Location costs, product quality, and implicit franchise contracts Haucap, Justus; Wey, Christian; Barmbold, Jens

Transcription

1 Location costs, roduct quality, and imlicit franchise contracts auca, Justus; Wey, Christian; Barmbold, Jens Veröffentlichungsversion / Published Version Arbeitsaier / working aer Zur Verfügung gestellt in Kooeration mit / rovided in cooeration with: SSG Sozialwissenschaften, USB Köln Emfohlene Zitierung / Suggested Citation: auca, Justus ; Wey, Christian ; Barmbold, Jens ; Wissenschaftszentrum Berlin für Sozialforschung ggmb (Ed.): Location costs, roduct quality, and imlicit franchise contracts. Berlin, 1998 (Discussion Paers / Wissenschaftszentrum Berlin für Sozialforschung, Forschungsschwerunkt Marktrozeß und Unternehmensentwicklung, Abteilung Wettbewerbsfähigkeit und industrieller Wandel 98-8). URN: htt://nbnresolving.de/urn:nbn:de:0168-ssoar Nutzungsbedingungen: Dieser Text wird unter einer Deosit-Lizenz (Keine Weiterverbreitung - keine Bearbeitung) zur Verfügung gestellt. Gewährt wird ein nicht exklusives, nicht übertragbares, ersönliches und beschränktes Recht auf Nutzung dieses Dokuments. Dieses Dokument ist ausschließlich für den ersönlichen, nicht-kommerziellen Gebrauch bestimmt. Auf sämtlichen Koien dieses Dokuments müssen alle Urheberrechtshinweise und sonstigen inweise auf gesetzlichen Schutz beibehalten werden. Sie dürfen dieses Dokument nicht in irgendeiner Weise abändern, noch dürfen Sie dieses Dokument für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, aufführen, vertreiben oder anderweitig nutzen. Mit der Verwendung dieses Dokuments erkennen Sie die Nutzungsbedingungen an. Terms of use: This document is made available under Deosit Licence (No Redistribution - no modifications). We grant a non-exclusive, nontransferable, individual and limited right to using this document. This document is solely intended for your ersonal, noncommercial use. All of the coies of this documents must retain all coyright information and other information regarding legal rotection. You are not allowed to alter this document in any way, to coy it for ublic or commercial uroses, to exhibit the document in ublic, to erform, distribute or otherwise use the document in ublic. By using this articular document, you accet the above-stated conditions of use.

2 WISSENSCAFTSZENTRUM BERLIN FÜR SOZIALFORSCUNG discussion aers SOCIAL SCIENCE RESEARC CENTER BERLIN FS IV 98-8 Location Costs, Product Quality, and Imlicit Franchise Contracts Justus auca* Christian Wey** Jens Barmbold* * University of Saarland ** Wissenschaftszentrum Berlin für Sozialforschung Setember 1998 ISSN Nr Forschungsschwerunkt Marktrozeß und Unternehmensentwicklung Research Area Market Processes and Cororate Develoment

3 Zitierweise/Citation: Justus auca, Christian Wey, Jens Barmbold, Location Costs, Product Quality, and Imlicit Franchise Contracts, Discussion Paer FS IV 98-8, Wissenschaftszentrum Berlin, Wissenschaftszentrum Berlin für Sozialforschung ggmb, Reichietschufer 50, Berlin, Tel. (030)

4 ABSTRACT Location Costs, Product Quality, and Imlicit Franchise Contracts by Justus auca, Christian Wey, Jens Barmbold In the literature on international trade, very little attention has been given to informational asymmetries between firms and consumers with resect to roduct quality. The few economic models that analyze the question of how asymmetric information about roduct quality might affect trade flows treat roduct quality as exogenous. In contrast, our model takes roduct quality as an endogenous variable, i.e. firms can choose the quality they wish to roduce. In this case, location costs can signal roduct quality under certain conditions and thereby affect international trade flows. More secifically, intra-industry trade in vertical differentiated exerience goods can be determined by information asymmetries about roduct quality. We would like to thank Andrzej Cieœlik, James Dunlevy, Jess Gasar, Michel Ghertman, Eckhard Janeba, Daniel Klein, Bertrand Quélin, Stehan Panther, Rudolf Richter, Dieter Schmidtchen, Ilya Segal, Udo Schmidt-Mohr, Deeak Somaya, Willy Sanjers, Mirja Weiss, and seminar articiants at UC Berkeley, the 1997 Meeting of the Euroean Economic Association in Toulouse, the 1997 Meeting of the Verein für Socialolitik in Berne, and the 1998 Meeting of the American Economic Association in Chicago for useful discussion and comments on this work. This aer was almost comletely rewritten while Justus auca was a visiting scholar at the Institute of Management, Innovation, and Organization (IMIO) at the aas School of Business, University of California, Berkeley, the hositality of which is most gratefully acknowledged. Moreover, Justus auca is grateful for financial suort from the Deutsche Forschungsgemeinschaft (DFG) under grant Ri Christian Wey acknowledges financial suort from the Rainer orstmann Fellowshi at Indiana University, Bloomington. This research was also financially suorted by the BMW AG.

5 ZUSAMMENFASSUNG Standortkosten, Produktqualität und imlizite Franchiseverträge In der internationalen andelstheorie ist der Tatsache, daß die Konsumenten in vielen Fällen nur unvollständige Information über die Produktqualität besitzen, kaum Rechnung getragen worden. Die wenigen Arbeiten, die die Wirkungen asymmetrischer Information zwischen erstellern und Konsumenten auf den internationalen andel untersuchen, behandeln die Produktqualität als exogene Variable. Im Gegensatz hierzu wird in diesem Beitrag Produktqualität als eine endogene Variable angesehen, die von den Unternehmen gewählt werden kann. Es wird argumentiert, daß in diesem Fall die Standortkosten zur Signalisierung der Produktqualität eines Unternehmens herangezogen werden können. Es zeigt sich, daß unter bestimmten Bedingungen die geeignete Wahl von standortsezifischen variable Kosten in Verbindung mit Standortaustrittskosten zu einem Trenngleichgewicht führen kann, in dem an einem relativ teuren Standort ochqualitätsrodukte und an einem relativ billigen Standort Niedrigqualitätsrodukte hergestellt werden. Aus diesen Überlegungen folgt, daß intra-industrieller andel in vertikal differenzierten Erfahrungsgütern durch Informationsasymmetrien über die Produktqualität und internationalen Unterschieden in den sezifischen Standortkosten erklärt werden kann. Aufgrund der Anreizwirkungen der Standortkostenstrukur auf die Wahl der Produktqualität wird das Verhältnis zwischen Landesregierung und Unternehmen als ein imlizites Franchiseverhältnis interretiert.

6 1. Introduction Traditional international trade theory redicts that a country s trading attern is mainly determined by two factors: (1) the comosition of its factor endowments, and (2) the intensity with which factors are required (see Jones and Neary, 1984). While government interventions such as tariffs, quotas, subsidies, etc. have also been argued to affect the volume and the structure of international trade, traditional models usually have assumed markets to be erfect. In shar contrast, the literature on strategic trade olicy as initiated by Brander and Sencer (1985) and reviewed by Brander (1995) has started to analyze the effects of trade olicy under conditions of imerfect cometition. As these models demonstrate, government intervention may be beneficial for a country in some situations. What is rather surrising in this context is the fact that otential effects of asymmetric information have only been exlored to a very limited extent - although we know at least since the ath breaking work of Stigler (1961) and Akerlof (1970) that markets with even small informational asymmetries are qualitatively different from markets with symmetric information. The few models that deal with trade olicy under asymmetric information almost exclusively focus on informational asymmetries either between firms and the government (Brainard and Martimort, 1997) or between domestic firms and foreign rivals (Collie and viid, 1993). Common to these models is the idea that government intervention may be beneficial for a country as it can make an outut exansion of the home firm credible. Moreover, as Qiu (1994) has demonstrated, if neither the domestic government nor a foreign rival can observe the firm s costs, the government can lay a second role: By offering a searation inducing menu firms might 1

7 eventually be enabled to signal their costs to a foreign rival. Almost no attention has been given to informational asymmetries between firms and consumers and the question of how these asymmetries may affect international trade. Particular excetions are the two models of W. Mayer (1984) and Bagwell and Staiger (1989) both of which discuss a country s otimal trade olicy when (foreign) consumers are uninformed about domestic roduct quality. As these models do, we will consider the case of an exerience good the quality of which is unknown to consumers. 1 owever, our model differs in imortant resects. First, W. Mayer (1984) and Bagwell and Staiger (1989) take roduct quality as exogenously given; once consumers know a roduct s quality any credibility roblem is resolved. In contrast, we will treat quality as a choice variable. That means, there is a moral hazard or commitment roblem connected with roduct quality. Second, W. Mayer (1984) assumes that the home country s government is informed about a roduct s quality. In contrast, we will consider the case in which even the government is uninformed about the firm s tye. The government is assumed not to know more than any consumer. Similar to Bagwell and Staiger (1989) and Qiu (1994) we will assume that neither the domestic government nor consumers know firm s roduction costs, although they know the distribution from which the tye is drawn. Basically, our model extents the recent work by auca, Wey & Barmbold (1997) according to which rational consumers can use the Made-in label as a signal for 1 The term exerience good was coined by Nelson (1970). According to his definition a roduct is considered an exerience good if its quality is not evident on insection and costly to determine before urchase. In contrast, goods the quality of which can be easily determined on insection have been labeled search goods (see Nelson, 1970). 2

8 roduct quality if countries differ in location secific costs and relocation is costly. In this case, roduction at a high-cost location can credibly signal high-quality roduction while roduction at a low-cost site indicates low roduct quality. owever, while auca, Wey & Barmbold (1997) take rices as given and do not account for differences in roduction costs, our analysis will also derive a monoolist s otimal ricing strategy and allow for differences in roduction costs. Moreover, the focus of auca, Wey & Barmbold (1997) is on the legal rotection of Made-in labels. In contrast, this aer will concentrate on otential imlications for international trade and FDI. Our model is also related to work by Chiang and Masson (1988) and Chisik (1998) who show that consumers may use country-of-origin labels in situations of adverse selection. owever, these models differ from ours in some imortant resects. Chiang and Masson (1988) analyze a case in which country-of-origin effects are the result of consumers statistically discriminating between goods on basis of the roducts country of origin. In this model, consumers only know the average roduct quality roduced by firms in a certain country. Accordingly, consumers form their exectation about roduct quality based on the roduct s lace of roduction. In contrast, our model rovides a signaling exlanation for country-of-origin effects. Similarly, Chisik (1998) analyzes country-oforigin effects in a signaling context. Nevertheless, his model is based on statistical discrimination effects, and country-of-origin labels are only used as signals for roduct quality when the signaling rocessing is noisy. In contrast, we consider a signaling rocess without noise. Furthermore, in Chisik s model high-quality roducers have lower signaling costs than low-quality firms. In our model, every firm faces the same signaling costs, but only some firms are able to bear these costs due to their low costs of high- 3

9 quality roduction. Finally, in Chisik (1998) firms have to make long-run quality choices so they are committed to their quality level once its is chosen. In contrast, we allow firms to choose quality in every single eriod of the game. The rest of the aer is now organized as follows. The next section will introduce and analyze the model. Then, section 3 will shortly relate the model to emirical findings in the international marketing literature, and section 4 will briefly discuss otential imlications for international trade, location choice (or foreign direct investment) and fiscal cometition. Section 5 offers concluding remarks. 2. The Model 2.1 Product Quality as a Problem of Moral azard Let us consider a firm that has develoed a new quality level, q, of some existing roduct category, and let us assume that it is the sole roducer of the new quality level. owever, while we assume the firm to have a monooly osition for the new quality level, let there also be a standard roduct of some lower quality, q L, the market for which is erfectly cometitive. 2 That means, the good under consideration can be roduced at two different quality levels, q L and q, with q L < q. For simlicity let q denote the difference between the two quality levels, i.e. q q - q L. Let us furthermore assume that the marginal cost of low-quality roduction is zero, while the marginal cost of high-quality roduction is given by some constant c >0. Since the market for the standard, low-quality roduct is assumed to be erfectly cometitive the 2 ence the model can also be interreted as a new roduct monooly in which both consumers and roducers have a given outside otion. 4

10 rice is given by L = c L =0. Consumers are assumed to have unit demand, i.e. consumers do not buy more than one unit of the roduct. Furthermore, we assume a continuum of consumers which are heterogenous in their taste for quality. More secifically, let consumers utility function take the form u( q )= Θ q - for j=l,, where denotes the rice aid. The j j arameter Θ reflects the variation in quality taste and is assumed to be uniformly distributed between 0 and 1. If the roduct under consideration is a search good, for which roduct quality is erfectly observable before urchase, a monoolist roducing at a quality level of q will face an inverse demand function of x( )=1-( / q), and the monooly rice for the highquality good is given by: = 1 2 ( q+c ) (1) M Assuming that there are no fixed costs, the monoolist s rofits are Π M 2 ( q ) -2 q c + c = 4 q 2 (2) Next let us consider the case of an exerience good in which roduct quality is not searchable ex ante, but can only be observed after urchase. That means, consumers only learn the roduct s quality after they have bought the good. Let us furthermore assume 5

11 that quality is endogenous: Product quality is not determined once and for all by some technology chosen, but the monoolist can choose between the two quality levels in every single eriod. In this case, the roducer faces an incentive to cheat consumers and offer low quality while charging the rice for a high quality good. ence, the market may ossibly fail as Akerlof (1970) has shown in his seminal aer. owever, as is well known from standard game theory (see, e.g., Eichberger, 1993), the roblem of market failure due to asymmetric information may be overcome if the game is reeated infinitely often and discount rates are sufficiently close to one: If consumers are willing to ay a rice remium for high-quality roducts and, at the same time, adot a boycott strategy in case of low-quality roduction the roducer s romise to deliver high-quality might become self-enforcing (Telser, 1980). Let us, for examle, consider the case of a ure grim strategy according to which consumers are willing to ay as long as the monoolist has never delivered low quality, but are only willing to offer L if the roducer has ever delivered low quality. In an infinite horizon model, the monoolist will refrain from roducing low quality if the following incentive comatibility constraint (IC) is met where i denotes the discount rate: - c i (3) 1+i The left hand side of this condition reresents the resent value of the rofit made er unit if the roducer does not cheat and honestly roduces high quality. The right hand side gives us the discounted hold-u rofit the roducer makes (also er unit) for one single time if he breaks his romise and delivers low quality while charging the rice for a 6

12 high-quality good. Since consumers will boycott him after a one time deviation, he will be only able to charge L in any eriod following a hold-u so that there are no further rofits to be made since L = c L. If now the resent value of rofits from honest high-quality roduction as given by the LS of condition (3) iis greater or equal to the hold-u rofit of a one time deviation as given by the RS, high-quality roduction is self-enforcing for the monoolist. That means, it is in the monoolist s own self-interest not to break his romise to deliver high quality. While the rice remium serves as a carrot, the boycott mechanism can be regarded as the stick, the combination of which makes the romise self-enforcing. For reasons of lasticity, let us rewrite the IC (3) in terms of the quality-assuring rice as: c (1+ i) (3a) If the rice charged by the monoolist is lower than this quality guaranteeing rice, the IC does not hold and the monoolist s otimal strategy is to offer low quality. owever, since ayoffs are assumed to be common knowledge, consumers will not believe in the romise of high-quality roduction for rices that are lower than the quality-assuring rice. ence, consumer demand for the new quality level can now be exressed as: x ( )= 1- for c( 1+i) q 0 for < c (1+ i) (4) 7

13 Given consumers strategy the roducer of a high-quality exerience good now faces the following constrained maximization roblem: Max ( - c )(1 - q ) s.t. c (1+ i) Solving for otimal rices gives us the monoolist s otimal ricing scheme in terms of c : c (1+ i) for c M ( c )= 1 2 ( q+c ) for c q q 1+2i 1+2i (5) Accordingly, the inverse demand function can be exressed as a function of the monooly rice M, and therefore also as a function of c. In this case, consumer demand is given by: x ( c )= 1 2 (1 - c q ) for q c (1+ i) 1- q 0 for for 1+2i q q c 1+i 1+2i q c 1+2i c 0 (6) ence, the roblem of market failure can be overcome if the costs of roducing high 8

14 quality are sufficiently low, or more exactly, if c q/(1+i). owever, if the costs of high-quality roduction exceed this threshold value, the market will fail to come into existence. In this case, the quality-assuring rice exceeds consumers willingness to ay so that the demand side breaks down. Furthermore, for a rice lower than the qualityassuring rice consumers will not buy an alleged high-quality roduct since they know that the roducer s incentive to be honest is lower than his incentive to cheat. 2.2 Product Quality as a Problem of Moral azard and Adverse Selection U to this oint we have considered a ure moral hazard roblem; ayoffs have been assumed to be common knowledge. owever, let us now assume that the cost of highquality roduction, c, is not directly observable for consumers. While consumers can observe rices, lack of knowledge about the costs of high-quality roduction leaves them uncertain about the monoolist s ayoffs, and therefore, his incentives to honestly roduce high-quality. For the sake of simlicity let us assume that there are two ossible tyes of monoolists, B and G, which are characterized by their costs of high-quality roduction, denoted as c B and c G, resectively. Let us also make the following two assumtions, where π denotes the robability of a monoolist being of tye G: (i) B G c q 1+i > c π q and (ii) c G >. π +i Due to assumtion (i) the quality-assuring rice that is incentive comatible for a tye-b 9

15 roducer exceeds consumers maximum willingness to ay as is shown by the demand function (6). ence, a tye-b roducer will not find it otimal to honestly roduce high quality since his costs are too high. For a tye-g roducer costs are sufficiently low so that honest high-quality roduction is self-enforcing. owever, since the qualityguaranteeing rice for a roducer of tye G exceeds zero, a roducer of tye B will now find it rofitable to retend to be a tye-g roducer and earn the hold-u rofit for one time rather than to reveal his tye and earn zero rofits in the low-quality segment of the market. Assumtion (ii) now imlies that the rior robability distribution over the two ossible G B tyes, ( π,c ;(1- π ),c ), induces sufficiently essimistic beliefs so that not a single consumer refers testing a new roduct at a quality-assuring rice of over buying a low-quality good at a rice of L =0. Put differently, there is not a single rice that would make consumers try the new roduct and at the same time rovide roducers with sufficiently high incentives to honestly deliver high quality. That means, consumers are sufficiently essimistic about a new roduct s quality so that rice cannot be used as a signal of roduct quality and demand breaks down. 3 ence, under incomlete information about the firm s roduction costs the roblem of market failure may return as we imly by assumtion (ii). 4 3 This result is exlained in Aendix 1. 4 To overcome this roblem various mechanisms have been suggested in the literature. Most rominently, burning money in form of introductory advertising (Klein and Leffler, 1981; Rogerson, 1987) or rice rebates (Shairo, 1983) have been argued to be a credible signal for roduct quality. owever, introductory advertising camaigns and rice rebates require large amounts of caital which might often not be available to new firms. If firms are not able to borrow against future rofits, the signaling mechanisms suggested are no longer availbale to firms. Moreover, we rule out warranties as a quality assuring rovision. First of all, warranties might not be available due to roblems of incomlete contracting, i.e. it might be too difficult to 10

16 2.3 Location Costs as Searation-inducing Means Now suose that there are two countries, E and W, which differ with regard to the location costs they imose on firms. More secifically, let us characterize the two countries by the differences in country secific fixed costs which are eriodically aid. The country secific fixed costs may take the form of high taxes or wages if labor is a fixed factor of roduction or if it is difficult to lay off ersonnel. Alternatively, one might also think of a situation in which firms have to recommit to a fixed caacity level which again induces a certain level of country secific costs er eriod. For the sake of simlicity let us assume that the low-cost country, E, is characterized by zero fixed costs er eriod 5 while in the high-cost country, W, firms have to ay country secific fixed costs of w er eriod. Moreover, we assume that market exit is costly. These market exit costs may result from comulsory layoff lans or from secific investments associated with lant location. That means, industrial locations are subject to site or country secificity. Let us denote these exit costs by R. Furthermore, assume that k marginal costs are deendent on the firm s location, S, where c j,s stands for the constant marginal costs to roduce quality level j with j {L,} at location S with S {E,W} for a firm of tye k with k {B,G}. Let us assume that conditions (i) and k (ii) hold for both countries, E and W. owever, note that the figure c j,s does not secify all ossible contingencies. Secondly, even if contracts can be contingent on any state of the world, warranties might be costly to enforce for consumers, if courts do not work costlessly and states of the world are difficult to verify. And finally, firms might be reluctant to give warranties if there are moral hazard roblems on the consumer side. Therefore, we want to focus on another mechanism that can hel consumers distinguishing between different tyes of roducers. It is also worth noting that Klein and Leffler (1981), Shairo (1983) and Rogerson (1987) do not focus on innovating firms. In addition, brand name investments and introductory rice rebates also have the major urose to dissiate any rofit in these models as cometitive markets are not comatible with ositive rofits. 5 This assumtion can be viewed as analogous to the standard laissez-faire assumtion made in the literature on strategic trade olicy (see, e.g., Qiu 1994). 11

17 include location costs. ence, the two locations can be fully described by the vectors k (0,c j,e ) k and (w,c j,w ). To kee the model as simle as ossible we take the countries location costs, 0 and w, as well as any exit costs, R, as given. The rocess leading to the cost figures is not analyzed. As a first ste let us also take firm locations as given, i.e. firms cannot choose their location of roduction. The case of caital mobility and location choice is analyzed in a second ste. The timing of our model is as follows: First, nature determines an innovator s location, S, and tye, k. Then, the innovating firm learns its location and tye and decides about market entry. After buyers have observed the innovator s location, S, the firm announces a rice for high-quality roducts,. Now buyers can either accet the rice of or reject it and offer L =0 instead. 6 This basically imlies that consumers comete for the roduct in Bertrand fashion as it is the case in simle highest-bid auctions: The firm sells to the highest bidders. Once the buyers have made their rice offers, the firm decides which quality to roduce, q L or q, and finally, the firm decides about market exit. If the firm does not exit, the next eriod starts with the firm s rice announcement. Let us now denote consumers beliefs as β(k S h t ) where h t is the history of the game. The history of the game h t is the sequence of tules ((,Q ),...,( 1 1 t-1,q t-1 )) that describes the moves of all layers in eriods 1,2,...,t -1 in terms of the rice,, acceted/offered by consumers and the quality, Q, roduced with Q {q,q }. L 6 That means, that the rice announced by the firm is basically a take-it-or-leave-it offer for buyers. By this assumtion, we rule out comlexities raised by bargaining issues. 12

18 Furthermore, all layers are assumed to have erfect memory. That means, consumers beliefs about the firm s tye, k, are conditional uon the location of the innovating firm and all actions observed in revious rounds of the game. While the seller recisely knows his ayoffs, buyers are ignorant about the seller s tye, i.e. his costs, and therefore also about the seller s ayoffs. For reasons of simlicity let now,s denote the otimal rice for a tye-g monoolist as given by the otimal ricing scheme (5). Then, consider the following air of strategies where σ S and σ B denote the strategy of sellers and buyers, resectively. σ S : If the location of roduction is W and nature has chosen G, always announce,w and roduce high quality for buyers acceting the rice. If, however, the location of roduction is W while nature has chosen B, do not enter the market at all. In all other cases, finally, announce L =0 and roduce low quality. σ B : If a seller is located at the high-cost country, buy the roduct for,w if the seller has never sold low quality for,w and if Θi,W / q. In any other case, offer L =0. If we now furthermore assume that roducers have a lexicograhic reference for roduction over non-roduction, we can state Proosition 1: The strategies σ S and σ B and the beliefs 13

19 ,W,t-i β(k = G S = W (,q ) ht for all i = 1,...,t -1)= 1 and,w,t-i L,t-i β(k = B S = E (,q ) ht for any i = 1,...,t -1)=1-π L,t-i form a Perfect Bayesian Equilibrium (PBE) in ure strategies if the following conditions hold: w i 1+i (1 -,W,W q ) (7) R w (8) i,w w ( - c )(1 - G q ),W,W (9) The roof is contained in Aendix 2. The intuition behind this roosition is as follows: Only a firm that can roduce highquality roducts at low costs can afford aying high taxes or wages. ence, only monoolists of tye G can survive at a high-cost location. Furthermore, for a tye-g monoolist rofits made by honestly selling high quality are at least as large as the rofits he makes by cheating consumers and delivering low quality while charging,w. That means, given consumers strategy a tye-g monoolist will never find it attractive to deviate from its strategy to be honest. For a tye-b firm that is located at a high-cost country, however, it is otimal not to enter the market at all since it does not make enough rofits to ay the high country secific costs. ence, on entering the market the 14

20 firm would make losses. Since consumers know this by assumtion they can conclude that for every active firm at a high-cost location the romise to deliver high-quality roducts is self-enforcing. In contrast, a tye-b monoolist located in a low-cost country will always find it rofitable to cheat consumers. While for a tye-g monoolist located at a low-cost country the romise to deliver high quality is not less self-enforcing than for one located at a high-cost country, the roblem is that consumers are unable to differentiate between tye-g and tye-b roducers located in a low-cost country ex ante. Since consumers beliefs are sufficiently essimistic, they will refer not to believe in anybody located in a low-cost country. ence, in equilibrium active roducers in the high-cost country will honestly deliver high quality while roducers in a low-cost country will honestly roduce low-quality goods. At a high-cost location the decision to enter the market serves as a signal for high-quality roduction. While in the low-cost country both tyes of firms will enter and roduce low quality, in the high-cost country only tye-g monoolist starts to roduce and deliver high-quality goods. Moreover, if we exlicitly assume that consumers do not only form beliefs about cost tyes, but directly about quality, we can reformulate Proosition 1 to obtain a Searating Perfect Bayesian Equilibrium. Suose consumer beliefs are given by b(q S h t ), i.e. consumer beliefs are not only contingent on location and the game s history, but the rice a consumer offers as well. Then, we can state Proosition 2: The strategies σ S and σ B and the beliefs b(q = q = S =W (,q ) ht for all i = 1,...,t -1)= 1,W 15,W,t-i and b(q = q S = E (,q ) ht for any i = 1,...,t -1)= 1 L,W,W,t-i L,t-i L,t-i

21 form a Searating Perfect Bayesian Equilibrium (PBE) in ure strategies if conditions (7), (8) and (9) hold. Finally, let us analyze the situation where caital is mobile and firms can choose their location of roduction before making the market entry decision. As location becomes a choice variable for firms now, we reformulate the sellers strategy as follows: _ σ S : If nature has chosen G, always choose W as the location of roduction, enter the market, announce,w, and roduce high quality for buyers acceting the rice. If, however, the nature has chosen B, choose E as the location of roduction, enter the market, announce L =0 and roduce low quality. We can now state Proosition 3: The strategies σ S _ and σ B and the beliefs,w,t-i β(k = G S = W (,q ) ht for all i = 1,...,t -1)= 1 and,w,t-i L,t-i β(k = B S = E (,q ) ht for any i = 1,...,t -1)= 1 L,t-i form a Perfect Bayesian Equilibrium (PBE) in ure strategies if conditions (7), (8) and (9) hold. At this oint it should be noted, however, that as in most signaling models multile equilibria may exist. Deending on consumers belief functions there may be ooling equilibria as well as other searating equilibria. owever, our model shows that the location of roduction can be a credible signal of roduct quality. Furthermore, the searating equilibrium ayoff-dominates the market failure equilibrium. 16

22 3. A short digression into the marketing literature Indeed, emirical evidence from the marketing literature shows that a roduct s country of origin is a cue used extraordinarily often by consumers to judge roduct quality ex ante. Since the 1960s several marketing studies have been conducted, most of which found that the location of roduction has a significant influence on buyers roduct evaluation. The imact a roduct s origin has on demand or consumer evaluation is commonly referred to as the country-of-origin effect. As Bilkey (1993,. xix) has ut it, country-of-origin effects reflect buyers oinions regarding the relative qualities of goods and services roduced in various countries. According to Tan and Farley (1987) these country-of-origin effects reresent the most researched international asect of consumer behavior. The extensive body of emirical work on country-of-origin effects has been reviewed by Bilkey and Nes (1982), Özsomer and Cavusgil (1991) and, more recently, Peterson and Jolibert (1995). In general, the studies have found that consumers have different ercetions about roducts made in different countries. The most salient finding is that consumers use a roduct s location of roduction to evaluate its quality ex ante. This is also reflected in consumers willingness to ay. To give an examle, in an emirical marketing study Johannson and Nebenzahl (1986) found that consumers in their samle were willing to ay 14 er cent more for a Buick built in Germany comared to one manufactured in the US while they were willing to ay 16.4 er cent less for a Buick made in Mexico (comared to the US). Similar findings were obtained for other cars as well as VCRs and microwaves (see Nebenzahl and Jaffe, 1993). 17

23 4. Possible Imlications for International Trade and Location Choice Concerning international trade flows our model would redict that high-cost countries exort high-quality roducts while imorting search (or heavily standardized) goods as well as low-quality exerience goods. The oosite should then be true for low-cost countries. 7 ence, our model can give a new reason for intra-industry trade as trade flows are determined by informational roduct differentiation. Moreover, the few models that discuss informational asymmetries between firms and consumers and focus on the question of how these asymmetries may affect international trade, i.e. W. Mayer (1984) and Bagwell and Staiger (1989), treat roduct quality as exogenous. Therefore, the olicy conclusions one might derive from these models are quite different from ours. In W. Mayer s model the otimal trade olicy consists in an exort subsidy that induces foreign consumers to try the roduct. Similarly, Bagwell and Staiger (1989) argue that a quality contingent tax/subsidy rogram might be welfare enhancing on the national level. In contrast, almost the oosite is true for our model: Not subsidies, but high country secific costs such as taxes can indirectly signal roduct quality. If caital is mobile and lant locations are endogenous the model might also rovide a new rationale for firms location choice. Firms that have low costs for roviding highquality roducts might consciously choose a high-cost country for roduction instead of 7 Interestingly enough, Vernon (1966) has made a very similar oint. As in our model, Vernon addresses the question of where new and where more standardized roducts are roduced. is finding is that new roducts are usually roduced in advanced countries, while more standardized goods are usually roduced in less develoed countries where labor is chea. While Vernon bases his finding on a comletely different exlanation, it is interesting to notice that his results could also be exlained by our model assuming that new roducts rather have exerience qualities while standardized roducts can be regarded as search goods. 18

24 a low-cost location - even if there is no difference in the level of ublic goods or infrastructure rovided. To be more recisely, location at a high-cost country might serve as a signal of roduct quality exactly because of the fact that the high-cost country does not offer any direct roductivity benefits. That means, firms may settle at high-cost locations urely for signaling considerations. A very similar oint has been raised by Bagwell and Staiger (1988) who consider a market entry game similar to the models of Milgrom and Roberts (1982) and Bagwell and Ramey (1988). According to their model, multinational firms might locate lants in high-cost countries to signal their roduction costs to rival firms. To ut it differently, in the model of Bagwell and Staiger (1988) location choice can be used strategically to change rivals ercetions about the firm s roduction costs. owever, the role the location of roduction may have as a signal for roduct quality is not exlored in their model. Country-of-origin labels may be esecially relevant as a signaling device if firms cannot sink enormous amounts of resources to gain credibility as has been suggested by Klein and Leffler (1981) and Shairo (1983). Introductory advertising camaigns or rice rebates require large amounts of caital which might often not be available to new firms. 8 ence, in the resence of caital market constraints firms might have to use other signals and might rather use the country-of-origin as a signaling device since it does not require tremendous amounts of start-u caital to be sunken to gain credibility. Instead the firm 8 Exlanations for the resence of caital constraints are beyond the scoe of this aer. owever, caital constraints have been reorted to exist - esecially for new firms (see, e.g., C. Mayer, 1988). 19

25 leases the country s brand name, Made in Country XY, and eriodically ays a franchise royalty in form of high taxes or wages. In this case, location choice can be interreted as the conclusion of an imlicit franchise contract between the firm and the host country or its reresentatives, i.e. the government. According to this interretation host countries (or governments) act as franchisors that rent out their brand name (e.g., Made in Germany ) to firms. Country secific exenditures can now be seen as franchise royalties roducers have to ay for using the brand name. In contrast to ordinary franchise contracts, however, there are no initial caital requirements, but a nondissiative exit cost bond which serves as a termination enalty. Quite interestingly, Bagwell and Staiger (1988) reort that according to the emirical findings of Swedenborg (1979) Swedish multinationals are in fact more likely to invest in countries characterized by comaratively high wage rates. In this context, our model might also offer a new ersective on tax cometition and location tournaments. Lowering taxes, relaxing environmental standards and granting subsidies to new firms might not necessarily attract new firms if it is the higher costs associated with a country that hels the firm signaling its roducts quality. Building uon this rationale location tournaments might not necessarily result in destructive tax cometition and subsidy races or social and environmental duming. Starting from our model one might think of scenarios in which both high-cost and low-cost locations might coexist. While firms in high-tax countries will then rather roduce high-quality exerience goods firms in low-tax countries will roduce search goods and low-quality exerience goods (assumed that all other things are equal and that there is a demand for such goods). In a general equilibrium aroach, a country s otimal tax olicy will be 20

26 deendent on the structure of demand for high- and low-quality exerience goods as well as on the demand for search goods (where signaling devices do not lay a role). Moreover, an additional factor that determines a country s otimal tax olicy consists in the degree of cometition for caital between governments, i.e. the structure of the franchise market. That means, the otimal tax olicy is also deendent of the strategies of other countries. To derive hard conclusions about otimal tax olicy in this context, however, further research is needed. While we have to leave the questions raised unanswered right now, we hoe that they may stimulate further research in this direction. 9 Finally, it should be noted, however, that the analysis is strictly alicable only to new roducts whose quality is unknown before urchase. The model says little about the location choice of multinational enterrises which already have an established brand name. In this case, the analysis of Wernerfelt (1988) might aly according to which a monoolist can use umbrella branding to signal roduct quality. That means, by using the brand name of an established roduct for a new roduct a multiroduct monoolist can ut its entire reutation at stake. Later sales of other roducts that wear the same brand as the new roduct serve as a bond and nondissiative signal of roduct quality. owever, for umbrella branding to work as a signal, one has to assume that the firm already enjoys a reutation which it can ost as a bond. In contrast, our analysis might aly to the case in which the firm does not already have a favorable reutation it can ut 9 The works of Bond and Samuelson (1986) and Raff and Srinivasan (1998) are somewhat related as they exlore how a country s tax olicy might be used to send signals about local roduction costs to foreign investors. owever, they do not exlorehow a firm s location choice might affect consumer beliefs and, thereby, demand. 21

27 at stake. 5. Conclusion This aer has endeavored to offer an additional reason for intra-industry trade. If firms can choose the quality they wish to roduce, and consumers find it difficult to determine roduct quality before urchase, the location of roduction can signal roduct quality under certain conditions, and thereby affect international trade flows. According to the model resented here, intra-industry trade in vertical differentiated exerience goods can be determined by information asymmetries about roduct quality. The model also rovides an exlanation for the emirically observed, but in economic theory widely neglected henomenon that consumers judge a roduct s quality by its location of roduction. According to the model resented here high country secific costs such as high taxes or wages and comulsory exit costs might serve as a natural screening device that enables consumers to differentiate between firms that can roduce high-quality goods at low costs and for which the romise to deliver high quality is selfenforcing and those firms that find it rofitable to hold u consumers. If the Made-in label is used as a signal for roduct quality, trade flows may be determined by countryof-origin effects. According to our theory high-cost countries would now rather exort high-quality exerience goods while imorting low-quality exerience goods and search goods while the oosite should hold for low-cost countries. Furthermore, we have argued that, in contrast to much of the recent trade literature, not a subsidy, but high location costs such as high taxes might hel a firm in its endeavors to 22

28 signal roduct quality. Moreover, if caital is at least mobile to some degree and lant locations endogenous, the model also gives a new rationale for location choice and foreign direct investment: Firms that can roduce high quality at low costs might consciously chose high-cost locations for roduction as a means to signal roduct quality. Location choice can now be seen as analogous to the conclusion of a franchise contract: The country (or its reresentatives) rent out the country s brand name ( Made in Country XY ) for which they eriodically receive a franchise royalty which is usually called tax in this context. Contrary to standard franchise contracts, however, we have emhasized on the role of exit costs which allow to waive initial caital requirements. Proceeding from this hyothesis we have finally suggested that intergovernmental cometition and location tournaments do not necessarily have to result in cut-throat tax cometition, subsidy races and the lowering of social and environmental standards as the standard literature on international ublic finance argues. While we have only exlored some ossible imlications of our model so far, we hoe that the questions raised might stimulate further research into this direction. 23

29 Aendix 1: Exlanation of Assumtion (ii) Assumtion (ii) imlies that not a single consumer is willing to try a new roduct of unknown quality. This result can be easily verified by comaring a consumer s ayoff from trying a new roduct with the certain ayoff of buying a low-quality roduct. Due to assumtion (i) it is otimal to deliver high quality for a tye-g roducer while it is a dominant strategy to cheat for a roducer of tye B. Now given the rior robability G B distribution, ( π,c ;(1- π ),c ), as well as consumers grim strategy as described above, for any articular a consumer s ayoff from trying a new roduct can be exressed as Θq- ΘqL- ΘqL π +(1- π )( + i 1+i i(1+ i) ) Comared to the certain ayoff from buying low quality, Θ q L /i, the condition for trying the new roduct reduces to Θ ( π +i). π q(1+ i) As can be seen from this condition, demand will be zero for ( π q(1+ i)) / ( π+ i). Since we know from condition (3a) that roducers only deliver a high-quality roduct and consumers only buy a roduct of allegedly high quality if G c (1+ i), this can be G reduced to c ( π q)/( π+i) which is, of course, assumtion (ii). 24

30 Aendix 2: Proof of the Proosition First, we must roof that given consumers strategy, σ, and their beliefs, β(k S h ), B t neither an innovator of tye B nor one of tye G can gain by deviation from his secified strategy, σ S, at any stage of the game. Second, we have to show that consumers strategies are otimal given σ S, and finally, consumer beliefs must be consistent with equilibrium strategies and observed actions. Ia) Suose an innovator is located at W and nature has chosen B. As imlied by assumtion (i) and consumers utility function, consumers are not willing to ay tye B s B quality-assuring rice, = c,s (1+ i). That means, the one-time hold-u rofit exceeds tye B s gains from honestly roviding high quality as imlied by condition (3). Since the firm is located at the high-cost country its net rofit on entering the market and cheating consumers is given by,w,w 1+i (1 q ) w i, if the firm roduces low quality and does not exit afterwards, and,w,w 1+i (1 q ) w 1+i R 1+i, if the firm roduces low quality and exits afterwards. The first exression cannot be ositive because of condition (7) which states that w/i exceeds the maximum rofit any roducer could make on a one-time deviation. 25

31 Furthermore, due to condition (8) the maximum rofit from deviating one single time and exiting the market afterwards is zero as well. Thus, the second exression is also nonositive. Therefore, we can conclude that for a tye-b roducer located at the high-cost country it is otimal not to enter the market. Ib) Now let us still assume that a firm is located at the high-cost country, but nature has chosen G. By honestly roducing high quality the tye-g monoolist can earn a rofit,w G stream of ( Π -w)/ i with Π=( - c )(1 - / q), which in this case is ositive,w due to condition (9). From (Ia) we already know that cheating consumers while being located at a high-cost country imlies losses (which is indeendent from the firm s tye). ence, a tye-g monoolist located at W will find it otimal not to deviate, but to honestly roduce high quality.,w Ic) Suose now that an innovator of any tye is located at E. Given consumers strategy and beliefs they will reject any rice other than L =0 due to assumtion (ii). ence, for firms located at the low-cost country it is otimal to roduce low quality. II) Now, we have shown that for innovators of tye B it is only otimal to enter the market and to roduce if they are located at the low-cost country. For innovators of tye G, however, it is otimal to enter the market in any case. Nevertheless, as has been shown under (I) it is only otimal to roduce high quality for the firm if its location is the high-cost country. Obviously, for consumers it is then otimal to accet,w if the seller is located at W, given sellers strategies and Θi,W / q. On the contrary, given the 26

32 G B robability distribution ( π,c ;(1- π ),c ) as secified by assumtion (ii) no consumers will find it otimal to samle a new roduct if the firm is located in the low-cost country as has been exlained in the text. Since both tye-b and tye-g roducers enter the market if they are located in the low-cost country consumers cannot distinguish between them and will otimally refrain from acceting any rice higher than L =0. III) Finally, consumers beliefs are consistent with equilibrium strategies, since every innovator that is located a the high-cost location and has entered the market will be of tye G. ence, the very fact that a firm located at W takes u roduction credibly signals consumers that its romise to roduce high quality is self-enforcing. In contrast, at the low-cost country every firm will enter the market. Therefore, a roortion 1-π of the firms located at the low-cost site will be of tye B, and accordingly, a roortion π of tye G. Thus, consumers beliefs are reinforced by equilibrium outcomes.. 27

33 References Akerlof, G.A., 1970, The Market for Lemons : Quality Uncertainty and the Market Mechanism, Quarterly Journal of Economics 84, Bagwell, K. and G. Ramey, 1988, Advertising and Limit Pricing, RAND Journal of Economics 19, Bagwell, K. and R.W. Staiger, 1988, Private Cost Information and the Multinational Enterrise, National Bureau of Economic Research Working Paer No Bagwell, K. and R.W. Staiger, 1989, The Role of Exort Subsidies When Product Quality is Unknown, Journal of International Economics 27, Bilkey, W.J., 1993, Foreword, in: N. Paadooulos and L. A. eslo, eds., Product- Country Images: Role and Imact in International Marketing (aworth Press, New York) xix-xx. Bilkey, W.J. and E. Nes, 1982, Country-of-Origin Effects on Product Evaluation, Journal of International Business Studies 13, Bond, E.W. and L. Samuelson, 1986, Tax olidays as Signals, American Economic Review 76, Brainard, L. and D. Martimort, 1997, Strategic Trade Policy with Incomletely Informed Policymakers, Journal of International Economics 42, Brander, J., 1995, Strategic Trade Policy, in: G. Grossman and K. Rogoff, eds., andbook of International Economics, Vol. 3, (Elsevier Science, Amsterdam) Brander, J. and B. Sencer, 1985, Exort Subsidies and Market Share Rivalry, Journal of International Economics 18, Chiang, S.-C. and R.T. Masson, 1988, Domestic Industrial Structure and Exort Quality, International Economic Review 29, Chisik, R., 1998, Exort Industry Policy and Reutational Comarative Advantage, mimeo, University of Connecticut, Storrs, CT. Collie, D. and M. viid, 1993, Exort Subsidies as a Signal of Cometitiveness, Scandinavian Journal of Economics 95, Eichberger, J., 1993, Game Theory for Economists (Academic Press, San Diego). auca, J., C. Wey and J. Barmbold, 1997, Location Choice as a Signal for Product Quality: The Economics of Made in Germany, Journal of Institutional and 28