottle price indexing) is not to be used inisolation but in combination with input monitoring as long as monitoringis informative

We observe that the amount of monitoring of grape growers and the inten-sity of incentives are jointly chosen in actual grape supply contracts (Boydet al 2000 Fraser 2005) This observation is also supported by theory as

370 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

setting intense incentives and carefully measuring performance can beEdgeworth complements (Milgrom and Roberts 1995)

Further monitoring can be rationalised in cases where there is scope forfree-riding due to commingling (Holmstrom 1982) for example when asingle grape grower supplies multiple grape varieties or when multiplegrowers supply their grapes to a single winery Considering the nature of abottle-price indexed contract monitoring could also be rationalised becauseboth parties may be reluctant to condition the sharing rule too heavily onthe market valuation of the final bottle since market risk and other exogenousfactors make it desirable to construct a performance measure that is moreclosely tied to both the individual growerrsquos and wineryrsquos contributions

2 Literature

Several aspects differentiate the model in this paper from previous modelsand analyses of double-moral hazard and multi-tasking As in Holmstromand Milgrom (1991) the desirability of providing incentives for any oneactivity decreases with the difficulty of measuring performance in any otheractivity that makes competing demands on an agentrsquos effort But whereasHolmstrom and Milgrom (1991) allow for a risk-averse agent in the presenceof hidden action of agents only this paper considers double-sided moralhazard and introduces a sharing contract to provide joint incentives throughpartial residual claimancy The two latter assumptions are also explored inBhattacharyya and Lafontaine (1995) and Brickley (2002) HoweverBhattacharyya and Lafontaine (1995) explore the nature of share contractsin franchising under risk neutrality in the absence of multi-tasking andinput monitoring Although Brickley (2002) allows for risk-averse agentsthe paper considers neither multi-tasking nor monitoring The model in thispaper includes input conditioning through effort monitoring and assumesmulti-tasking and a risk-averse agent Finally and in contrast to the above-mentioned papers the model specification also permits an analysis of strictsharing in double-moral hazard settings when agents are risk-averse

There are several multi-tasking studies with empirical applications outsideof the wine sector to which this paper is related to including the analysis ofSlade (1996) on gasoline-service stations and Hueth and Melkonyanrsquos(2004) quasi-case-study approach to identity preservation issues in a multi-tasking model where agents perform one inside and one outside activity inthe context of the sugar beet and processed vegetable industries

Since the following model is based on a second-best sharing contract in thecontext of multi-tasking it also closely relates to the literature on sharecontracts (eg Eswaran and Kotwal 1985 Luporini and Parigi 1996 Danaand Spier 2001) Pure risk sharing has been shown to motivate sharingarrangements when both agent and principal are risk averse and benefitfrom insurance (eg Stiglitz 1974) but one-sided moral hazard has alsobeen put forward to rationalise share contracts (eg Stiglitz 1974 Mathewsonand Winter 1985) Furthermore double-sided moral hazard as an explanation

Contracting in the wine supply chain 371

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

for sharing contracts has not only been employed in the economics of share-cropping (Stiglitz 1974 Eswaran and Kotwal 1985) but it has also been putforward to explain revenue sharing in the context of franchising and supplychain contracts (eg Rubin 1978 Lal 1990 Bhattacharyya and Lafontaine1995 Brickley 2002 Corbett et al 2005)

The remainder of the paper is structured as follows Section 3 providesbackground information about the wine industry focusing on contractual pro-visions used in real-world contracts Section 4 develops a sharing contractunder the assumptions of multi-tasking double-moral hazard and effort moni-toring and derives propositions which are contrasted with stylised facts fromprevious empirical wine industry studies Section 5 concludes the paper

3 Background contractual use in the wine industry

Formal wine grape supply contracts are used extensively in many key wineproducing regions including Australia (Fraser 2005) California (Moulton1988 Bedwell 2000 Goodhue et al 2002 2004) Argentina (Fares et al2002) Brazil (Zylbersztajn and Miele 2005) New Zealand (Boyd et al2000) France (Montaigne and Sidlovits 2003) and Spain (Olmos 2008)As a result of producer surveys the use of contracts has perhaps been bestdocumented in the case of California (eg Goodhue et al 2004) and Australia(eg Fraser 2005)

31 Industry structure

In several key producing regions where formal contracts are in use the wineindustry has expanded significantly over the past two decades resulting inchanges in market structure In Australia the number of growers has increasedby more than 30 per cent to 4822 between 1994 and 1998 At the same timethe number of wineries expanded by nearly 50 per cent to 1197 establish-ments (Shepherd and OrsquoDonell 2001) By 2005 the industry had expandedto approximately 6000 grape growers and 2000 wineries (AWBC 2007RMN 2007) In California during 2005 there were 2275 wineries and4600 grape growers (2972 wineries in 2009 Wine Institute 2011) comparedwith 750 wineries and about 5600 growers in 1987 (Moulton 1988 WineInstitute 2007a) Since vineyards also expanded considerably (1988297000 acres 2005 445141 acres Wine Institute 2007b) the growersrsquoaverage vineyard size increased significantly (by 73 per cent) At the sametime wine processing has become more consolidated in both California(Goodhue et al 2008) and Australia (Smith and Marsh 2007) Consideringthe above evidence on the differences in the total number of growers versusthe total number of wine-processing firms (wineries) in a given region ndashsuggesting that wineries are typically much larger operations than growersndash it may be a reasonable assumption that a winery is generally less riskaverse compared with a single grower (risk aversion is typically assumed tobe inversely related to wealth Laffont and Matoussi 1995 390) Thus for

372 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

the following model the simplifying assumption of a risk-neutral winery ismade an assumption which has also been applied to processors inprincipal-agent models that were employed to study other agriculturalmarkets (Hueth and Ligon 2002)

However the above figures on firm numbers and size also raise the questionof whether the bargaining power between grape growers and wineries istypically evenly or unevenly distributed when grape contracts are settledThis question is relevant with regard to the following model since it implicitlyassumes equal bargaining power between principal and agent Evidence fromCalifornia New Zealand and Chile suggests that bargaining power is not onlya function of scale (on both the growerrsquos and wineryrsquos side) but also a func-tion of perceived quality as greater grower bargaining power was found inhigh-quality grape regions (Moulton 1988 Gwynne 2006) Fairweatheret al (1999) provide an extensive discussion of the New Zealand wine indus-try on the issue of tension between contracting grape growers and wineriesTheir analysis suggests that there is no evidence for a one-sided excessivebargaining power Davis and Ahmadi-Esfahani (2005) suggest that a recentgrape excess supply in Australia may also have been caused by lucrativegrape contracts which implies that wineries have not consistently extractedrents at the expense of growers Other evidence from Australia (Scaleset al 1995) also suggests that the bargaining power of wineries although gen-erally of concern to growers has at times shifted towards growers Scales et al(1995) report that grape growersrsquo bargaining power was found to havestrengthened during times of growing export opportunities for wine andwhen alternative markets expanded for grapes (ie markets for dried vinefruits) Nevertheless we have evidence that the bargaining power rests onthe buyerrsquos (wineryrsquos) side in periods of excess (grape) supply (Fraser2005) Other documented evidence on the extent of bargaining power differ-ences and their implications for contracting and pricing are scarce A study ofthe New York State wine industry used a small survey among wineries toeconometrically explore the relationship between grape prices and prices ofwine (Hefetz and White 1999) The study concludes that lsquoThe clear and sig-nificant relations between retail prices and grape prices result from the sharingof revenue from wine sales between the grape growers and the wine makersrsquo(Hefetz and White 1999 16) Thus the study provides an example of aseemingly balanced bargaining power between growers and wineries

32 Type of contract and contractual provisions

Surveys in the main grape-growing regions of Australia and California foundthat 85 per cent (2001) and 72 per cent (1999) respectively of growers havewritten contracts (Goodhue et al 1999 Fraser 2005) These contracts aretypically written over the supply of bulk wine over grape must or overfresh grapes and have been observed to be offered on a take-it-or-leave-itbasis (eg Sidlovits and Kator (2007) for documented evidence fromHungary) In California fresh grape contracts between growers and wine

Contracting in the wine supply chain 373

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

processors are most frequently used (Goodhue et al 1999) From Fraser(2005) and Boyd et al (2000) we have evidence that these contracts havetypically a relatively low degree of customisation which supports themodelling assumption of simple uniform linear compensation schemesemployed in this study

Typical contract specifications include provisions on production practice(viticultural management for example the documented use of chemicalpesticides in grape production as evidence from Spain Australia and Cali-fornia suggests Goodhue et al 1999 Fraser 2005 Olmos 2008 respect-ively) price incentives (bonusespenalties for quality attributes of freshgrapes) and monitoring (Moulton 1988 Boyd et al 2000 Goodhueet al 2002 Benavente 2004 Fraser 2005 Olmos 2008) or a subsetthereof However Goodhue et al (2004) suggest that requiring specificpractices is uncommon in California The same observation is made byScales et al (1995) in the context of Australia Considering this obser-vation together with the fact that winegrapes are not necessarily specificto a given vintner (Goodhue et al 2004) it appears that relation-specificassets and the role of quasi-rents in grape contracting are generally notlikely to be of major importance in the wine industry This observationmay lend support to the use of a standard linear principal agent model(Section 4) rather than analysing wine grape contracting through thelens of the property rights approach (eg Grossman and Hart 1986Hart and Moore 1990)

Considering further contractual provisions we have no evidence that grapesupply contracts control for exogenous weather variables (rainfall temperature)despite expectations in this respect that arise from Holmstrom (1979) as well asfrom previous empirical studies by Ashenfelter et al (1995) and Byron andAshenfelter (1995) who show that these weather variables can have a signifi-cant impact on price and wine quality

Grape grower monitoring in the form of winery fieldmen is used extensivelyby the wineries (Boyd et al 2000 Goodhue et al 2002 Fraser 2005Zylbersztajn and Miele 2005 Olmos 2008) As a result of such monitoringefforts wineries have been observed to generate historical performance scor-ecards for individual growers which are used when new contracts are put intoplace (eg Zylbersztajn and Miele 2005) However we have also evidence forwinery monitoring such that growers infer winery processing and marketingefforts from trade publications winery reports and other industry participantsWineries also submit reports to growers about the composition of their grapejuice or about results from wine tastings (Montaigne and Sidlovits 2003Omond 2003)

33 Use of residual claimancy in contracts

We expect to observe the use of residual claimancy in the wine industryunder certain conditions When free-riding occurs as a function ofcommingling grapes from several growers during processing the

374 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

underlying moral hazard problem (Holmstrom 1982) does not lend supportto the use of contracts conditioning grape grower payments on the retailvalue of their individual output However when moral hazard remainsan issue (in the absence of commingling) due to the imperfect ability ofthe winery to fully infer the quality contribution of the grower to thewine at grape processing some form of residual claimancy could beexpected to provide appropriate incentives Under these circumstanceswe would anticipate that the use of residual claimancy can in part substi-tute for quality monitoring (see Arrunada et al (2005) for example fromfranchising) But what empirical evidence do we have for the use ofresidual claimancy via bottle-price indexing

Evidence from New Zealand suggests that mixed payment schedules com-posed of a base price and an incentive-related margin are common where thegrower compensation for a particular grape variety is related to the price ofwine produced from that particular grape variety (Boyd et al 2000) Thiscombination of a base price plus commission with bottle-price indexing isreflected in the model developed further below

The use of grape grower contracts conditioning grower compensation onwine retail prices is also documented for Australia France and California(Moulton 1988 Hueth et al 1999 Fraser 2002 Montaigne and Sidlovits2003) Australian evidence suggests that about 20 per cent of grape contractsuse this form of residual claimancy (Fraser 2002) Current figures for bottleprice contracts in California are not available although this figure has beenestimated to have been below 5 per cent during the 1980s (Moulton 1988)Exact figures for France and New Zealand are also not available (Boydet al 2000 Montaigne and Sidlovits 2003) Documented evidence for theuse of residual claimancy in other regions is missing although someauthors propose bottle price-indexing to be part of a solution to overcomefuture coordination issues in developing wine markets in Eastern Europe(Sidlovits and Kator 2007 9)

For those regions where we have documented evidence for residualclaimancy bottle retail prices enter the compensation scheme in differentways In the USA and New Zealand retail bottle prices are used fromwines that originated from the same vineyard or the same grape varietyyet from wines that were released in the previous year (Moulton 1988Boyd et al 2000) In Australia grape growers are compensated basedon retail prices of the forthcoming bottles from the current vintage(Fraser 2005) In France an average retail price is used to derive anindex formula based on forthcoming bottles from the current vintage aswell as from past vintages (Montaigne and Sidlovits 2003)

4 Model

The following model assumes a one-shot game in which a risk-averse grapegrower contracts with a risk-neutral winery over the supply of fresh grapes Inaddition to modelling multiple tasks on the growerrsquos and the wineryrsquos side the

Contracting in the wine supply chain 375

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

model allows for moral hazard on both the growerrsquos and the wineryrsquos partTwo factors are assumed to contribute to the performance indicator accordingto which both winery and grower agree to share the outcome from productionprocessing and marketing the market valuation of the outcome from grapeproduction and wine processing as reflected in the bottle retail price andinformation from effort monitoring

The model shows that a sharing contract can provide incentives to bothprincipal (winery) and agent (grower) such that the efficient contract maxi-mises surplus for all incentive compatible contracts An agency relationshipis considered in which a grape grower allocates his total production effortsamong several activities n = 1 N where the vector of efforts isdenoted by a = (a1 aN) The winery allocates its processing and market-ing efforts amongst activities m = 1 M where the vector of efforts is

given by e = (e1 eM) and a [ RN+ e [ RM

+ respectively For both

grower and winery each element of her effort vector measures managerialeffort in a distinct activity (variable inputs) such that w frac14 [aT eT] Assumethat efforts are observed with noise

w = ae

[ ]+ 1aw

1ew

[ ] such that w = a

e

[ ] (1)

Observational error in measuring quality outcomes is present such that therealisation of

1w = 1aw

1ew

]

[normally distributed with zero mean and covariance matrix Sw is unobservedby both parties The degree of the wineryrsquos inference problem regarding thegrower efforts a is given by the variance of 1aw and the degree of thegrowerrsquos inference problem regarding the winery efforts e is given by the var-iance of 1ew An example for the former case could be the difficulty of thewinery to observe the actual pesticide applications employed by the growerafter signing the contract which may deviate from the contractually specifiedpesticide applications As an example for the latter the grower may observethe marketing campaign of the winery in the marketplace but the actual mar-keting budget that was allocated to specific wines and thus to the correspond-ing grape batches from a given grower may be difficult to observe by thegrower Further a second source of randomness is allowed for It isassumed that both the winery and the grower are exposed to exogenousshocks that make it impossible for both sides to perfectly control their contri-bution to wine and grape quality respectively

1k =1ak

1ek

[ ]

with mean zero and covariance matrixsum

k1N(0sum

k) These shocks

provide scope for moral hazard because although grower and winery

376 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

cannot affect the states-of-nature per se they can affect the outcomerealised in those states The inference problem of the winery withregards to the growerrsquos efforts and the growerrsquos inference problem withregards to the wineryrsquos efforts therefore relates to the potential of mitigat-ing or enhancing the wine quality outcome in an unobserved manner incertain states-of-nature What are examples of such exogenous shocksA shock that can affect the outcome of the wine-quality-contributingefforts of a given grower could be a certain disease in the vineyardsAfter harvest the grower could thus supply grapes of lower qualityblaming the pest However the winery may contract with other growersfrom the same region and thus find out that the grower in questioncould have enhanced certain grape (and thus wine) qualities by additionaleffort following the disease incident Similarly the winery may beexposed to an external shock on its quality-contributing efforts thatcould for example originate from an input supplier or from the retailerend A quality shock due to a defective cork may be an example for theformer Due to extreme weather conditions in a given region (or simplydue to a lack of storage care under regular conditions) the wine retailermay affect the wine quality through its storage quality efforts such thata certain wine quality variation that impedes the wineryrsquos marketingefforts maybe outside of the control of the winery A winery may thereforesuggest vis-a-vis its grower that it suffered a quality shock that was outsideof its control originating from the retail level However if the growerwould supply the same grapes to multiple wineries (or use some othermonitoring device) she may be able to control to what extent thewinersquos final quality (and thus market success) at the retail level is dueto the wineryrsquos processing andor marketing effort as well as due to thewineryrsquos unobserved efforts that may enhance a given quality shockwhich originated from the retailer end

41 Quality outcome from effort allocation

Considering both sources of randomness the wine quality outcome fromeffort allocation becomes

q = Fw +F1w + 1k (2)where F denotes a matrix of productivities It is the objective of bothgrower and winery to specify a joint performance indicator that relies onthis outcome q To achieve this the relationship between efforts andquality outcome could be modelled more explicitly This has two advan-tages First it allows us to transform grape and wine quality attributesinto monetary values via grape grower and winery characteristics (see dis-cussion below) Secondly this enables us to take production processingand marketing realities into account We generally observe that a combi-nation of inputs (production processing marketing) is responsible in

Contracting in the wine supply chain 377

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

determining a given quality attribute For example the residual sugar level(Brix) in grapes is influenced by irrigation weeding and pruning (eg Jones

and Ough 1985) Assuming that Fw = y and y = [yTa yT

e ] we have

Faawa +Faewe

Feawa +Feewe

[ ]= ya

ye

[ ] (3)

In this way Equation (3) tells us how grower and winery efforts translateinto quality attributes of the wine As an example consider Faawa whichtells us how a growerrsquos effort affects quality attributes that he delivers tothe final product Also Faewa reveals that the way in which growerefforts affect the wineryrsquos quality contribution is not necessarily symmetricto the way by which winery efforts affect the growerrsquos quality contri-butions The extent to which wineries can affect the final bottle qualityby making lsquobadrsquo wine out of lsquogoodrsquo grapes may differ from the extent towhich grape growers can affect the processing efforts of the winery andthus final bottle quality Consider that the grower can impact the processingabilities of the winery by affecting the fermentation qualities of the wineThe fermentation process can be impeded by an undesired use of certainpesticides and fertiliser (Wade et al 2004 Downey et al 2006 Lundand Bohlmann 2006) Further the final bottle quality may be impededby grape grower efforts in terms of credence attributes (Darby and Karni1973 Emons 1997) even if the fermentation quality is not affectedgrape supply contracts in Australia for example contain a chemical useclause due to potential chemical residues resulting from chemicalapplications to the grapes (Fraser 2005)

Further to obtain a monetary compensation scheme let vector za denotegrower characteristics that are observable by the winery including charac-teristics such as location of the vineyard grape varieties planted or pro-duction methods used Winery characteristics that can be observed bygrowers are denoted as ze and are assumed to include the type of proces-sing technology employed (eg type of oak barrels) the size of vintage

and the brand name These attributes denoted by zT = zTa zT

e

[ ] are a

reflection of the contract terms to which both parties have committed toThey also form the basis for the monetary valuation of wine attributesby consumers and marketers at the retail level as these attributes are fre-quently visible on the wine bottle or elsewhere in the marketplace Wecould thus allude to hedonic studies which have derived implicit pricesfor wine qualities and labelling attributes (eg Oczkowski 1994Nerlove 1995 Steiner 2004) Perhaps most intriguing in the context ofthe following model is the hedonic study by Golan and Shalit (1993)which derives a producer pricing schedule for grape growers based onthe monetary valuation of wine attributes by consumers and marketers atthe retail level

378 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

42 Performance indicator with monitoring

Not only are winery fieldmen assumed to monitor grape growers but it is alsoassumed that winery monitoring is observed such that growers infer wineryprocessing and marketing efforts from winery reports trade publicationsand other industry participants The resulting lsquomonitoring scorecardsrsquo(grades) are weighted by a contractually determined matrix V which reflectsthe relative importance that a given effort is perceived to have for the finaloutcome from contracting Further it is assumed that matrix Q convertsefforts w into grades such that each effort which has been monitored receivesa single grade (diagonal matrix) or several efforts are used to determine asingle grade Therefore it is assumed that v(s) = V Qw and in order toconvert these grades into monetary terms the assumption is that zT[V Qw]The total performance indicator is thus

m = zT (F(w + 1w) + 1k) + V Qw[ ] (4)To summarise the notation and model structure so far see Table 1

43 Model timeline

The timeline of the model is as following When the contract is signed thewinery puts forward a sharing rule to the grower As part of this sharing ruleboth parties agree ex ante that certain weights will be placed on the informa-tional outcome from monitoring activities The contract also specifies ex antethat grape growers are compensated based on retail prices of the forthcoming

Table 1 Notation and structure of performance measure

(I) Output information

w Effort (variable inputs)

w = w + 1w Observable effort (with observational noise) E(1w) frac14 0

q = Fw + 1k Production function (with external shocks) E(1k) frac14 0

F Productivities

z Contract provisions (eg grape variety type of oak barrel)

(II) Supervision

s 5 Qw Information outcome of monitoring activity

V Weights on monitored variables

zT[V Qw] Conversion of monitoring grades into dollars

v(s) = Vs rArr performance indicator m = zT[(F(w + 1w) + 1k) + VQw]w (N times 1) 1w (N times 1) 1k (M times 1)q (M times 1) rArr F (M times N)n (M times 1) rArr V (M times S)s (S times 1) rArr Q (S times N)

Contracting in the wine supply chain 379

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

bottles from the current vintage After the contract is signed and given the infor-mation from monitoring we obtain v(s) After both grower and winery havecommitted their efforts and wine is produced we observe to which contractterms z they have adhered to This is a reflection of their production functionthat transformed efforts with observation noise 1w and in the presence ofrandom shocks 1k Assuming that the contract provisions z have market valueand are ultimately responsible for the price of the bottle of wine revenue is gen-erated that can be shared through a performance indicator as in Equation (4)

The underlying cost-of-effort function (lsquodisutilityrsquo) is assumed convex andmonotonically increasing since the cost-of-effort matrices K1 and K2 areassumed symmetric positive semidefinite and considered in monetary terms

Assumption 1

Cprimew(w) 0 Cprimeprime

ww(w) ge 0 forallw [ RN where

Cprimew (w) = partC(w)

partwand Cprimeprime

ww (w) = part2C(w)partwpartwT

Hence C1(a) = 12aTK1a defines the growerrsquos quadratic cost of effort and

C2(e) = 12eTK2e denotes the wineryrsquos quadratic cost of effort (seeHolmstrom and Milgrom (1991 1994) and Slade (1996) for quadratic cost

functions in the context of multitasking) Therefore C(w) = 12wTK3wwhere

K3 = K1 Kae

KaeT K2

[ ] (5)

Since the cost-of-effort matrices K1 and K2 are assumed symmetric positivesemidefinite efforts can be represented as substitutes in the growerrsquos andwineryrsquos cost-of-effort function respectively In those cases it is assumedthat if the incentive intensity is increased on one effort this will causesubstitution away from other types of efforts

Given the outcome from monitoring and the outcome from grape and wineproduction as in Equation (4) it is assumed that the optimal second-best

incentive scheme l takes the following linear form

l = a (zT[(F(w + 1w) + 1k) + V Qw]) + b (6)where a denotes the commission rate on the dollar outcome (there exists nosharing contract that implements first-best under double moral hazard evenwith risk-neutral agents the contract will be second-best as long as weassume that the budget-balancing constraint is satisfied see Holmstrom(1982) and Bhattacharyya and Lafontaine (1995) for proof)

380 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

The size of a reflects thus how strongly powered the incentives are for thegrower such that in a high-powered incentive contract the agentrsquos total returnswill be relatively sensitive to the contracting outcome If a frac14 0 the growerceases to be an independent supplier whereas with a frac14 1 the growerwould become full residual claimant The scalar b denotes a fixed ex antebase payment upon which grower and winery agree when signing the contractWe could also let b take negative values which would allow for thepossibility that the grower borrows capital from the winery For example inAustralia we observe that wineries desire specific grape varieties fromgrape growers in the production of certain wines (Scales et al 1995) apractice that could be implemented by wineries lending capital to growersIn New Zealand we have evidence that wineries provide capital to growersfor converting vineyards (Boyd et al 2000)

The sharing rule could be modified such that another parameter is deter-mined ex ante to the allocation of grower and winery effort Instead ofleaving the relative weights allocated to production processing and marketingoutcome versus monitoring outcome in the performance indicator unspecifiedboth parties could agree ex ante on a base split that is variable the qualityoutcome could for example be linked with the performance of the consumerprice index (CPI) resulting in a flexible sharing rule over l

E[mlowast] = zT[lF+ (1 minus l)V Q]w = zTMlowastw (7)Indeed linking wine bottle contracts to the CPI is practised in the Australianwine industry (Scales et al 1995 Fraser 2002) However since this modifi-cation does not change the key results in which we are interested we willproceed as in Equation (6)

Given Equation (6) the wineryrsquos problem is to allocate the surplus such thatexpected profits are maximised subject to the constraints that the winery andthe grower comply with the efforts specified in the contract and subject to thecondition that the reservation utility for the grower is assured While theresulting contract is assumed to be efficient it leaves the surplus allocationunspecified since retail prices are unobserved ex ante the winery onlychooses a and agrees with the grower on the weights that shall be placedon the monitoring outcome

44 Setting up incentive compatibility conditions

Assuming a risk-averse grower we are interested in the variance of thepayment scheme (6) as this serves to derive the risk premium Furtherthrough the covariance matrix we can analyse random complementaritiesbetween tasks Random complementarities arise when the random allocationof efforts to one quality task increases the marginal expected benefit of allo-cating efforts to another task for example as a result of a random positivedemand shock (eg Slade 1996) Consider that in a stochastic environmentthe covariation of uncertainty across contract provisions implies that the

Contracting in the wine supply chain 381

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

grower is exposed to greater compensation risk Given the greater sensitivityof the grower to a given incentive (to reallocate efforts) in this environmentwe would expect that incentive contracts are used which are of lower power(eg Holmstrom and Milgrom 1991) Therefore the optimal share parametershould be decreasing in random complementarities where risks are correlatedacross tasks

From Equation (6) we obtain

Var(m(a e z)) = E(zT(F1w + 1k)(F1w + 1k)Tz)

= zT Fsumww

FT +Fsumwz

+sumzw

FT +sum

zz

( )z

sumij

= 1i 1Tj

( )

= zsum

z

(8)where

sum=

sumaa

sumaesumT

ae

sumee

[ ]Given the above sharing rule as in Equation (6) the grower profits aregiven by

Pa = amminus 1

2aTK1a + b (9)

From the moment generating function for the multivariate normal 1 N(0S) we obtain the CARA expected utility of profits as

E[u(Pa)] = minus exp minusr am+ bminus 1

2aTK1a

( )+ 1

2r2a2zTSz

(10)

where r denotes the constant absolute risk aversion coefficientr = minusUprimeprimeUprime r 0

Noting that

E[m] = zT(F+ V Q)w

zTMw = zTa Maaa + zT

a Maee + zTe Meaa + zT

e Meee(11)

It follows that the certainty equivalent (CE) for the grower is

CE = a(zTMw) + bminus 1

2aTK1a minus 1

2ra2zTSz (12)

The CE utility is thus given by the expected compensation minus the private costof efforts minus the risk premium As long as the CE utility satisfies the growerrsquosreservation level he will accept the contract However since it is assumed thatcontracting takes place in an environment of randomness in ex post observedretail prices the associated information asymmetries require that incentive

382 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

compatibility constraints are met In setting up the incentive compatibility con-ditions it is assumed that the grower chooses his own efforts a such that thewineryrsquos expected profits are maximised This optimisation problem of thegrower excludes the effort choice of the winery e but includes the variancendashcovariances since we assume a CARA model of grower choice

a [ arg max a (zTMw) + bminus 1

2aTK1a minus 1

2ra2zTSz

(13)

The necessary first-order conditions for Equation (13) are thus

a zTa Maa + zT

e Mea

( )minus aTK1 le 0 (14a)

a ge 0 (14b)

a a zTa Maa + zT

e Mea

( )minus aTK1

[ ]= 0 (14c)

Assuming that an interior solution exists we can solve Equation (14a) as asystem of equalities for a such that

a(e) = aKminus11 (MT

aaza + MTeaze) (15)

Since Equation (15) reveals the level of grower effort that maximises thegrowerrsquos certain equivalent income in Equation (12) it gives us the incentivecompatibility condition that needs to be satisfied to achieve a feasible contract

45 Setting up the wineryrsquos unconstrained maximisation problem

In the following step we substitute the growerrsquos effort choice function (15)into the growerrsquos CE income function (12) to obtain the indirect certaintyutility of the grower

Denoting

E[m] = da[m(a)] + de[m(e)]

we can write the indirect CE as

CE(a a) = b+ ada[m(a)] + ade[m(e)] minus1

2aTK1a minus 1

2r Var(ama) (16)

Together with the participation constraint

b+ ada[m(a)] + ade[m(e)] minus1

2aTK1a minus 1

2r Var(ama) ge U(n) (17)

in which U(n) is the default utility level of the grower and (n) is its certainmonetary equivalent As long as the CE utility is greater than the defaultutility of the grower the growerrsquos performance incentives are not affectedand b serves only as a redistribution (surplus transfer) between winery and

Contracting in the wine supply chain 383

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

grower without affecting the agentrsquos performance incentives (Holmstrom andMilgrom 1994) From Equation (17) an inequality constraint is implied onthe wineryrsquos choice for the fixed base payment b

b ge nminus ada[m(a)] minus ade[m(e)] +1

2aTK1a + 1

2r Var(ama) (18)

Thus to induce the growerrsquos voluntary participation we impose a partici-pation constraint which together with the incentive constraint is necessaryto deliver an incentive feasible contract

For convenience let

K1 = Kminus11 0

0 0

[ ] K2 = 0 0

0 Kminus12

[ ]

such that K1 + K2 = Kminus13

Considering the growerrsquos optimal effort level from Equation (15) we obtain

E(m(a)) = a zTa MaaKminus1

1 MTaaza + zT

a MaaKminus11 MT

aaze

((19a)

+ zTe MaaKminus1

1 MTaaza + zT

e MeaKminus11 MT

eaze

)(19b)

= azTMK1MTz (19c)

Given the wineryrsquos profit function as

P = (1 minus a)mminus 1

2eTK2e minus b (20)

the expected profit criterion becomes

E(Pe) = (1 minus a)E[m] minus n+ ada[m(a)] + ade[m(e)] minus1

2aTK1a

minus 1

2r Var(ama) (21)

From Equation (21) we know that the participation constraint is binding sinceE(Pe) is strictly decreasing in b Together with a strictly positive reservationutility of the grower n we know that grapes and wine are produced

In order to obtain the wineryrsquos unconstrained maximisation problem maxae

E(Pe) we substitute the right-hand side of Equation (15) for a in Equation (21)and the right-hand side of Equation (18) as equality for b into Equation (21)The optimal level of winery efforts is thus given by

partEPe

parte= (1 minus a+ ade)(zT

a Mae + zTe Mee) minus eTK2 (22)

384 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

This yields

e(a) = (1 minus a+ ade)Kminus12 (MT

aeza + MTeeze) (23)

Therefore Equation (19c) becomes

E[m(e)] = 1 minus a+ ade( )zTMK2MTz (24)

which yields unconstrained profits of the winery

maxae

E(P_

e) = zT

1 minus a+ ada minus1

2a

( )aK1

+ 1 minus a+ ade minus1

2(1 minus a+ ade)

( )((1 minus a+ ade)K2

)M

T minus 1

2ra2S

⎡⎢⎢⎢⎣⎤⎥⎥⎥⎦z

(25)

= zT M minusaminus 3

2+a2da

( )K1 +

1

2

(1+a2 +a2d2

e minus 2aminus 2a2de + 2ade

)K2

MT minus 1

2ra2S

[ ]z

(26)

To find the optimal level of winery efforts we use the following equation

partE(P_

e)parta

= zT M(1minus 3a+ 2ada)K1+

(a+ad2e + 1minus 2ade + de)K2

MT minus raS

[ ]z = 0 (27)

Therefore

alowast = 1

1+ r zTSz zTMKminus1

1 MTz( )( ) (28)

Several comparative static results can be obtained by considering the lsquooptimalshare parameterrsquo a and by varying the assumptions with regard to the degreeof effort contractibility the growerrsquos risk aversion and the growerrsquos disutilityof effort

Proposition 1 As long as the grower is risk-averse and observational error inmeasuring quality and randomness on the supply and demand side is presentthe optimal contract involves strict sharing

Proof

if 1w 1k r 0 rArr a 0 a 1

QED

Contracting in the wine supply chain 385

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Proposition 2 Increasing risk aversion of the grower is associated with ahigher Pareto optimal share that goes to the winery (and vice versa)

Proof

r 1 alowast 0

r 0 alowast 1

QED

In a single-moral hazard setting and with both risk-neutral principal andagents we would not expect strict sharing (ie the optimal contract cannothave a frac14 0 or a frac14 1) since the agent who provides the only unobservableinput would be the single residual claimant However Bhattacharyya andLafontaine (1995) show that this result changes in the context of double-moralhazard to a strict sharing arrangement Therefore Proposition 1 is consistentwith previous model evidence of strict sharing in the presence of double-sidedmoral hazard when both parties are risk-neutral and only a single task isperformed (Bhattacharyya and Lafontaine 1995 Corollary 1) Since themodel in this paper accounts for a risk-averse agent in the context of multi-tasking Proposition 1 suggests that under the above model conditionsstrict sharing and thus partial residual claimancy can extend to a double-moral hazard setting when the agent is risk-averse and performs multipletasks Therefore the prediction of strict sharing also distinguishes the abovemodel from the Holmstrom and Milgrom (1991) model

To motivate Proposition 1 further consider that strict sharing is not onlyconsistent with previous incentive theories of franchising (eg Rubin 1978Mathewson and Winter 1985) but also with empirical evidence fromfranchising (Kaufmann and Lafontaine 1994) In those franchise contractsex post rents can serve as a coordination device because the ex post rentsthat are left downstream by the franchisor create franchisee incentives(Kaufmann and Lafontaine 1994) This rationale that ex post rents in astrict sharing contract serve to motivate agents lends support to the use ofstrict sharing as a meaningful coordination device in the context of bottleprice contracts as they are observed in Australia New Zealand Californiaand France

The prediction from Proposition 2 regarding the relationship between thegrowerrsquos risk aversion and the allocation of the Pareto optimal share to thegrower is comparable to the prediction from standard linear compensationcontracts in the absence of multi-tasking and bilateral moral hazard theagentrsquos commission rate is predicted to be a decreasing function of risk aversionand goes to zero when risk aversion goes to infinity (Holmstrom and Milgrom1987) What is novel compared with the standard linear contract with one taskand compared with predictions from Holmstrom and Milgromrsquos (1991)risk-neutral multitask model is the prediction of what this relationship

386 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

between risk aversion and partial residual claimancy means for cooperation in abottle price contract ie to what extent the model can provide us with furtherinsights into the extent of higher levels of cooperation between winery andgrape grower Consider that Holmstrom and Milgrom (1991) put forward theiranalysis of multi-tasking in a one-sided moral hazard setting as a lsquorudimentarytheory of ownershiprsquo (p 26) which takes the variance of measurement errorand thus measurement costs as an important determinant of integration Thusas long as observational errors in measuring quality are present in the abovemulti-task setting (Proposition 1) we would not only anticipate higher levelsof cooperation compared with the absence of multi-tasking and measurementerrors but in the context of a risk-averse grower (Proposition 1) expect thatwith increasing growerrsquos risk aversion the increasing efficiency of the wineryof bearing risks is likely associated with higher levels of cooperation (towardsownership and thus integration) such that closer coordination is associatedwith a lower Pareto optimal share that goes to the grower

Considering empirical evidence from California we observe that largergrape growers ndash which are likely to be less risk-averse compared withsmaller growers ndash are more likely to supply lower quality grapes to wineriesthan smaller producers (Goodhue et al 2004) Compared with smallergrowers these grape growers are observed to contract with wineries overprice incentives rather than production practices evidence from Californiasuggests that winegrape acres have a significant positive effect on theinclusion of price incentives in the grape contracts (Goodhue et al 2004)Since the use of production practices ndash which effectively amounts to ahigher degree of coordination between grower and winery (compared withcontracting over price incentives) ndash is more frequently associated withgrowers supplying higher quality grapes Propositions 1 and 2 seem to fitthe observation that wineries are more likely to coordinate closely (towardsintegrating grape growing into the winery) with smaller and likely morerisk-averse growers (while possibly receiving a greater Pareto share althoughwe have no published evidence from industry or other studies of a higherPareto share going to the winery)

Proposition 3 As the magnitude of the growerrsquos disutility of effort increasesthe Pareto optimal share which goes to the grower decreases

Proof

as K1 1 alowast 0

QED

Proposition 4 With increasing observational error in measuring quality thePareto optimal share which goes to the grower decreases

Contracting in the wine supply chain 387

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Proof

as 1w 1 alowast 0 (S 1 in Equation (28))

QED

From Propositions 3 and 4 we anticipate that as a growerrsquos disutility ofeffort increases (for example by expanding into the production of higherquality grapes) and as the role of quality measurement error and thus the dif-ficulty (cost) of monitoring performance increases a greater Pareto share goesto the winery The prediction that more extensive grower monitoring costs areassociated with a reduced partial residual claim going to the grower ie thatpartial residual claimancy substitutes for monitoring in the above context ofmulti-tasking and bilateral moral hazard has interesting parallels to relatedcontracting studies We have empirical evidence from share contracts in fran-chising that residual claimancy can save on monitoring costs (Arrunada et al2005) Similarly in a model of one principal contracting with two agentswhere each agent performs an outside and an inside activity residual clai-mancy of agricultural producers (in the sense of contracts being linearly con-ditioned on each growerrsquos revenue) is predicted to substitute for performancemeasurement (Hueth and Melkonyan 2004) Furthermore Hueth et al (1999)predict that a substitution relationship between residual claimancy and moni-toring could be expected to hold stronger in some agricultural organisationalforms rather than in others1

But what does evidence from the wine industry suggest with respect to theabove Evidence from Australia and California suggests that the use of pro-duction practice provisions (indirect monitoring) and a high direct monitoringintensity is predominant in high-quality grape regions (Scales et al 1995Goodhue et al 2004 Fraser 2005) The more extensive use of productionpractice provisions in grape supply contracts of high-quality regions islikely a reflection of the difficulty (costs) of identifying and measuring thekey grape characteristics that determine grape and thus wine quality inthose regions (Scales et al 1995) This suggests that the winery uses theseprovisions as an indirect monitoring mechanism in an attempt to addressthe incentive problems created by multi-tasking when grower effortsbetween quality tasks differ in measurability at harvest time

Further in those cases where wineries contract with high-quality grapegrowers we anticipate that these wineries which likely face greater qualitymeasurement costs (multitasking problems) and thus higher processingquality risks would receive a higher Pareto share under bottle price contracts

1 lsquo a processor might let growers earn some portion of their payment through revenue shar-

ing with the processing firm Presumably growers would feel greater connection with the organ-

ization and would be more inclined to produce high-quality output It is noteworthy that

agricultural marketing cooperatives ndash organizations that do use this form of residual claimancy

ndash have a notoriously difficult time monitoring the quality of their membersrsquo productionrsquo (Hueth

et al 1999 384)

388 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

This increasing Pareto share could thus be expected to be associated with anincreasing internalisation of contracting externalities associated with multi-tasking Considering the above model framework (Propositions 2ndash4) wewould expect that wineries aim to internalise those externalities of incentivedesign by closer cooperation (towards integration) between grape productionand their own processing operation Such evidence of wineries integratinghigh-quality grape growing into their own operation comes from Australia(Scales et al 1995) and Spain (Olmos 2008)

We also observe that production practices in high-quality regions encom-pass the use of winery and wine-specific grape varieties (Scales et al1995) such that wineries supply growers with new vines on the conditionthat they have a right to buy the vintages from those vines for a certainnumber of years (Boyd et al 2000) Thus we have further evidence thatcloser coordination between wineries and growers occurs in cases wherequality measurement cost and growersrsquo disutility of efforts are high andmulti-tasking problems are likely to be more significant

Proposition 5 With increasing uncertainty of measuring effort contributionit becomes more efficient to allocate a greater Pareto optimal share to the lessrisk-averse party

Proof

When 1w 1k≫ 0 r 1 alowast 0

QED

Proposition 5 is supported by empirical evidence in the sense that closercoordination (towards integration of grape growing into the processing oper-ation) is observed in high-quality regions where bottle-price indexing isobserved (Scales et al 1995 Boyd et al 2000) and where the uncertaintyof measuring bilateral efforts is likely higher compared with low-qualityregions These observations support Proposition 5 since in those wine-producing regions in which (i) effort allocation inference problems and (ii)quality shocks due to states of nature and related moral hazard issues aremost likely prevalent (ie high-quality wine regions where both the likelihoodand the magnitude of states of nature and their quality implications is expectedto be most significant) it is likely more efficient to allocate the winery agreater Pareto share through closer coordination (towards integration) Fur-thermore in those instances ((i) and (ii)) it is also likely more efficient to allo-cate the winery as the less risk-averse party a larger Pareto share since it likelycan bear these risks with regard to the measurability of effort contributionmore efficiently than the grower

Contracting in the wine supply chain 389

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

5 Conclusions

This paper explores internal organisational design problems of firms in the wineindustry as they relate to moral hazard multi-tasking and residual claimancyThe paper uses a quasi-case-study approach to analyse contract specificationsand their relationships with performance measurement and control in thewine industry of certain regions Stylised facts are employed from previousindustry studies to explore the applicability of the theory to the wine industryin regions where we have documented evidence for residual claimancy insupply contracts

To explore predictions from a particular agency relationship between agrape grower and a winery over the supply of grapes for wine production amulti-tasking model is first developed which encompasses residual claimancythrough bottle-price indexing (revenue sharing) and double-sided moralhazard while allowing for asymmetric quality contributions of the contractingparties

The modelrsquos comparative static results are employed to provide insights intowhat factors determine the Pareto optimal shares of the contracting parties Themodel predicts that as long as there are observation errors in measuring effortsand as long as the grape grower is risk averse the Pareto optimal rate lies strictlybetween 0 and 1 hence partial residual claimancy is supported Thus strictsharing (partial residual claimancy) finds support in a bilateral moral hazardsetting when the agent is risk-averse and faces multiple tasks Further themodel predicts that with increasing magnitude of the growerrsquos disutility ofeffort and with increasing difficulty of monitoring performance a greaterPareto share goes to the winery under partial residual claimancy At the sametime the model predicts that when partial residual claimancy is used closercoordination (towards integration of grape growing into the processingoperation) is to be observed in high-quality regions where the uncertainty ofmeasuring bilateral efforts is likely higher compared with low-quality regions

Although no formal tests are presented several model predictions fit withempirical evidence in those wine industries where we have documented evi-dence that residual claimancy is used namely in Australia France NewZealand and California Nevertheless some model predictions may also beapplicable to other markets than to the above wine markets for tworeasons First when considering Hueth et alrsquos (1999) argument that the pro-prietary nature of contracts implies limited empirical research (and thus evi-dence) on contracts the actual use of residual claimancy in global winemarkets may be more widespread than the published academic literaturesuggests Secondly we have evidence for multi-tasking and the use of residualclaimancy in the sugar beet sector as well as in the market for fresh tomatoes(Hueth and Ligon 1999 Hueth and Melkonyan 2002) and we have evidencethat residual claimancy is used in the asparagus the broccoli the carrot theonion the orange the potato the squash and the peaches sector of California(Hueth et al 1999 Wolf et al 2001 Hueth and Melkonyan 2004)

390 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

The model has made a number of convenient assumptions and has therebyleft aside important issues that warrant further analysis Given the static natureof the model it ignores ex post contract renegotiation and an analysis of itsconsequences (eg Tirole 1999) The model implies equality in bargainingpower between grape grower and winery in implementing contracts Althoughthe paper has presented empirical evidence which suggests that the balance ofbargaining power is not consistently skewed towards the winery ongoingindustry consolidation at the processor level in increasingly global winemarkets as well as grape supplyndashdemand disequilibria (eg Fraser 2005)could change this balance of bargaining power

Acknowledgements

I thank Jeff LaFrance Ben Hermalin Urvashi Narain Carlo Cafiero Maximilian Auffhammer

Ethan Ligon Rob Weaver Aaron Green Peterangelo Vallis Kirby Moultondagger Dale Heiendagger Jim

Lapsley Andy Beckstoffer James Omond and Karen Ross for very helpful discussions and sug-

gestions I thank the anonymous referees the editor and seminar participants at the University of

Texas AampM and the University of Alberta for helpful suggestions on earlier versions of this

paper Thanks especially to Ethan Ligon and the Department of Agricultural amp Resource Econ-

omics University of California at Berkeley for invaluable support and hospitality during a visit

which only enabled me to initialise this paper I would also like to thank my colleagues at the

University of Alberta Department of Rural Economy for their endured support during my

time in Canada while finalising this paper Funding provided by the Volkswagen Foundation

and the Social Sciences and Humanities Research Council of Canada is gratefully

acknowledged

References

Ackerberg D and Botticini M (2002) Endogenous matching and the empirical determi-

nants of contract form Journal of Political Economy 110 564ndash591

Alchian A and Demsetz H (1972) Production information costs and economic organ-

ization American Economic Review 62(5) 777ndash795

Arrunada B Garicano L and Vazquez L (2005) Completing contracts ex post How car

manufacturers manage car dealers Review of Law and Economics 1(1) 150ndash173

Ashenfelter O Ashmore D and Lalonde R (1995) Bordeaux wine vintage quality and

the weather Chance 8 7ndash14

AWBC (2007) Australian wine sector at a glance 2006 httpswwwawbccomau

winefactsdatafreeaspsubcatid93 Australian Wine and Brandy Corporation

Accessed 26 March 2007

Bedwell B (2000) The state of the grapes ndash a world view In Paper presented to the

Washington State Grape Society Annual Meeting Grandview Washington 16ndash17

November 2000 httpwwwgrapesocietyorg2000meetingproceedingsstateofgrapes

html Accessed 21 February 2008

Benavente J (2004) Technological change in Chilersquos grape and wine producing industry

World Bank Technical Report WB0131 Chile Department of Economics The Univer-

sity of Chile Santiago

Contracting in the wine supply chain 391

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Bhattacharyya S and Lafontaine F (1995) Double-sided moral hazard and the nature of

share contracts Rand Journal of Economics 26 761ndash781

Boyd H Evans L and Quigley N (2000) The Efficiency of Contractual Arrangements in

Private Agricultural Product Markets Victoria University of Wellington Law Review

31 Wellington New Zealand Institute for the Study of Competition and Regulation

(ISCR)

Brickley J (2002) Royalty rates and upfront fees in share contracts Evidence from

franchising Journal of Law Economics and Organization 18 511ndash535

Byron R and Ashenfelter O (1995) Predicting the quality of an unborn Grange

Economic Record 71 400ndash414

Corbett C DeCroix G and Ha A (2005) Optimal shared savings contracts in supply

chains Linear contracts and double moral hazard European Journal of Operational

Research 163 653ndash667

Dana J and Spier K (2001) Revenue sharing and vertical control in the video rental

industry Journal of Industrial Economics 49 223ndash245

Darby M R and Karni E (1973) Free competition and the optimal amount of fraud

The Journal of Law and Economics 16 67ndash88

Davis T and Ahmadi-Esfahani F (2005) Hedonic modeling for Australian wine 49th

Annual conference of the Australian Agricultural and Resource Economics Society

Novotel Australia

Downey M Dokoozlian N and Krstic M (2006) Cultural practice and environmental

impacts on the flavonoid composition of grapes and wine A review of recent research

American Journal of Enology and Viticulture 57 257ndash268

Emons W (1997) Credence goods and fraudulent experts Rand Journal of Economics 21

107ndash119

Eswaran M and Kotwal A (1985) A theory of contractual structure in agriculture

American Economic Review 75 352ndash367

Fairweather J Campbell H and Manhire J (1999) The lsquogreeningrsquo of the New Zealand

wine industry Research Report No 7 Studies in Rural Sustainability Dunedin

New Zealand Department of Anthropology University of Otago

Fares M Ayouz M and Martin G (2002) Contract choice quality and industrialization

in the Argentine winegrape industry Test of transaction cost theory In Paper pre-

sented at the International Society for New Institutional Economics Annual Meeting

Cambridge MA 27ndash29 September 2002

Fraser I (2002) Relations between wine grape growers and wineries in Australia Survey

results Australian Viticulture 6 68ndash79

Fraser I (2003) The role of contracts in wine grape supply coordination An overview

Agribusiness Review 11 1ndash16

Fraser I (2005) Microeconometric analysis of wine grape supply contracts in Australia

Australian Journal of Agricultural and Resource Economics 49 23ndash46

Golan A and Shalit H (1993) Wine quality differentials in hedonic grape pricing

Journal of Agricultural Economics 44 311ndash321

Goodhue R Heien D and Lee H (1999) Contract use in the California Winegrape

Economy Davis AIC Issues Brief No11 Agricultural Issues Center Davis University

of California

392 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Goodhue R Heien D Lee H and Sumner D (2002) Contract use widespread in wine

grape industry California Agriculture 56(3) 97ndash102

Goodhue R Heien D Lee H and Sumner D (2004) Contracts and quality in the

California winegrape industry Review of Industrial Organization 23 267ndash282

Goodhue R Green R Heien D and Martin P (2008) California wine industry evolving

to compete in 21st century California Agriculture 62(1) 12ndash18

Grossman S and Hart O (1986) The costs and benefits of ownership A theory of vertical

and lateral integration Journal of Political Economy 94 691ndash719

Gwynne R (2006) Governance and the wine commodity chain Upstream and down-

stream strategies in New Zealand and Chilean wine firms Asia Pacific Viewpoint 47

381ndash395

Hart O and Moore J (1990) Property rights and the nature of the firm Journal of

Political Economy 98 1119ndash1158

Hefetz A and White G (1999) Grape purchasing and disease management strategies for

premium wine grapes Staff paper SP 99ndash02 Department of Agricultural Resource and

Managerial Economics Cornell University Ithaca New York

Hennessy D and Lawrence J (1999) Contractual relations control and quality in the hog

sector Review of Agricultural Economics 21(1) 52ndash67

Holmstrom B (1979) Moral hazard and observability Bell Journal of Economics 10

74ndash91

Holmstrom B (1982) Moral hazard in teams Bell Journal of Economics 13 324ndash340

Holmstrom B and Milgrom P (1987) Aggregation and linearity in the provision of

intertemporal incentives Econometrica 55 303ndash328

Holmstrom B and Milgrom P (1991) Multitask principal-agent analyses Incentive con-

tracts asset ownership and job design Journal of Law Economics and Organization 7

24ndash52

Holmstrom B and Milgrom P (1994) The firm as an incentive system American

Economic Review 84 972ndash991

Hueth B and Ligon E (1999) Producer price risk and quality measurement American

Journal of Agricultural Economics 81(3) 512ndash524

Hueth B and Ligon E (2002) Estimation of an efficient tomato contract European

Review of Agricultural Economics 29(2) 237ndash253

Hueth B and Melkonyan T (2002) Quality measurement and contract design Lessons

from the North American sugarbeet industry Food System Research Group Working

Paper FSWP2002ndash6 Madison University of Wisconsin 19

Hueth B and Melkonyan T (2004) Identity preservation multitasking and agricultural

contract design American Journal of Agricultural Economics 86(3) 842ndash847

Hueth B Ligon E Wolf S and Wu S (1999) Incentive instruments in fruit and

vegetable contracts Input control monitoring measuring and price risk Review of

Agricultural Economics 21(2) 374ndash389

Jones R and Ough C (1985) Variations in the percent ethanol (vv) per Brix conversions

of wines from different climatic regions American Journal of Enology and Viticulture

36(4) 268ndash270

Contracting in the wine supply chain 393

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Kaufmann P and Lafontaine F (1994) Costs of control The source of economic rents for

McDonaldrsquos franchisees Journal of Law and Economics 37(2) 417ndash453

Laffont J-J and Matoussi MS (1995) Moral hazard financial constraints and sharecrop-

ping in El Oulja Review of Economic Studies 62 381ndash399

Lal R (1990) Improving channel coordination through franchising Marketing Science 9

299ndash318

Lund S and Bohlmann J (2006) The molecular basis for wine grape quality A volatile

subject Science 311 804ndash805

Luporini A and Parigi B (1996) Multi-task sharecropping contracts the Italian mezza-

dria Economica 63 445ndash457

Mathewson G and Winter R (1985) The economics of franchise contracting

The Journal of Law and Economics 28 506ndash526

Milgrom P and Roberts J (1995) Complementarities and fit strategy structure and

organizational change in manufacturing Journal of Accounting and Economics 19

179ndash208

Montaigne E and Sidlovits D (2003) Long Term Contracts and Quality in the Wine

Supply Chain The Case of the Appellation lsquoVins des Sables du Golfe du Lionrsquo

Budapest Hungary International Society for New Institutional Economics Budapest

University Department of Economics

Moulton K (1988) Grape pricing in California and its relation to product quality Paper

presented at the General Assembly of the OIV 5ndash9 September 1988 International

Organisation of Vine and Wine Paris France

Nerlove M (1995) Hedonic price functions and the measurement of preferences The case

of Swedish wine consumers European Economic Review 39 1697ndash1716

Oczkowski E (1994) A hedonic price function for Australian premium table wine

Australian Journal of Agricultural Economics 38 93ndash110

Olmos M (2008) Why use contracts in viticulture Journal of Wine Research 19 81ndash93

Omond J (2003) Grape Supply Contracts Seminar paper by LAAMS Group Pty Ltd

Melbourne Vic Omond amp Co wine law specialists

RMN (2007) Wine 2006 Crush 14 Davis Rural Migration News Department of Agri-

cultural Economics University of California httpmigrationucdavisedurmnmore

phpid=1211050 Accessed 26 July 2007

Rubin P (1978) The theory of the firm and the structure of the franchise contract

The Journal of Law and Economics 21 223ndash233

Scales W Croser B and Freebairn J (1995) Winegrape and wine industry in Australia

Report by the committee of inquiry into the winegrape and wine industry Canberra

Australia Australian Government Publishing Service

Shepherd A and OrsquoDonell V (2001) Australian Wine Grape Industry Australian Bureau

of Agricultural and Resource Economics httpabareonlineshopcomproduct

aspprodid=11711 Accessed 24 March 2007

Sidlovits D and Kator Z (2007) Characteristics of vertical coordination in the Hungarian

wine sector In Paper prepared for presentation at the joint IAAE ndash 104th EAAE

Seminar Agricultural Economics and Transition lsquoWhat was expected what we

observed the lessons learnedrsquo Budapest Hungary 6ndash8 September 2007

394 B Steiner

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from

Slade M (1996) Multitask agency and contract choice An empirical exploration

International Economic Review 37 465ndash486

Smith K and Marsh I (2007) Wine and economic development technological and

corporate change in the Australian wine industry International Journal of Technology

and Globalisation 3(2ndash3) 224ndash225

Steiner B (2004) Valuing labelling attributes with hedonic price analysis Australian

wines in the British wine market Agribusiness ndash An International Journal 20

287ndash307

Stiglitz J (1974) Incentives and risk sharing in sharecropping Review of Economic

Studies 42 219ndash255

Tirole J (1999) Incomplete contracts Where do we stand Econometrica 67 741ndash781

Wade J Holzapfel B Degaris K Williams D and Keller M (2004) Nitrogen and

water management strategies for wine-grape quality Acta Horticulturae 640 61ndash67

Wine Institute (2007a) Bonded Wineries San Francisco CA Wine Institute httpwww

wineinstituteorgindustrykeyfactscawineriesphp Accessed 23 March 2007

Wine Institute (2007b) California Wine A Signature California Industry San Francisco

CA Wine Institute httpwwwwineinstituteorgcaliforniawineimpactpdf Accessed

22 March 2007

Wine Institute (2011) Number of California Wineries San Francisco CA Wine Institute

httpwwwwineinstituteorgresourcesstatisticsarticle124 Accessed 27 February

2011

Wolf S Hueth B and Ligon E (2001) Policing mechanisms in agricultural contracts

Rural Sociology 66(3) 359ndash381

Zylbersztajn D and Miele M (2005) Stability of contracts in the Brazilian wine industry

Revista de Economia e Sociologia Rural 43 353ndash371

Contracting in the wine supply chain 395

at New

University of Southern D

enmark on July 3 2012

httperaeoxfordjournalsorgD

ownloaded from