Buyer-Induced Exclusive Dealing 1

Size: px
Start display at page:

Download "Buyer-Induced Exclusive Dealing 1"

Transcription

1 Buyer-Induced Exclusive Deling Dvid E. Mills Deprtment of Economics P.O. Box 4008 University of Virgini Chrlottesville, VA (phone) (fx) (Preliminry Drft: Februry, 04 ) The uthor thnks the Bnkrd Fund for Politicl Economy for providing finncil support for this reserch.

2 I. Introduction Exclusive deling rrngements require buyer to purchse good exclusively or chiefly from single supplier. The buyer my be retiler or distributor who resells the good, or my be n end user. In most instnces these rrngements - whether requirement or n greement - re imposed on the buyer by the seller. The seller my seek n exclusive deling rrngement for severl procompetitive resons such s preventing competing suppliers from free riding on the seller s investments in distributor s sles effectiveness (Mrvel, 98). But n exclusive deling rrngement my hve n dverse effect on competition if it forecloses the supplier s competitors from sufficiently lrge portion of the mrket for sufficient period of time. Exclusive deling my exclude even n entrnt who is more efficient thn the incumbent seller who imposes exclusive deling on its buyers. When there re economies of scle in production nd mny buyers hve exclusive dels with the incumbent, n entrnt cnnot grner enough business to rech the minimum efficient scle of opertions. Although the extensive literture on exclusive deling concerns minly those rrngements imposed by sellers, some exclusive deling rrngements re instigted by buyers rther thn sellers. These rrngements rise in mrkets for homogeneous goods where sellers only compete in prices, but they lso rise in mrkets for differentited products nd brnded consumer goods. For instnce, chin of convenience stores my sell single brnd of light bulbs or sun glsses, or chin of fst-food resturnts my sell single compny s fountin There is substntil literture on the uses of exclusive deling to deter entry, beginning with Aghion nd Bolton (987) nd followed by Rsmusen, Rmseyer nd Wiley (99), Bernheim nd Whinston (998) nd Segl nd Whinston (000) mong others. This literture ws precipitted by, nd ultimtely qulified, Bork s (978) clim tht n exclusive deling greement cn only increse rther thn decrese consumer benefits becuse retil competition compels the retiler to ct s n gent for consumers. Bork resoned tht the mnufcturer would hve to compenste the retiler for ny loss in consumer benefits creted by exclusive deling, nd tht this requirement would mke nticompetitive exclusive deling unprofitble for the mnufcturer. With this line of resoning, the only exclusive deling greements tht would be consummted re those tht increse consumer welfre.

3 beverges. In the helth sector, insurers nd helth mintennce orgniztions my impose tight restrictions on ptients choice of drugs nd cre providers. The UK Competition Commission conducted survey of nerly 500 suppliers to grocery retilers in by the UK in 006. This survey reveled tht 35 percent of suppliers hd been sked to enter into n exclusivity greement by customer, nd [o]verll 9 percent of ll suppliers ctully entered into n exclusivity greement (006, p. 39). Lrge customers were the most ctive solicitors of exclusivity greements. Of those suppliers receiving these requests, twothirds were mde by one of the four lrgest supermrket chins in the UK. These responses include both exclusive deling nd exclusive distribution greements. 3 Buyer-induced exclusive deling rrngements hve different motivtions nd effects thn seller-imposed dels. In mny distribution chnnels, there is more mrket power t the distribution stge thn upstrem t the mnufcturing stge becuse consumers re more disposed to switch brnds within store thn switch stores within brnd (Steiner, 985). Lrge retilers or distributors my exercise buyer power to obtin preferentil terms of sle from suppliers tht re not vilble to smll buyers. Dobson (008) observed tht buyer-led restrints... occur most commonly... [when] the buyer holds some brgining dvntge over suppliers tht ensures their complince or consent (p. 93). One tctic for such buyer is to use the prospect of exclusivity to ply one supplier off ginst nother to reduce purchse prices s Glbrith (95) once rgued. Steuer (000) wrote tht lrge customer with buying power my nnounce to would-be suppliers tht it will commit to buy from only one of them nd tht if they hope to be selected they hd better offer their products on the most ttrctive terms.... It is n ll-or-nothing gme, with ech supplier 3 An exclusive deling rrngement occurs when the supplier is the retiler s exclusive source for some line of goods. An exclusive distributor rrngement occurs when the retiler is the supplier s exclusive customer. 3

4 knowing tht it must offer the best terms to obtin ny of tht customer s business (p.39-40). Similrly, Abbott nd Wright (forthcoming) note tht lrge buyers my use exclusive deling to intensify competition by mnufcturers for their business nd to improve purchse terms (p. 8). The U. K. Competition Commission s survey of suppliers in the UK grocery mrket provides some evidence tht lrge retilers pit suppliers ginst ech other in contests to win distribution. The Commission reported tht 30 percent of the suppliers indicted tht they hd tendered for business vi uctions (006, p. 65). Of those suppliers who climed they hve bid for business vi uction, 7 percent indicted tht the uction ws with one of the four lrgest supermrket chins. This survey lso indictes tht 79 percent of the suppliers who bid for business vi uction erned lower gross mrgins on these sles thn on other sles. When distributor or retiler commits to single supplier of brnded consumer good, exclusive deling reduces consumers choices. Fced with limited brnd selection, those consumers who do not find their preferred brnd in stock t retiler must either switch brnds or switch stores. Hving driven its cquisition costs down by plying one supplier off ginst nother, the retiler my reduce retil prices to discourge store switching nd encourge brnd switching (Klein nd Murphy, 008). Whether consumer welfre decreses becuse brnd selection is limited or increses becuse retil prices re lower is uncler priori. Tht is the question exmined in this pper. Both Klein nd Murphy nd Gbrielsen nd Sørgrd (999) investigte buyer-induced exclusive deling. Both show tht under certin conditions, exclusive distribution contrcts between retilers nd suppliers cn reduce retil prices of the exclusive brnd enough to increse consumer welfre even though some consumers do not purchse their preferred brnd. This 4

5 possibility is consistent with Glbrith s (95) erly rgument tht price concessions won by lrge buyers upstrem trnslte into lower prices for consumers downstrem. These issues re relted to those ddressed in two clusters of ppers in the literture on exclusive deling rrngements. The first cluster includes contributions by Mthewson nd Winter (987), O Brien nd Shffer (997) nd Bernheim nd Whinston (998). These ppers explore interctions in verticl structures similr to the one explored here (i.e., two mnufcturers nd single retiler), but where the inititive in verticl contrcting is ssigned to the upstrem suppliers rther thn to the downstrem buyer. Mthewson nd Winter s suppliers compete in liner triffs coupled with the option to offer the retiler n exclusive deling requirement. They find tht exclusive deling my rise in equilibrium, nd tht totl welfre my increse s result. O Brien nd Shffer, nd Bernheim nd Whinston do not restrict suppliers to liner triffs nd find tht where nonliner triffs re fesible, exclusive deling does not rise in equilibrium. This pper sks similr questions, but the inititive in verticl contrcting is ssigned to the buyer rther thn the suppliers. This is in keeping with the pper s focus on the presumed buyer power of lrge buyers. Ppers such s Mthewson nd Winter (997), Mrvel nd Yng (008), Dn (0) nd Chen nd Li (03) re in second cluster of relted ppers. These exmine the welfre effects of exclusive supply rrngements orchestrted by buying groups. Independent buyers of intermedite products form buying groups nd consolidte orders to exercise buyer power when purchsing from sellers of competing brnds. Hospitl buying groups, for instnce, re formed to negotite the purchse of hospitl supplies nd equipment for its members. Diverse stte nd municipl gencies often pool their purchsing in the sme wy. These buying groups exist minly becuse they re ble to leverge their sles volume to wrest dvntgeous terms from 5

6 suppliers. These orgniztions differ from lrge retilers or distributors with buyer power in tht they re comprised of end users of the goods purchsed. This pper exmines verticl structures in which retilers or distributors with buyer power resell rther thn consume the goods in question. Section II models the interctions between retiler with mrket power nd two suppliers of differentited goods s brgining gme where the retiler negotites with both suppliers simultneously. The outcome of this gme, given by the Nsh brgining solution, hs the retiler distributing the brnds sold by both suppliers nd mximizes the firms joint surplus. This outcome is tken s the benchmrk for comprison with the outcome produced when the retiler plys one supplier off ginst the other in pursuit of n exclusive del. Section III models interctions between the retiler nd the suppliers s three-stge non-coopertive gme in which the retiler hs the option to commit to choosing single supplier before it solicits terms of sle from the suppliers. The nlysis produces generlized necessry nd sufficient condition for predicting when the retiler opts for inducing n exclusive deling rrngement. In section IV, specific demnd structure is introduced to exmine more thoroughly the implictions of this necessry nd sufficient condition. This demnd structure incorportes elements of both verticl nd horizontl product differentition. The nlysis shows tht the retiler opts for n exclusive supply rrngement when consumers preferences for the brnds they fvor re not too strong. Consumers re not necessrily injured if the retiler pursues n exclusive supply rrngement. Depending on prmeter vlues, the retiler s preference for exclusive deling my increse or reduce consumer welfre. Also, some prmeter vlues show the retiler preferring to distribute both brnds when consumer welfre would be greter with 6

7 buyer-induced exclusive deling. In short, the interests of consumers my or my not line up with those of retiler who induces exclusive deling. II. Brgining without Exclusivity The verticl structure consists of single retiler or distributor nd two mnufcturers tht produce differentited brnds of the sme good. The mnufcturers hve no fixed costs nd hve constnt mrginl costs which, for nlyticl ese, we ssume re zero. The retiler hs mrket power when reselling the good to consumers. This mrket power my be due to the firm s size, loction, or other distinguishing chrcteristics. 4 Or in the cse of grocery stores, mss merchndisers nd the like, some degree of mrket power is due to consumers shopping for severl items simultneously insted of single items. Shopping for severl items simultneously conserves shopping costs, but it reduces consumers in-store demnd elsticities for specific goods. 5 In ddition, lrge retilers hve significnt popultions of loyl customers who incur switching costs if they shop elsewhere. 6 The retiler s operting costs re fixed nd sunk. The mrginl cost of hndling nd reselling unit of either mnufcturer s good is constnt nd zero. Consumers inverse demnds f i( q,q )for the two goods t the retiler s estblishment hve fori, j,. fi fi f 0, 0 nd 0 q q q q i j i j 4 Size is essentil. Inderst nd Shffer (008, p. 630) write tht [t]he min source of buyer power... is the bility to substitute wy from ny given supplier s input.... [I]n generl the profitbility nd thus the credibility of substitution should increse with the buyer s reltive size. 5 Bliss (988, p. 38) identifies this cptive buyer effect s contributing fctor to retilers mrket power in the sle of specific goods. 6 Dobson (005) ttributes some of the buyer power cquired by lrge retilers tht distribute mny products to the symmetry between the lrge number of products nd suppliers these firms hve nd the smll number of products nd customers served by the suppliers. 7

8 Events in the verticl structure tke plce in two stges. At the first stge the retiler negotites simultneously nd seprtely with ech mnufcturer. These negotitions determine the quntity q i 0of ech good the retiler cquires nd the pyment T i 0 ech mnufcturer receives from the retiler. This representtion llows the firms to negotite nonliner pricing schedules. At the second stge the retiler sets the retil price p i f i ( q,q ) 0,i, of ech good. The retiler cnnot price discriminte mong consumers. The outcome of the firms negotitions is given by the Nsh brgining solution. Following Chipty nd Snyder s (999) formultion of Nsh brgining when single gent (in this cse, buyer) brgins simultneously with two or more gents (in this cse, suppliers), the firms exchnge those quntities of ech good tht mximize their joint surplus. The firms divide tht surplus ccording to the rule: the retiler nd ech mnufcturer eqully divide the incrementl surplus creted by their greement on the ssumption tht the retiler nd the other mnufcturer exchnge the surplus-mximizing quntity. 7 Although it is unnecessry to go into detils, the Nsh brgining solution is formlly implemented by dynmic, noncoopertive brgining gme, s Rubinstein (98) demonstrted. Let i i, V( q,q ) q f ( q,q ) be the firms strictly concve joint surplus function, nd let * * ( q,q )be the unique quntities tht mximize this surplus: q * rg mx[v( x,q * )] nd q * rg mx[v( q *,x )] () x x 7 In ddition to Chipty nd Snyder (999), similr formultions of simultneous bilterl brgining hve been used by Horn nd Wolinsky (988), Stole nd Zwiebel (996) nd Inderst nd Mzzrotto (008). The interctions between retiler nd its mnufcturers hve been modeled in vrious wys. When there is single prty on one side nd more thn one prty on the other side, Whinston (006) distinguishes multi-stge, noncoopertive gmes where the single prty mkes tke or leve offer to other prties in the first stge from gmes where the other prties submit bids in the first stge. O Brien nd Shffer (997) model these interctions by llowing one prty to mke tke or leve offer of menu of contrcts from which the other prties choose. 8

9 These re the equilibrium quntities exchnged in the Nsh brgining solution. The distribution of * * V( q,q )mong the firms in this equilibrium depends on the firms incrementl pyoffs. To clculte these pyoffs, ssume tht the retiler nd mnufcturer i believe tht negotitions between the retiler nd mnufcturer j led to the efficient quntity q * j. 8 These beliefs men tht the firms nticipte tht mnufcturer j nd the retiler will exchnge q * j units even if their own negotitions brek down so tht qi 0. With this, the incrementl pyoff shred by the retiler * * * nd mnufcturer is V( q,q ) V(0,q ) becuse q 0if negotitions brek down between tht mnufcturer nd the retiler. Similrly, the incrementl pyoff shred by the retiler nd * * * mnufcturer isv( q,q ) V( q,0 ). The equilibrium pyments(t,t )re clculted to distribute the incrementl pyoff of * * ech trnsction eqully between the prties to the trnsctions: T V( q *,q * ) * * * * V(0,q ) nd V( q,q ) T V( q,0 ) () * * These pyments re the mnufcturers profits: * T * nd * T * (3) The retiler s prices re p * f ( q *,q * ) nd p * f ( q *,q * ), nd its profit is: * * * * * * * V( q,0 ) V(0,q ) R V( q,q ) T T (4) The vlues in equtions () (4) represent the outcome if the firms strike n efficient brgin without provision for n exclusive supplier. They provide benchmrk for comprison with the outcome, considered next, tht ccompnies n exclusive supplier rrngement. 8 McAfee nd Schwrtz (994) cll these pssive beliefs. 9

10 III. Choosing n Exclusive Supplier Even though simultneous bilterl brgining in this verticl structure llows the firms to mximize their joint surplus, the retiler s profit my be greter if the firm forgoes negotiting with both mnufcturers nd insted pursues n exclusive supply contrct with one of the mnufcturers. The retiler s mrket power mens tht some consumers could be prompted to switch brnds rther thn switch stores if the retiler does not offer their preferred brnd. By soliciting bids from the mnufcturers for n exclusive supply contrct, the retiler cn exploit wht Glbrith clled the opportunity of strong buyer to ply one seller off ginst the other (95, p. 3). In effect, the retiler cn refuse to contrct with both mnufcturers nd insted wrd n exclusive supply contrct to the mnufcturer who offers the retiler the best del. With this, events in the verticl structure tke plce in three stges. At the first stge the retiler solicits( q i,t i )bids from ech mnufcturer. 9 The mnufcturers submit bids in the second stge tht specify both qi ndt i. At the third stge the retiler contrcts with one of the mnufcturers nd sets the relevnt retil price. This gme hs perfect equilibrium in which one of the mnufcturers wins the contrct to supply the retiler. The firms equilibrium strtegies re found using bckwrd recursion. Suppose tht the retiler commits to purchsing from single mnufcturer t the first stge, nd tht the mnufcturers submit bids( q i,t i )t the second stge. Then t the third stge, the retiler ccepts mnufcturer s offer if [V( q,0 ) T ] [V(0,q ) T ] nd sets p f ( q,0 ). With this 9 It is not unusul for lrge buyers nd their suppliers to hve long-term, fixed-quntity contrcts. Noll (005, p. 603), for instnce, observes tht lrge buyers often do not exercise their dominnce by posting low buying price nd witing for sellers to rrive. Insted the common prctice is for buyers nd sellers to negotite long-term contrct tht specifies both price nd quntity. 0

11 outcome, the retiler s profit isv( q,0 ) T nd mnufcturer s profit ist. If [V( q,0 ) T ] [V(0,q ) T ], the retiler ccepts mnufcturer s offer nd sets p f (0,q ). Here the retiler s profit isv( 0,q ) T nd mnufcturer s profit ist. If [V( q,0 ) T ] [V(0,q ) T ], the retiler chooses supplier t rndom. Now consider the second stge, gin ssuming tht the retiler commits to purchsing from single mnufcturer t the first stge. At the second stge, if mnufcturer i offers ny ( q i,t i )which is less profitble for the retiler thn some j j ( q,t )wheret j 0, mnufcturer j would respond by mking n offer tht is more profitble for the retiler thn( q i,t i ). Of course, neither mnufcturer would mke n offer( q i,t i )wheret i 0. Let ( q ˆ ˆ,q )be the unique quntities: qˆ rg mx[v( x,0 )] nd qˆ rg mx[v(0,x )] (5) x These stnd lone quntities mximize the joint surplus of the retiler nd one mnufcturer when the retiler does not contrct with the other mnufcturer. With no loss of generlity, ssume tht good is the more populr brnd in the sense tht it offers the greter stnd lone surplus: V( q ˆ ˆ,0 ) V(0,q ). Becuse mnufcturer will not offer the retiler ny( q ˆ,T )where x T 0, mnufcturer cn profitbly undercut ny offer tht mnufcturer mkes. In equilibrium, mnufcturer offers, nd the retiler ccepts, the bid( q ˆ ˆ,T )where: Tˆ V( q ˆ,0 ) V(0,q ˆ ) (6) The retiler chrges consumers the price ˆp f ( q ˆ,0 ) nd erns ˆ V( 0,q ˆ ) (7) R

12 in profit. Tht is, the retiler contrcts with the mnufcturer tht cn provide the greter stnd lone surplus, but only retins profit equl to the stnd lone surplus tht the excluded mnufcturer might hve provided. Mnufcturer s profit is ˆ T ˆ (8) nd mnufcturer hs no sles nd erns no profit. This equilibrium is the outcome of Bertrnd-like competition between mnufcturers for n exclusive supply contrct with the retiler. In symmetric verticl structures like the one in this model, it is plusible (nd in fct generlly ssumed) tht the single prty cn initite the form of negotitions with the verticl level tht hs two or more prties. The retiler hs the option to brgin simultneously with both mnufcturers for non-exclusive terms, or to commit itself to single source of supply nd solicit n exclusive deling rrngement with one of the mnufcturers. Whether the retiler opts for exclusive deling depends on the firm s comprtive profit in the two equilibri. * Compring R nd ˆ in equtions (4) nd (7) gives: R Proposition : The retiler opts for n exclusive supplier iff V( q,0 ) V(0,q ) V(0,q ˆ ) This necessry nd sufficient condition for the retiler to seek n exclusive deling rrngement is tht the stnd lone surplus with the excluded good is no less thn the verge of the surpluses defined by the firms disgreement points if the retiler brgins without n exclusivity provision. Ech of the stnd lone quntities is greter thn the corresponding quntity when the joint surplus of ll three firms is mximized (See the Appendix for Proof):

13 Proposition : ˆq q for i, i i This Proposition indictes tht when the retiler opts for n exclusive supplier, the retiler s unit sles of tht supplier s brnd is greter thn where both brnds re distributed: ˆq q *. Aprt from this result, it is not possible to compre outcomes in the two equilibri without further restricting the demnd functions f i ( q,q ),i,. In principle, the price of the exclusive good ˆp my be greter or less thn the price of the sme good p * if the retiler negotites contrcts with both mnufcturers. In order to nlyze more thoroughly the retiler s option to choose n exclusive supplier, nd to explore welfre implictions of this business strtegy, it is necessry to depict consumers demnd for the goods more completely. The next section considers demnd specifiction tht extends the nlysis. IV. A Useful Demnd Specifiction There re two types of consumers. Type consumers prefer brnd over brnd, nd conversely for type consumers. The retiler hs continuum [0, ] of consumers where (,) re type nd the remining re type. This is in keeping with the previous ssumption tht brnd is more populr. Every consumer of either type hs tste prmeter θ, where θ is uniformly distributed on [0, ]. Consumers purchse single unit of one brnd or else purchse nothing. The retiler cnnot observe consumers tste prmeters or types, nd so cnnot price discriminte mong consumers. A representtive type consumer's utility is p if she buys unit of brnd nd is p if she buys unit of brnd, where. If she purchses neither good, the consumer s 3

14 utility is 0. A representtive type consumer's utility is pif she buys unit of brnd, p if she buys unit of brnd, nd 0 if she buys neither good. These preferences llow for both verticl nd horizontl differentition. Within types, ll consumers prefer the sme brnd, lbeit with different intensities. This is the verticl element. Across types, consumers rnk the brnds differently. This is the horizontl element. IV. A. Distributing Both Brnds If the retiler elects to distribute both brnds, the firm brgins simultneously with ech mnufcturer nd sells the quntities( q,q )defined by eqution (). To find these quntities, * * we must derive the demnds for both brnds by consumers of ech type. Let q be the number of units of brnd j sold to type i consumers so tht q q q. j j j Consider type consumers first. Ordering consumers by descending vlues of, type consumers inverse demnd for brnd is: q ( ), (9) neglecting the brnd option. Similrly, neglecting the brnd option, type consumers inverse demnd for brnd is: q ( ). (0) Where both brnds re vilble, type consumers demnds for the brnds depend on both p nd p. Non-negtive quntities of q nd q require prices tht stisfy: p [0,] nd p [0, p ]. () i j 4

15 For prices tht stisfy (), type consumers demnds re derived from two conditions bsed on equtions (9) nd (0): q q ( ) p ( ) p q q p 0. () (3) These conditions re depicted in Figure. Condition () identifies the mrgin between those type consumers who choose brnd nd those who choose brnd. A type consumer is indifferent between purchsing unit of brnd nd brnd when the consumer s surplus is the sme with either purchse. Condition (3) identifies the mrgin between those type consumers who choose brnd nd those who purchse neither brnd. A type consumer is indifferent between purchsing unit of brnd nd purchsing neither brnd when prices re such tht brnd confers no surplus. Combining equtions () nd (3) gives type consumers inverse demnds for ech brnd over q [0,] nd q [0, ] : q q q q f ( ) nd f (4) With these demnds for type consumers, it follows tht (See the Appendix for Proof): Lemm : The firms joint surplus from sles to type consumers is mximized when q nd q 0; these quntities re sustined by prices p nd p [,]. Cpturing s much surplus s possible from sles to type consumers involves selling t prices tht produce no sles of brnd. 5

16 Now consider sles to type consumers. Using the sme resoning s before with type consumers, we cn derive type consumers demnds to get the nlogous result (the proof is nlogous to the proof of Lemm ): Lemm : The firms joint surplus from sles to type consumers is mximized when by prices p [,] nd p. q nd 0 q ; these quntities re sustined Tken together, Lemms nd indicte tht mximizing the firms joint surplus from sles to ll consumers involves no consumer purchsing unit of their less preferred brnd: Proposition 3: The firms joint surplus from sles to ll consumers is mximized t q nd q ; these quntities re sustined by prices p p. The vlues in Proposition 3 re the equilibrium prices nd quntities under simultneous bilterl brgining between the retiler nd the mnufcturers. When the quntities in Proposition 3 re produced the firms joint surplus is * * V( q,q ). To find how this surplus is distributed mong the firms in equilibrium, we must 4 use the disgreement pyoffs: * * * V( q,0 ) pq 4 * * * ( ) V(0,q ) pq 4 (5) These vlues indicte tht the incrementl surplus ttributble to the trnsction between the retiler nd mnufcturer is: 6

17 * * * V( q,q ) V(0,q ). 4 (6) Similrly, the incrementl surplus ttributble to the trnsction between the retiler nd mnufcturer is * * * ( ) V( q,q ) V( q,0 ). 4 (7) The retiler s pyments to the mnufcturers(t,t )re clculted to distribute the incrementl * * surplus of ech trnsction eqully between the retiler nd the relevnt mnufcturer: * * * * V( q,q ) V(0,q ) T 8 * * * * V( q,q ) V( q,0 ) ( ) T 8 (8) * * * * The retiler retins the residul surplusv( q,q ) T T. In sum, the firms profits in the equilibrium with simultneous bilterl brgining re: * * ( ) *, nd R (9) IV. B. Distributing Single Brnd Now suppose the retiler elects t the outset to contrct with n exclusive supplier insted of distributing both brnds. To determine whether the retiler s profit would be greter thn * R if the firm opts for n exclusive supply rrngement, we must compre * R in eqution (9) to ˆ R s defined in eqution (4). To clculte ˆ R, recll tht the retiler contrcts with the mnufcturer tht cn provide the greter stnd lone surplus, but only retins profit equl to the stnd lone surplus tht the excluded mnufcturer might hve provided. The retiler contrcts with 7

18 mnufcturer if V( q ˆ ˆ,0 ) V(0,q ) ; otherwise the firm contrcts with mnufcturer. The retiler s profit ˆ is the lesser of V( q ˆ,0 ) nd V(0,q ˆ ). R The quntities qˆ nd qˆ re defined by eqution (5). If the retiler sells only brnd, then the inverse demnds for tht brnd on the prt of type nd type consumers re: q q f ( ) nd f. (0) To get the totl inverse demnd for brnd when brnd is not offered, we invert the functions (0), dd them together to get the totl demnd for the brnd, nd then invert bck: f ( q ) () Using (), the vlue of q tht mximizesv( q,0 ) nd the mximized vlue ofv( q,0 ) re: qˆ nd V( q ˆ,0 ) 4( ) () A similr derivtion for brnd when brnd is not offered yields: qˆ nd V( 0,q ˆ ) 4( ) (3) With nd, equtions () nd (3) indicte thtv( q ˆ,0 ) V(0,q ˆ). This mens tht if the retiler seeks n exclusive supply rrngement, mnufcturer will be the exclusive supplier (See the Appendix for Proof): 8

19 Proposition 4: If the retiler opts for n exclusive supplier, the firm buys ˆq B( )( ) units from mnufcturer for ˆT. 4( )( ) The retiler chrges consumers ˆp. ( ) If the retiler opts for n exclusive supplier, mnufcturer s brnd is excluded from the retiler s distribution. Using equtions (7) nd (8), the firms profits re: ˆ T ˆ, ˆ 0 nd ˆ R 4( ) (4) Compring Propositions 3 nd 4 confirms tht ˆq * q,s indicted by Proposition. * * The propositions lso show tht ˆq q q. The retiler sells more units of the more populr brnd with n exclusivity rrngement, but the firm s totl unit sles re the sme whether or not it chooses n exclusive supplier. The retil price of the selected brnd is less thn where the retiler distributes both brnds: ˆp p *. The price p flls just enough for incresed unit sles of the more populr brnd to offset the displced unit sles of the less populr brnd. IV. C. The Retiler s Choice The retiler will elect to pursue n exclusive supply rrngement ifˆ. Otherwise * R R the firm will engge in simultneous bilterl brgining with the mnufcturers. Compring * R nd ˆ in equtions (9) nd (4) indictes whether the retiler opts for n exclusive deling R rrngement: Proposition 5: The retiler opts for n exclusive supplier iff 9

20 To interpret Proposition 5, notice tht the prmeters nd correspond, respectively, to the extent of verticl nd horizontl product differentition between the mnufcturers brnds. The prmeter is mesure, within types, of the intensity of consumers preferences for the brnds they fvor. The greter is, the greter is the premium consumers re willing to py for their fvored brnd. This prmeter reflects the extent of verticl product differentition. The prmeter reflects the size symmetry of the two consumer groups. A smller vlue of indictes tht consumers preferences reflect greter horizontl product differentition. But s, horizontl product differentition disppers ltogether. Proposition 5 shows tht if consumers brnd preferences re wek ( is smll), the retiler opts for n exclusive supply rrngement. In lieu of distributing both brnds, it is more profitble for the retiler to distribute only brnd nd chrge ˆp slightly below p * to induce some consumers to switch brnds. If consumers brnd preferences re strong enough, then the retiler exploits them by distributing both brnds rther thn selling only brnd with ˆp well below p *. This result explins why retiler s distribution strtegy my be different for different consumer good ctegories. For instnce, convenience store tht sells only one brnd of light bulbs my sell more thn one brnd of beer becuse consumers brnd preferences re more pronounced for beer thn light bulbs. Proposition 5 shows tht the miniml strength of consumers brnd preferences for the retiler to opt for distributing both brnds depends on the prmeter. The threshold vlue of is lower where is lrge becuse one of the brnds is substntilly more populr tht the other. When one brnd is fvored by lrge mjority of consumers, the retiler does not wnt to reduce the price of tht brnd enough to cuse the minority to switch brnds. 0

21 IV. D. Comprison with Previous Results Klein nd Murphy nd lso Gbrielsen nd Sørgrd explored issues similr to those in this pper. 0 Some of the results in this pper re consistent with the findings these uthors reported. The results in these erlier ppers, like those here, pertin to specific stylized models of consumer demnd nd to interctions between two mnufcturers nd single retiler. In this pper, the retiler uses the lure of n exclusive supply rrngement to ply the mnufcturers off ginst ech other nd win dvntgeous terms. Klein nd Murphy lso ssume tht, hving committed to the selection of n exclusive supplier, the retiler plys the mnufcturers off ginst ech other. Gbrielsen nd Sørgrd do not llow this kind of plying off. Both Klein nd Murphy nd by Gbrielsen nd Sørgrd begin with reduced-form ggregte demnd functions nd ssume tht in the bsence of n exclusivity rrngement, mnufcturers wholesle prices re determined by Bertrnd interctions. However, these ppers re dissimilr in regrd to how retil prices re set. Gbrielsen nd Sørgrd ssume tht the retiler mrks wholesle prices up in double-mrginliztion fshion for resle to consumers. Klein nd Murphy ssume tht downstrem competition compels the retiler to chrge retil prices tht merely cover costs. Gbrielsen nd Sørgrd s chrcteriztion of retil pricing ssumes tht the firms cnnot void squndering significnt shre of the joint surplus ltent in the verticl structure. Klein nd Murphy s ssumption tht retil prices re competitive rules out ny exercise of downstrem mrket power. In this pper, retil prices re determined differently depending on whether the retiler opts for n exclusive supplier. If the retiler distributes both 0 Klein nd Murphy s (008) nlysis of buyer-induced exclusive deling is not the focl point of their pper. The min thrust is tht retiler s bility to shift incrementl sles from one brnd to nother cn reduce wholesle nd retil prices.

22 brnds, simultneous bilterl brgining drives retil prices to levels tht mximize the firms joint surplus. If the retiler opts for exclusive deling, the retiler exercises its downstrem mrket power by setting retil prices unilterlly. A comprison of Propositions 3 nd 4 indictes tht buyer-induced exclusive deling reduces the retil price of the brnd selected by the retiler. Both Klein nd Murphy nd Gbrielsen nd Sørgrd get similr result. Although Klein nd Murphy s nlysis provides no indiction tht retiler would ever reject n opportunity to contrct with n exclusive supplier, Gbrielsen nd Sørgrd model the retiler s decision bout whether to use exclusive deling. They predict tht the retiler is more likely to seek n exclusive supplier where one of the brnds is preferred by significnt mjority of consumers. Proposition 5 mkes the opposite prediction becuse plying off is less remunertive for the retiler when few consumers prefer the second brnd. Gbrielsen nd Sørgrd lso predict tht the retiler is more likely to seek n exclusive supplier where the brnds re strongly differentited. This prediction is counterintuitive. It mens tht the retiler denies consumers vriety where consumers vlue vriety most. Proposition 5 predicts tht the retiler will not opt for n exclusive supplier where the brnds re strongly differentited ( is lrge). IV. E. Welfre Effects The retiler chooses unilterlly whether to pursue n exclusive supply rrngement, but the firm s choice hs pronounced redistributive effects for other mrket prticipnts. The effect on the suppliers is the lest surprising. If exclusive deling is profitble for the retiler, then it reduces profits for both suppliers (See the Appendix for Proof): Mthewson nd Winter (987) find tht seller-induced exclusive deling reduces wholesle prices but my or my not reduce retil prices.

23 * Proposition 6: ˆ i i for i, iff The reduction in profit is more extreme for mnufcturer thn for mnufcturer becuse brnd is not distributed t ll when the retiler contrcts exclusively with mnufcturer. Consumers lso re ffected by the retiler s decision bout whether to pursue n exclusive supply rrngement. Consumer welfre my be greter or less when the retiler opts for n exclusive supplier (See the Appendix for Proof): Proposition 7: Consumer welfre is greter with n exclusive supplier iff 4 4 This Proposition shows tht if consumers brnd preferences re sufficiently strong (i.e., is lrge), n exclusive supply rrngement increses consumer welfre. This hppens even though consumers who prefer the excluded brnd lose tht option. With exclusive deling, the retiler cuts the price of the exclusive brnd to induce some of these consumers to switch. The requisite strength of consumers preferences for consumer welfre to increse with exclusive deling is greter the smller is the consumer group who prefer the excluded brnd (i.e., is lrge). Even where the strength of consumers preferences is not gret, n exclusive deling rrngement increses consumer welfre if one of the goods is fvored by lrge mjority. A comprison of Propositions 5 nd 7 shows tht the interests of the retiler nd its consumers re sometimes, but not lwys, ligned. The prmeter thresholds in these Propositions re depicted in Figure. Proposition 5 shows tht the retiler chooses n exclusive supplier when prmeter vlues fll below the upper threshold, but elects to distribute both brnds when prmeter vlues lie bove this threshold. Similrly, Proposition 7 shows tht 3

24 consumer welfre is greter if the retiler contrcts with n exclusive supplier when prmeter vlues re bove the lower threshold, nd conversely for vlues below this threshold. The prmeter vlues tht fll in between the thresholds re those where the retiler tkes on n exclusive supplier nd consumers re the beneficiries of this decision. Depending on the reltive size of the two consumer groups, this lignment of the retiler s nd consumers interests occurs when consumers re willing to py percent more for their preferred brnd thn the lterntive. With sufficiently high vlues of, the retiler will not contrct with n exclusive supplier even though consumers prefer tht outcome. And with sufficiently low vlues of, the retiler will contrct with n exclusive supplier even though consumers prefer the two-brnd outcome. Even though consumer welfre my increse when the supplier opts for n exclusive supplier, totl welfre does not. If exclusive deling is profitble for the retiler, then it reduces totl welfre (See the Appendix for Proof): Proposition 8: Totl welfre is less with n exclusive supplier iff The welfre implictions of Propositions 7 nd 8 gree t some but not ll points with those in the ppers by Klein nd Murphy nd by Gbrielsen nd Sørgrd. Klein nd Murphy s model suggests tht buyer-induced exclusive deling lwys increses consumer welfre s well s totl welfre. In Gbrielsen nd Sørgrd s model, buyer-induced exclusive deling my hve positive effect on totl welfre but will reduce welfre if the products re sufficiently differentited (999, p. 35). Zenger (00) shows tht this result depends on the ssumption in Klein nd Murphy s model tht the demnd for the vilble brnds is symmetric. If one of the brnds is notbly more populr, then exclusive deling my reduce rther thn increse totl welfre. 4

25 V. Conclusion The motivtion nd effects of exclusive deling rrngements tht re solicited by lrge retilers nd distributors with mrket power re different thn exclusivity rrngements orchestrted by mnufcturers nd suppliers with mrket power. When retiler plys the suppliers of different brnds of consumer good off ginst ech other by offering exclusive distribution to win dvntgeous wholesle pricing, there re mixed implictions for consumers. The first impliction of buyer-induced exclusive deling is tht consumers served by the retiler hve fewer brnd choices. The excluded brnds re the lest populr. At the sme time, the nlysis here indictes tht exclusivity reduces the retil price of the brnd tht is selected for exclusive distribution. Retil prices for brnds supplied exclusively re reduced to encourge those customers who prefer n excluded brnd to switch brnds rther thn switch stores. Consumers who prefer the brnd selected for exclusive distribution re better off, but those who prefer n excluded brnd my or my not be better off. Totl consumer welfre my increse or decrese s result of this exclusivity. The conditions tht cuse consumer welfre to increse when retiler selects n exclusive supplier re different thn, but overlp with, the conditions tht cuse the retiler to opt for n exclusive supplier. The retiler s interest nd the interest of consumers re in lignment unless consumers brnd preferences re either too wek or too strong. If consumers hve sufficiently wek brnd preferences, the retiler distributes single brnd even though the price of the exclusive brnd is not so low tht consumers prefer tht option to more brnd vriety. If consumers hve sufficiently strong brnd preferences, the retiler distributes both brnds in order 5

26 to exploit those preferences with higher prices even though consumer minority would prefer second option even t higher price. The possibility tht buyer-induced exclusive deling reduces retil prices enough to increse consumer welfre even though some consumers do not purchse their preferred brnd hrkens bck to Glbrith s notion tht the counterviling power of lrge retilers is beneficil to consumers. Glbrith s clim does not pply to every lrge retiler or distributor with buyer power, but reserch on the downstrem effects of counterviling power supports his clim in certin circumstnces. Heretofore, ll such circumstnces hve required the preservtion of competition t the retil level even though the lrge retiler exercises counterviling power upstrem. 3 This pper identifies different mechnism whereby counterviling power reduces retil prices nd increses consumer welfre. A lrge retiler or distributor my reduce cquisition costs by plying suppliers off ginst ech other nd offering consumers limited selection of brnds. The firm reduces retil prices to encourge brnd switching in lieu of store switching. If retil prices re reduced sufficiently, consumer welfre my increse. The nlysis in this pper hs implictions for verticl reltionships in mrkets where firms bundle their own goods X with complementry products Y. When complementry products re differentited nd re produced by two or more independent sellers, the producer of X my bundle X with the Y of single supplier. This denies the firm s customers the opportunity to choose different brnd of Y when they purchse X. For instnce, utomobile mnufcturers select the mnufcturers of the tires nd udio systems tht re instlled in new crs. Microsoft selects the softwre products tht occupy the Windows desktop on new personl computers. A hospitl my select the nesthesiology prctice tht its ptients must use for surgicl procedures, 3 For exmple, see von Ungen-Sternberg (996), Dobson nd Wterson (997), Chen (003) nd Mills (03). 6

27 nd house pinter my select the brnd of pint s(he) uses on job. Where the producer of X chooses brnd of Y for its customers, rther thn llowing the customers to choose, the firm my leverge its bility to choose to win dvntgeous terms of sle from Y producers. The welfre effects of this prctice re similr to buyer-induced exclusive deling in the distribution sector. Exclusive deling rrngements tht re sought by suppliers with mrket power sometimes hve nticompetitive exclusionry effects. It is unlikely tht n exclusive deling rrngement instigted by retiler or distributor would be motivted by the gol of excluding suppliers or impeding the entry of new suppliers. 4 But it is worth sking whether dominnt retiler or distributor s buyer-induced exclusive deling might nevertheless be exclusionry becuse it restricts the distribution of less populr brnds or the brnds introduced by new entrnts. It is not likely tht buyer-induced exclusive deling hs serious exclusionry effects. Even when the consumer popultion served by lrge retiler hs (, ) combintion tht mkes exclusive deling dvntgeous for tht retiler, this does not prevent the brnds excluded by tht retiler from being distributed by other retilers who serve different customer popultions. And of course, the retiler is not interested in driving the excluded supplier out of business becuse tht would strengthen the strtegic position of the retiler s exclusive supplier. At present, the consensus view of exclusive deling rrngements is tht they cn be either nticompetitive or efficiency-promoting depending on severl fctors. This view is predicted minly on the nlysis of seller-induced exclusive deling. This pper s nlysis of buyer-induced exclusive deling only reinforces the consensus view. 5 Exclusive deling rrngements rise for different resons in different commercil environments, nd their effects 4 This kind of exclusion is the primry concern of the ppers cited in footnote. 5 In his survey of wide vriety of buyer-led verticl restrints, Dobson (008) reches similr conclusion. 7

28 on competition nd welfre re not lwys the sme. Submitting these prctices to the rule of reson, rther thn per se prohibition, remins the best ntitrust response. 8

29 p ( q ) p p = = ( q ) q q q q Figure 3 β Figure 9

30 Appendix Proposition : ˆq q for i, i i Proof: Reclling thtv( q,q ) q f ( q,q ),we hve: i i, * * * * * * V( q,q ) * * * f ( q,q ) * f ( q,q ) f ( q,q ) q q 0. (5) q q q * Becuse q 0nd f 0, eqution (5) implies tht: q * * * * * f ( q,q ) f ( q,q ) q 0 q (6) Also, we hve: V( q ˆ,0 ) f ˆ ( q,0 ) f ˆ ˆ ( q,0 ) q 0. (7) q q * f f Becuse q 0nd becuse 0 nd 0, equtions (6) nd (7) indicte tht ˆq q. q q q Similrly, ˆq q. Lemm : The firms joint surplus from sles to type consumers is mximized with q nd q 0; these quntities re sustined by prices p nd p [,]. Proof: The firms joint surplus from sles to type consumers is p q equtions () nd (3): p q, or using q q q q V( q,q ) q ( ) + q. 30 (8)

31 First order conditions tht re sufficient for mximizing this expression re: V q q q 0 q V q q 0 q (9) Solving the equtions in (9) simultneously gives q nd q 0, which quntities re demnded by type consumers with p nd p [,]. Proposition 4: If the retiler opts for n exclusive supplier, the firm buys ˆq units from B( )( ) mnufcturer for ˆT. The retiler chrges consumers 4( )( ) ˆp. ( ) Proof: Equtions () nd () together show tht ˆq nd ˆp. ( ) Substituting the vlues of V( q ˆ,0 ) nd V(0,q ˆ ) in eqution () nd (3) into eqution (6) gives B( )( ) ˆT. 4( )( ) * Proposition 6: ˆ i i for i, iff Proof: Equtions (9) nd (3) show thtˆ (4) indicte thtˆ * iff: *. Also, Proposition 4 nd equtions (8) nd 3

32 ( )( ). ( )( ) (30) This inequlity holds for ny (,) nd. 4 Proposition 7: Consumer welfre is greter with n exclusive supplier iff. 4 Proof: Proposition 3 gives the vlues of prices nd quntities in the equilibrium without exclusivity. With these vlues, consumers surplus is: S* (3) 8 Proposition 4 gives the vlues of ˆp nd q ˆ in the equilibrium with n exclusive supplier. With these vlues, consumers surplus is: Ŝ ( ) (3) Compring expressions (3) nd (3), we get: Ŝ S* which estblishes the result. 4 iff 4 (33) Proposition 8: Totl welfre is less with n exclusive supplier iff Proof: The prices nd quntities in Proposition 3 indicte tht totl welfre in the equilibrium without exclusivity is: 3 W* (34) 8 3

33 The vlues of ˆp nd q ˆ in Proposition 4 indicte tht totl welfre in the equilibrium with n exclusive supplier is: Ŵ ( ) (35) Compring expressions (34) nd (35), we get: Ŵ W * iff (36) which estblishes the result. 33

34 References Abbott, A. F. nd Wright, J. D. (forthcoming). Antitrust Anlysis of Tying Arrngements nd Exclusive Deling. Antitrust Lw nd Economics. Hylton, K. N., ed. Northhmpton, MA: Edwrd Elgr Aghion, P., nd Bolton, P. (987) Contrcts s Brrier to Entry. Americn Economic Review, 77, Bernheim, B. D. nd M. D. Whinston (998). Exclusive Deling. Journl of Politicl Economy, 06, Bliss, C. (988). A Theory of Retil Pricing. Journl of Industril Economics, 36, Bork, R. H. (978). The Antitrust Prdox. New York: Bsic Books Chen, Y. nd Li, X. (03). Group Buying Commitment nd Sellers Competitive Advntges. Journl of Economics nd Mngement Strtegy., Chen, Z. (003). Dominnt Retilers nd the Counterviling-Power Hypothesis. RAND Journl of Economics, 34, 6-65 Chipty, T., nd C. M. Snyder (999). Buyer Size nd Brgining Power. Review of Economics nd Sttistics, 8, Dn, J. (0). Buyer Groups s Strtegic Commitments. Gmes nd Economic Behvior. 74, Dobson, P. W. nd M. Wterson (997). Counterviling Power nd Consumer Prices. The Economic Journl, 07, Dobson, P. W. (005). Exploiting Buyer Power: Lessons from the British Grocery Trde. Antitrust Lw Journl, 7, Dobson, P.W. (008) Buyer-Led Verticl Restrints. Issues in Competition Lw nd Policy. Collins, W. D. ed. Chicgo: Americn Br Assocition, Section of Antitrust Lw, Gbrielsen, T. S. nd Sørgrd, L. (999) Discount Chins nd Brnd Policy. Scndinvin Journl of Economics, 0, 7-4 Glbrith, J. K. (95). Americn Cpitlism: The Concept of Counterviling Power. Boston: Houghton Mifflin Horn, H., nd Wolinsky, A. (988). Bilterl Monopolies nd Incentives for Merger. RAND Journl of Economics, 9,

35 Inderst, R. nd Mzzrotto, N. (008). Buyer Power in Distribution. (In W. D. Collins (ed.), Issues in Competition Lw nd Policy, Vol. II (pp ). Chicgo: Americn Br Assocition.) Inderst, R. nd Shffer, G. (008). Buyer Power in Merger Control. (In W. D. Collins (ed.), Issues in Competition Lw nd Policy, Vol. II (pp ). Chicgo: Americn Br Assocition.) Inderst, R. nd Vlletti, T. M. (009). Price Discrimintion in Input Mrkets. RAND Journl of Economics, 40, -9 Klein, B. nd Murphy, K. M. (008). Exclusive Deling Intensifies Competition for Distribution. Antitrust Lw Journl 75, Mrvel, H. (98). Exclusive Deling. Journl of Lw & Economics, 4, -5 Mrvel, H. P., nd H. Yng (008). Group Purchsing, Nonliner Triffs, nd Oligopoly. Interntionl Journl of Industril Orgniztion, 6, Mthewson, G. F., nd R. A. Winter. (987). The Competitive Effects of Verticl Agreements: Comment. Americn Economic Review, 77, (997). Buyer Groups. Interntionl Journl of Industril Orgniztion, 5, McAfee, R. P., nd Schwrtz, M. (994). Opportunism in Multilterl Verticl Contrcting: Nondiscrimintion, Exclusivity, nd Uniformity. Americn Economic Review, 84, 0-30 Mills, D. E. (03). Counterviling Power nd Chin Stores. Review of Industril Orgniztion, 4, 8-95 Noll, R. G. (005). Buyer Power nd Economic Policy. Antitrust Lw Journl, 7, O Brien, D. P. nd Shffer, G. (997). Nonliner Supply Contrcts, Exclusive Deling, nd Equilibrium Mrket Foreclosure. Journl of Economics & Mngement Strtegy, 6, Rsmusen, E. B., Rmseyer, J. M. nd Wiley, Jr., J. S. (99), Nked Exclusion. Americn Economic Review, 8, Rubenstein, A. (98). Perfect Equilibrium in Brgining Model. Econometric, 50, Segl, I. R. nd Whinston, M. D. (000). Nked Exclusion: Comment. Americn Economic Review, 90,

36 Stole, L. A. nd Zwiebel, J. (996) Orgniztioinl Design nd Technology Choice under Intrfirm Brgining. Americn Economic Review, 86, 95- Steiner, R. L. (985). The Nture of Verticl Restrints. Antitrust Bulletin, 30, Steuer, R. M. (000). Customer-Instigted Exclusive Deling. Antitrust Lw Journl, 68, 39-5 U.K. Competition Commission. (006). Reserch on Suppliers to the UK Grocery Mrket. Accessible t von Ungen-Sternberg, T. (996). Counterviling Power Revisited. Interntionl Journl of Industril Orgniztion, 4, Whinston, M. D. (006). Lectures on Antitrust Economics. (Cmbridge: MIT Press) Wright, J. D. (03). Simple but Wrong or Complex but More Accurte? The Cse for n Exclusive Deling-Bsed Approch to Evluting Loylty Discounts. Accessible t Zenger, H. (00). When Does Exclusive Deling Intensify Competition for Distribution? Comment on Klein nd Murphy. Antitrust Lw Journl 77, 05-36