Version classiqueVersion mobile

The Economics of Geographical Indications

Daniela Benavente

Chapter 2 – GI Club Reputation Dynamics (Quality Endogenous to the Firm)

Texte intégral

2.1. Abstract

1This chapter presents a model of club reputation that incorporates the main references in the extant literature as special cases (Shapiro 1982 on individual reputation and Winfree and McCluskey 2005 on collective reputation). Reputation is assumed to be traceable through the GI label and quality is endogenously determined at the firm level, with reputation as the state variable. In contrast with previous research, it is shown that the TRIPS legal construct regulating GIs is potentially compatible with an equilibrium involving a self-fulfilling level of quality (and reputation) that is above the minimum, under the condition that the GI club has a reduced membership of firms. However, the establishment of a minimum level of quality is still the first best policy to improve firm profits. In the same framework, it is demonstrated that under bottom-up firm-driven processes of club formation, in which firm profits are maximized, firm quality and profit levels are higher, and club membership levels are lower, than under top-down State-driven processes, in which club profits are maximized.

2.2. Introduction

« À l’égard du vin, de l’huile, et des autres choses que l’on est dans l’usage de goûter avant d’en faire l’achat, il n’y a point de vente tant que l’acheteur ne les a pas goûtées et agréés. »
Art. 1587 of the French Civil Code of 1804

2The reputation of a firm – or its goodwill – is a valuable asset. But building-up a reputation only makes sense in a world of imperfect information, and with reference to an experience good – i.e. goods for which quality is unobservable until they are consumed. If products’ attributes were perfectly observable prior to purchase, previous production of high-quality items would not enter into consumers’ evaluations of a firm’s product quality, as quality beliefs could be derived from inspection. But when product attributes are difficult to observe prior to purchase, consumers may plausibly use the quality of products produced by the firm in the past as an indicator of present or future quality. Thereby a firm possesses a good reputation if consumers believe its products to be of high quality (Shapiro 1982 and 1983).

3Now, compared to individual reputation, the collective reputation of a group of firms is similar to a common property resource. If there is unconstrained use of a particular GI or appellation of origin, firms might perceive the shadow price of such a use to be zero and decide to free-ride on reputation by “cheating” on quality, thus making short term profits and degrading the rent-value of the GI or appellation. A firm’s decision to produce high-quality items is dynamic since the benefits accrue only in the future. As reputation formation may then be considered a form of signalling activity, a seller cannot command a price associated with high quality until his reputation is established (Winfree and Mc Cluskey 2005).

4Translated into economics, the TRIPS Agreement holds that a GI is used to designate a particular variety of a good of higher-than-generic quality level, for which supply is restricted due to the geographical confinement of production. The TRIPS Agreement does not require that a particular standard be associated with a GI-protected good, it only stipulates the existence of some intrinsic quality. The GI label gives traceability to the variety – through the legal protection of the use of the GI but not at the firm level (either because traceability is not possible or because the brands using the GI are weak). Restricting the use of the GI label does not constitute per se a barrier to entry, since a GI, in its weakest version, is simply a label indicating origin. Nothing prevents firms from making identical copies of the GI-good, so long as the GI label is not used to indicate that such copied products originate in the GI region (TRIPS basic protection), and firms do not claim to produce a “like” product (TRIPS extra protection granted to wines and spirits).

5Correspondingly, protection of a GI is legitimated as a means of preventing the use of a GI: “in a manner which misleads the public as to the geographical origin of the good” (TRIPS Article 22.2.a) – consumer protection – or in a manner “which constitutes an act of unfair competition” (TRIPS Article 22.2.b) – protection of the investment and goodwill of the firms establishing the reputation of the GI.

6In this chapter, a simple model of club reputation is developed, which outlines the role of shared costs in the establishment of the equilibrium quality/reputation level within a club of producers bearing a particular GI. The model involves the dynamic optimization of quality (the choice variable) of an isolated profit-maximizing firm subject to a certain process of club reputation (the state variable). The model is demonstrated to include, as special cases, the two primary authorities in the extant literature, Shapiro (1982) and Winfree and McCluskey (2005), who model individual and collective reputation respectively.

2.3. Literature review: Individual and collective reputation

7Several theoretical papers have addressed the issue of reputation both in the context of the firm and elsewhere. To my knowledge, however, none have attempted to specifically model the reputation of a club, although Winfree and McCluskey (2005) model the collective reputation of a group of firms as a common resource good, and apply their model to a geographical indication, Washington apples. What follows is a brief review of the current literature regarding reputation.

8A finding that is common to this body of work is that the adoption of a minimum standard of quality is the first best policy to sustain a high level of reputation (individual, collective or club). The model proposed in the next section is no exception.

9Shapiro (1982) shows that a profit-maximizing monopolist subject to a state equation for reputation commands a steady-state equilibrium with a mark-up that stems exclusively from being in an imperfect information setting. He also demonstrates that if the monopolist could credibly commit to the level of quality where marginal revenue equals marginal cost (the perfect information equilibrium), both quality and profits would increase in comparison with the steady-state. In a second paper (Shapiro 1983), Shapiro shows that premia for high-quality products are returns on the asset value of reputation (thought of as an initial information cost). In doing so, he uses a dynamic model characterized by perfect competition but imperfect consumer information: the quality of the product is observable only after purchase, but brand reputation is common knowledge. Shapiro then goes on to show that the role of reputation in ensuring higher quality is imperfect at best, and that correspondingly it need not confer market power nor imply a barrier to entry. He considers heterogeneous consumers in a welfare analysis utilizing comparative statics, as responses to shifts in minimum-quality and high-quality premia.

10Winfree and McCluskey (2005) build on Shapiro (1982) to model collective reputation as a common property “extracted” by firms (p. 206). They outline a model of simultaneous dynamic optimization over continuous time. A particular number of risk-neutral, identical firms are assumed to produce a specialty product (experience good), with no entry, no exit, and no firm-level traceability (p. 208). Each firm maximizes profits with respect to quality subject to the state equation for reputation. They show that the steady-state Cournot-Nash equilibrium is a saddle-point, and that as the number of firms rises, the steady-state level of quality is reduced, indicating the existence of a free-rider problem, as returns to quality are diluted whereas costs are not internalized. With the monopoly as a benchmark (quality and reputation both being previously proved by Shapiro to be higher), the policy implication is that it is in the interests of the industry to set a minimum quality standard to eliminate free-riding. This would allow for the achievement of a higher collective reputation, a Pareto improvement compared with the earlier Cournot-Nash equilibrium. Winfree and McCluskey also show that with the group working as a quality cartel, the possibility of trigger strategies exists. In that context, quality wars have lagged effects on reputation (as proved by Tirole 1993) that price wars do not.

11In this line, a recent paper (McQuade et al. 2010) develops a framework wherein both quantity and quality are choice variables. The authors first demonstrate that in the context of a constant per unit cost of production that is strictly convex in quality, a single level of quality is chosen by a firm. In this situation, a unique symmetric and interior Nash equilibrium exists, an important result that was assumed in previous models (and the one proposed here). As the market becomes less concentrated, the equilibrium quality level is found to decline, such that minimum quality standards can improve economic welfare. The real novelty of the paper is its assessment of the occurrence of the Nash equilibrium when the final product is assembled from components manufactured in-house by each Cournot competitor. Minimum quality standards can then be imposed on the final product or, alternatively, on a single component. Similarly, Rouvière and Soubeyran (2008) deal with the issue of entry into an industry where firms share a collective reputation. They show that free entry is not socially optimal and that the imposition of a minimum quality standard is optimal and does not necessarily constitute a barrier to entry.

12Fleckinger (2007) explores the situation in which consumers know the average quality offered by a set of producers, but not the quality of one given product. First, he shows that when the strategic variables are quantity and quality, the equilibrium welfare function is convex in the number of competing firms as a consequence of decreasing quality and increasing quantity. Second, he assesses the potential trade-offs of quantity against quality through a number of regulatory tools, to demonstrate that among one-instrument policies, entry and quantity regulations perform better than price-based regulation.

13Tirole (1993) addresses the issue of reputation from yet another angle. He studies the joint dynamics of individual and collective reputations and derives the existence of stereotypes from history dependence (the key concept), “rather than from a multiplicity of equilibria or from the existence of a common trait as is usually done in the literature” (p. 1). He models group reputation as an aggregate of individual reputations. The reputation of a firm (the principal) results from the actions of its employees (the agents), which are of three types: honest, dishonest, and random. He uses a matching game wherein agents do not, once hired, meet the same principal twice, and decide if they cheat based on their type and payoff structure. After episodes of bad behaviour, either the group becomes stuck in a bad reputation steady-state, or it takes several periods to re-establish trust and reach a high reputation steady-state. A stochastic version of Tirole’s model was recently developed by Levin (2009).

2.4. A model of club reputation

14In this section I develop a simple model of club reputation. The model is a generalization in continuous time of Shapiro (1982) and Winfree and McCluskey (2005), who focus, as discussed, on individual and collective reputation respectively, and includes these models as special cases. Additions to their models are inspired by the previously outlined club goods theory.

2.4.1. Model set-up

  • 20  The assumption of a firm producing one item of a single quality is made without loss of generality (...)

15The model assumes N identical, risk-neutral, profit-maximizing firms, indexed by i. Each firm is assumed to produce a fixed number of items of an experience good, fixed at one for simplicity, of a level of quality q.20 There is a minimum level of quality q0 under which sales are not possible, which might be a legal minimum (technical regulation in place), a level at which quality reductions may be detected by inspection, a warranty level (Shapiro 1983) or a level of quality at which profits are zero.

16Current quality is the control variable for the firm; it is not observed by consumers before purchase but predicted based on reputation. Reputation is the state variable and is assumed to be common knowledge for both consumers and firms. Club reputation evolves with a lag g, which captures the notion that consumers “may only adjust reputation partially in response to observed quality” (Shapiro 1983, p. 671), but that is used here to represent the fact that reputation is built over time, and that as such, club reputation, based on perceptions of average quality, is time-sensitive, implying an adaptive process. Quality q and reputation R are normalized and assumed to be in the same unit. At the steady-state, reputation equals quality q = R.

  • 21  Since the quantity sold by the firm is normalized to one, marginal price equals marginal revenue.

17Production costs are a function of quality, while the inverse demand is a function of reputation (c(q) and p(R) are positive and increasing). The cost function is strictly convex, with marginal costs increasing at a higher pace than marginal revenue (implying p'(R) > 0; c'(q) > 0; c"(q) > 0 and c"(q) > p"(R)|q=R).21 Per period firm profits are assumed to be time-additive. The discount rate is assumed to be fixer at r. Following Shapiro (1982), prices are assumed to be independent of demand (firms are price-takers), although there may be no consumers (and thereby no “active sellers”) at a given price-quality combination. This is analogous to the fact that price is determined solely by costs under perfect competition.

18The model departs from previous models in two broad aspects:

  1. Winfree and McCluskey assume no entry, no exit, and no firm-level traceability. For the model at hand on club reputation, the assumptions, inspired by the TRIPS Agreement are that:
    – The good has traceability at the club level through the GI label (“Geographical indications are indications which identify a good…”)
    – The GI club has limited membership due to the geographical confinement of output (“…as originating in the territory of a Member, or a region or locality in that territory…”)
    – The GI good constitutes a differentiated variety within its broad product category (“…where a given quality, reputation or other characteristic of the good is essentially attributable to its geographical origin.” Article 22.1 of the TRIPS Agreement)

    • 22  It is shown below that the main findings hold for an S-shaped function (at low-quality levels, up (...)

    Following club goods theory, “club-house” costs are assumer, G(R), which are shared by the members of the club, such that each firm pays a fee ƒ = G(R)|N. These costs of establishing, maintaining, and sustaining the GI club (lobbying, marketing, advertising, administrative costs, etc.) are a function of reputation (the higher the reputation, the higher the incentive to have a club-house; alternatively, the higher the investment in the GI, the higher its reputation). G(R) is increasing and strictly concave, since sustaining a reputation is costly, but once the entire system is in place, sustaining a slightly higher quality level at the margin (at the steady-state q = R) involves only minimal adaptations of the facilities. The assumption thereby is that G'(R) > 0 and G"(R) < 0.22

2.4.2. The firm’s maximization problem

19The firm’s problem is to maximize the present value of profits subject to the state equation for reputation and the constraint that profits be non-negative:

20Subject to:


22Where, following Winfree and McCluskey (2005), it is assumed that firms adopt a stationary Markovian strategy (the firm conditions its actions at a certain period only by the state of affairs at that given period), so that qjø(R). The current-value Hamiltonian is:

23Where li is the current-value shadow price of reputation – a co-state variable. Apply the Pontryagin recipe to obtain the isoclines (derivations in Appendix A):

2.4.3. Steady-state equilibrium

24The steady-state equilibrium is found at the intersection of the isoclines Ṙ = 0 q ̇ = 0 and including the symmetry assumption. The solution is:

25Above, p'(R*) is the marginal revenue of the representative firm (because output is set at unity); c'(q*) is the output marginal cost and G'(R*)/N is the club-house marginal cost per firm. The solution involves a price schedule where marginal revenue covers marginal costs at a premium

26Under the assumption that p"(R) < 0 (downward-sloped marginal revenue curve), the model implies a lower equilibrium level of quality/reputation than under collective reputation (in the latter case G = 0).

2.4.4. System stability: Saddle-point path equilibrium

27The Jacobian matrix is:

  • 23  Although a concave marginal revenue function is usually assumed, the model does not rule out a con (...)

28The determinant of the Jacobian matrix at the steady-state provides information about the behaviour near the steady-state. Under the assumptions of the model (c" > 0; G"≥ 0 and c" > p"), the determinant is unambiguously negative, implying eigenvalues of opposite signs and a saddle-point path equilibrium. The stability condition D < 0 is a curvature condition that asserts that the firm’s marginal revenue is either declining p" < 0, or that if it rises, it rises no faster than the “relevant” marginal cost increases (which includes the premium).23

2.5. Quality and membership: Network and dilution effects

29The implicit derivative of quality with respect to number of firms at the equilibrium point is:

30As the number of firms increases, the average quality may or may not decrease, depending on the parameters of the model. At low levels of membership, quality will increase if:

  1. The discount rate r is low. A low discount rate implies a relatively high value of future cash flows, making it unprofitable for firms to cash in early on the extra profits obtained on cheating on quality. This is a firm-level optimization process that implies, in the aggregate, a higher quality level.

  2. The speed of consumer learning g is relatively high. This is an intuitive result: if consumers are quick in adapting their expectations of quality based on consumption, the payoff of increasing quality is not delayed, making it profitable to increase quality.

  3. Marginal costs of production c' are low. The lower the additional cost from marginal increases in quality, the more it pays to increase quality.

  4. Marginal GI-related costs G'are relatively high. High marginal GI-related costs imply high benefits from cost-sharing, something difficult to achieve at low levels of membership.

31Rearranging terms, these conditions reduce to one simple condition: quality will increase with the number of firms as long as the premium over production cost

32is lower than the per-firm GI-related marginal cost

33(otherwise, quality will decrease).

34This suggests a potential scenario in which there is an inflexion point at some level of N, up to which the equilibrium quality increases, implying positive club “network effects”, and after which it decreases, due to negative crowding and “dilution effects”. This finding is crucial, as it departs from most theoretical work on collective reputation. Winfree and McCluskey (2005), for instance, find that as the number of firms increases, the average quality unambiguously decreases due to a free-rider problem, essentially due to the fact that the returns to quality are diluted, while the costs are not. Once shared costs are included, the optimum is affected, the race-to-the-bottom over quality can be prevented as network effects prevail at early stages of the development of the reputation of the GI (reduced number of producers).

2.5.1. The bottom-up, firm-driven approach to optimal membership

35Proposition 1: A GI club with club-house shared costs increasing with reputation will exhibit a steady-state saddle-path point equilibrium with a self-fulfilling level of quality that is a function of the level of membership N. Quality increases with the number of firms as long as the premium over production cost

36 is lower than the per-firm GI-related marginal cost

37, due to positive club network effects. In this case, a level of membership Nf > 1 that maximizes profits at the firm level exists, which does not correspond to the monopolist. In the early stages of the development of a GI good, with a reduced number of firms/output, the “race-to-the-top” over quality is a market outcome under the assumption of club reputation.

38Figure 2.5.a illustrates the equilibrium steady-states as a function of the number of firms. The self-fulfilling level of quality is maximized at a level of membership that is greater than 1 (10 firms in the example). Since that level is itself above the minimum level of quality, firm profits are positive.

Figure 2.5.a: Equilibrium steady-states as a function of the number of firms

Figure 2.5.a: Equilibrium steady-states as a function of the number of firms

39Proposition 2: For a membership N > Nf, the club presents partial rivalry through reduced profits as membership increases, due to crowding or dilution effects. In this model output per firm is set at 1 for simplicity, but there is no restriction imposed on the amount of output that could be sold. Membership refers exclusively to the number of firms in the market.There is an unambiguous race-to-the-bottom over quality as membership increases.

40Figure 2.5.b shows the same Figure with club costs assumed to be zero (G = 0), an assumption that corresponds to Winfree and McCluskey’s results, but with reduced membership due to GI level traceability. However, if membership is kept low artificially, which is something that could happen, for example, due to the geographical confinement of output, the self-fulfilling level of quality at the steady-state, given N, is above q0 (Cf. Proposition 1).

Figure 2.5.b: Equilibrium steady-states without GI club costs

Figure 2.5.b: Equilibrium steady-states without GI club costs

41If the development of GIs “in nature”, prior to any legal protection (i.e. Florida Oranges, Antigua Coffee, Parma Ham, Basmati Rice, and Feta Cheese), is any indication of this mechanism at play, the message of the model is powerful. It implies that as long as the use of the GI name is restricted (through fair competition, laws against passing-off, or the legal protection of the GI), and the GI product is a highly-valued and differentiated variety (as expressed by the inverse demand function), the market equilibrium will automatically command both a higher quality level and a higher industry-level profit, exploiting some of the potential of the GI variety under imperfect information.

42The model also justifies the protection of GIs when GIs are confronted with free-riding by non-legitimate users (on both consumer and goodwill-protection grounds), as it improves quality and profits due to traceability (a signalling effect well-known in branding theory) in a perfect competition / imperfect information setting.

43The policy implications of this result for developing countries are attractive. They do not need to impose cumbersome and expensive structures for the development of GIs, they only need to ensure that the use of the GI label is fair and protected from counterfeits for some of its economic potential to develop as a market outcome ( increase in the quality of the product as membership in the GI-club increases, due to club network effects). This is the level of protection granted under the current TRIPS Agreement (conditional on national protection).

2.5.2. The top-down, State-driven, approach to optimal membership

  • 24  This question is related to the literature on (Old and New) Institutional Economics, which focuses (...)

44Although GI protection in traditional countries (France, Italy) often involves the establishment of quality standards, conformity assessment, and administrative structures, the only clear requirements under TRIPS for a GI to be protected as such are: to be a variety with some intrinsic quality or reputation, and to originate in “a territory (…) or a region or locality in that territory” (TRIPS Article 22.1). The territory is usually formally delimited by the State, following a request by some communal entity or group of producers, generally simply validating a historically defined territory in accordance with tradition.24

45This issue is of particular relevance regarding a GI club for two reasons. First, there is a public good dimension to the GI. Globally, GIs have a strong rural bias, and are attached to goods that embody some sort of cultural heritage, typicity, tradition, and historicity (Reviron 2009, p. 6). Those features usually imply a strong hold of State central and communal institutions, and of civil society in all GI-related matters, with a top-down approach to the protection of GIs. This is usually the approach adopted in developing countries. In continental Europe, however, the protection of GIs has been more often a process driven by GI good producers through professional or inter-professional associations (cf. Italian Consortia), a bottom-up approach, which is then eventually validated by the State.

  • 25  Club goods theory, as a theory of consumption, sheds light on this question. Consider a golf club. (...)

46Second, different club-house institutional arrangements will lead to different levels of optimal membership. To simplify, assume that a “social planner” delimits the territory, thereby indirectly setting N, the club “membership”, and total output. What is the optimal membership? If the club-house is member-driven, the firms will set membership at the level that maximizes profits at the firm level, Nf. If, however, the GI is controlled by the State, the State might want to set a level of membership that maximizes profits at the club level Nc, to fully exploit the potential of the GI and/or to increase its own tax revenues (which in most countries are set at a fixed percentage of profits). At the limit, if the country’s legal regime allows for it, the State might even set a State monopoly to fully appropriate the GI profits (cf. the case of Cuban Habanos, to which Shapiro’s model applies).25

47To explore this issue under the current framework, I will assume that in the neighbourhood of [Nf, Nc] production technology constraints limit output per firm to a fixed amount independent of membership, set at one unit per firm for simplicity, such that the levels of profits under different levels of membership are comparable. In the scenario in which the social planner maximizes club profits, the game for the social planner is the following:

  1. The social planner assumes GI club firms maximize profits based on the state equation for reputation and the exogenous number of firms in the market, and solves for the steady-state equilibrium, obtaining the result above.

  2. The social planner maximizes profits for the GI club based on such a steady-state and sets club membership Nc indirectly (Nc is a mnemonic for club).

  3. Firms maximize their profits taking Nc as exogenous.

48The process might be gradual, starting from a low N, eventually stabilizing at (Nc, qc, Rc). Step 1 was solved above. In step 2, the social planner assumes a steady-state / symmetric equilibrium, so the problem is no longer dynamic, but static:

49Subject to:

50Profits are maximized at:

51First, the club profit-maximizing level of membership, Nc, exists in the range of positive profits π = p – c – G/N ≥ 0. A monopoly may no longer be viable. Second, the possible range for quality is above the generics level and below the perfect information optimum, so that

52The second-order condition for a relative maximum is that the second derivative be negative. In the range where quality increases with N, where the left-hand side (LHS) is strictly positive, firm profits always increase as N increases, and so do club profits. At qmax, where

  • 26 The level of membership that maximizes profits at the firm level is always that of qmax unless G is (...)

53, firm profits reach their maximum.26 In the range where quality decreases with N, profits initially increase with N up to Pmax, and then begin decreasing.

54Proposition 3: The top-down State-driven approach to profit maximization (maximization of club profits) implies a greater optimal level of membership and a lower level of quality than the bottom-up, firm-driven process of profits maximization (maximization of firm profits).

55Figure 2.5.c shows that the membership that maximizes profits at the firm level is always lower than the membership that maximizes profits at the club level. In addition, the latter implies a lower level of quality in equilibrium.

56This result can be interpreted with reference to the membership elasticity of firm profits eπ,N. Club profits reach their maximum when the membership elasticity of firm profits equals –1: they increase at higher levels and decrease at lower levels. Notably, if eπ,N = 0, then quality is maximum but club profits still have room to increase with increased membership.

57Corollary 1: In the absence of strict standards of quality, the top-down (State-driven) versus bottom-up (firm-driven) approaches to GI protection and territory delimitation, with their impact on optimal club membership, involve a trade-off between club profits/ firm profits/ total sales/ quality.

Figure 2.5.c: Club versus firm profits maximization

Figure 2.5.c: Club versus firm profits maximization

2.6. Minimum quality standards and branding

58In the previous analysis, the emphasis was on understanding within-club dynamics under the hypothesis of endogenous quality and reputation, corresponding to the minimum level of protection and requirement granted under the TRIPS Agreement. But GIs are also a hybrid tool of differentiation as they signal origin and, more often than not, a production process and a minimum level of quality (in addition to their intrinsic characteristics as varieties). The role of GIs is to mitigate market failures deriving from information asymmetries, in conjunction with other differentiation and signalling devices available to firms such as brands, safety labels, standards, and technical regulations (TRs).

59Regarding standards, the World Trade Organization offers a framework of analysis in its Agreement on Technical Barriers to Trade (TBT Agreement). “Technical regulations” are specifications that products must fulfil mandatorily, out of security or safety concerns. “Standards”, on the other hand, are adopted on a voluntary basis, and often include specifications developed for conformity with other products (e.g. videocassette recorder VCR v. digital video disc DVD), not necessarily denoting some higher quality. That said, some standards – certification labels, which define high-quality, environmental, or “healthy” standards, such as the French “Label Rouge” for meat, or the French “Bio” and American “Organic” labels – do denote a minimum quality. These procedures might be either privately or publicly set and performed, and are available to all firms after compliance is determined through an independent certification process.

60In addition, firms might develop an above-quality product and signal that level through proper branding and firm-level traceability. One cannot determine a priori a ranking of the qualities inherent to each instrument, although it can be safely assumed that a GI firm that has a strong brand should have a higher quality than that of its club sisters (otherwise the investment in the brand is not justified).

61Marette (2005) suggests that the concentration ratio (CR) in the industry determines the preferred device (on average). For example, brands are broadly favoured in the sectors of mineral water and baby food, which have a CR3 of 80% (for baby food, it reaches 95% in the United States with Gerber, Beech-Nut, and Heinz). With a CR3 of just 25 per cent, the cheese and meat sectors are dominated by GIs and so-called “labels”. Other sectors with lower levels of concentrations tend to simply abide by the technical regulations in place, which correspond with the legal minimum.

Table 2.6.a: Industry concentration and preferred differentiation device

Table 2.6.a: Industry concentration and preferred differentiation device

62Considering that under symmetry the industry level of concentration is given by the reciprocal of the number of firms, these facts are coherent with the framework developed above. Notably, since the reduced membership around a GI implies a self-fulfilling, above-generics level of quality, it seems that imposing a minimum quality standard would inflict only a minor additional cost to firms, against an increase in profits. On the other hand, a monopolist or a large firm has an advantage in investing in its own brand (Shapiro’s result).

2.7. Champagne: strong brand development (case study)

63Among wine and spirits GIs, Champagne’s strong brand development sets it apart. Intangible Business, a company based in London that specializes in valuing brands and other intangible assets, issues a report on the Top 100 brands in the spirits and wine sectors in terms of both market share and broad measures of brand awareness and price premia (The Power 100: the world’s most powerful spirits and wine brands, 2006 to 2009 editions). The following stylized facts stand out (Table 2.8.a):

64Of the ten most powerful sectors, Champagne ranked eighth in 2006, after Whisky, Vodka, Rum, flavoured spirits, still light wine, light aperitif, and Cognac, and before Gin and Tequila, with twelve brands in the Top 100 constituting a combined market share of 39 per cent (volume based). The 2009 ranking does not single out Champagne from other sparkling wines (ranked seventh), although nine Champagne brands made it to the Top 100, with a combined market share of 45.6 per cent.

65Of the top ten countries of origin, France came third in 2009 (fourth in 2006), after Scotland and the United States and before Russia, Cuba, Italy, Mexico, Sweden, Ireland, and England. Of the 18 and 21 French brands that made it to the global Top 100 in 2009 and 2006, not a single one was of wine (aside from Champagne). This is a sector in which France has strong GIs (Bordeaux, Bourgogne, Beaujolais to name a few), and yet brand power is dominated by the United States (with eight brands), Australia (six brands), Chile, Spain, and South Africa (one brand each, Concha y Toro, Torres, and Kumala respectively).

66These facts strongly support the thesis of this book: that once a club has achieved market saturation, and under the condition that club market concentration is high, firms have an interest in differentiating themselves from their within-club peers through the development of their brands and a constant level of quality.

67In 2009, the appellation Champagne covered 4,776 farmers, 66 cooperatives, and 293 businesses operating in ans area of 33,106 hectares, 85 per cent of which belonged to farmers. Total sales in 2009 amounted to 352 million bottles (12,263 kg/hectare), worth EUR 3.7 billion (of which EUR 1.6 billion were exports). There is, however, strong market concentration, as Table 2.7.a shows a C10 of 82 per cent and a C20 of 96 per cent for 2009.

Table 2.7.a: Champagne market concentration 2009

Table 2.7.a: Champagne market concentration 2009


68Of total sales, the Maisons de Champagne accounted for 66 per cent of sales in volume, 3/4 of sales in value, and 90 per cent of exports. As their site puts it, this is equivalent to the sale of 30 Airbus or of more than one week of purchases of oil for the entire country.27 Additionally, this leads to important price premia – according to the Comité Interprofessionnel du Vin de Champagne, Champagne represents eight per cent of wine sales in volume and 32.9 per cent of wine sales in value. Market concentration reveals economies of scope as well. LVMH and Pernod Ricard rank first among Champagne brand holders (with five and two brands in the Top 100 each in 2006, reduced to three and one in 2009), but are also significant players in other wine and spirits sectors (in the Top 100, Pernod Ricard ranks second in 2009 with 19 brands – Absolut, Chivas Regal, Ballantines, Ricard, Havana Club, Beefeater, Kahlua – and LVMH fourth with four brands (Cognac Hennessy coming in fifth place).

69Interestingly, few Champagne brands made it to the Top 10 ranking regarding market share and brand awareness (Moët & Chandon ranked 5 in both in 2006 and 4 and 7 respectively in 2009) and growth prospects (Veuve Clicquot 6 in 2006). But Champagne brands stand out in the Top 10 for heritage (Dom Perignon 1 in 2006 and 2009, Veuve Clicquot 5 in 2006, Moët & Chandon 9 in 2006, 5 in 2009) and premium price positioning (Dom Perignon 1 in 2006 and 2009, Krug 2 and Bollinger 7 in 2006). Regarding the Top 10 on soft measures on branding, in 2009 Champagne brands ranked 4, 9 and 10 on brand relevance (capacity to relate to the brand and propensity to purchase) and 2nd and 8th on brand perception (loyalty and how close a strong brand image is to a desire for ownership).

2.8. Concluding remarks

70In this chapter, I argue that firms in a GI club enjoy a reputation which is traceable through the GI label, and develop a dynamic model of club reputation in which quality is endogenously determined at the firm level, with reputation as the state variable. Contrary to previous work, it is shown that the GI legal construct under TRIPS is potentially compatible with an equilibrium involving a self-fulfilling level of quality (and reputation) that is above the minimum and that does not fatally lead to a race-to-the-bottom over quality, with the conditions that the GI club has a reduced membership of firms and a highly differentiated variety. The establishment of a minimum level of quality, however, is still the first best policy to improve firm profits. In the same framework, it is demonstrated that under bottom-up (firm-driven) processes of club formation, levels of quality and profits are higher, and levels of club membership lower, than under top-down (State-driven) processes, where the contrary holds true.

Table 2.8.a: Case study on Champagne brands

Table 2.8.a: Case study on Champagne brands


20  The assumption of a firm producing one item of a single quality is made without loss of generality, as proved by McQuade, Salant and Winfree (2010), who show that starting from a similar model in which firms are allowed to produce multiple qualities at different levels of output, the solution involves a single mass point of quality for each firm and an interior equilibrium of output and quality that is symmetric for all firms.

21  Since the quantity sold by the firm is normalized to one, marginal price equals marginal revenue.

22  It is shown below that the main findings hold for an S-shaped function (at low-quality levels, up from some quality level) or for a convex function, but at the cost of a loss of generality.

23  Although a concave marginal revenue function is usually assumed, the model does not rule out a convex marginal revenue function, which happens, for example, in the case of extremely rare wine vintages such as Champagne, wherein increases in quality are marketed at huge price premia. In addition, under the assumption that the function is concave, a saddle-point path equilibrium is not completely ruled out either, though it comes at the price of an additional condition. Wherever the determinant is positive, the system is unstable: i.e. there is a direction in a 2-dimensional space in which the system will not tend to return back to the equilibrium point.

24  This question is related to the literature on (Old and New) Institutional Economics, which focuses on the social and legal norms and rules that underlie economic activity (Commons 1931, Coase 1937, Williamson 1985). In an interesting non-theoretical application of this literature, Barjolle and Chappuis explain why the artisanal food supply chains are organized as “hybrid forms” (between the two poles of market and hierarchy – or the integrated firm), with a case study for a well-known GI, Gruyère (Barjolle and Chappuis 2000).

25  Club goods theory, as a theory of consumption, sheds light on this question. Consider a golf club. It has been proved that if the golf club is member-driven, then it is optimal for the club to maximize average net consumer utility. If the golf club, however, has a single profit-maximizing owner, then it is optimal for the owner to maximize its profits, even if it implies a lower average utility per club member. Refer to Cornes and Sandler 1996, p. 420 for a discussion and an example.

26 The level of membership that maximizes profits at the firm level is always that of qmax unless G is decreasing, so that G' ≤ 0, a situation that would imply a subsidy, is ruled out by assumption.


Table des illustrations

Fichier image/jpeg, 44k
Fichier image/jpeg, 48k
Fichier image/jpeg, 36k
Fichier image/jpeg, 48k
Fichier image/jpeg, 72k
Fichier image/jpeg, 36k
Fichier image/jpeg, 28k
Fichier image/jpeg, 84k
Fichier image/jpeg, 80k
Fichier image/jpeg, 24k
Fichier image/jpeg, 16k
Fichier image/jpeg, 24k
Fichier image/jpeg, 24k
Titre Figure 2.5.a: Equilibrium steady-states as a function of the number of firms
Fichier image/jpeg, 792k
Fichier image/jpeg, 212k
Titre Figure 2.5.b: Equilibrium steady-states without GI club costs
Fichier image/png, 61k
Fichier image/jpeg, 220k
Fichier image/jpeg, 36k
Fichier image/jpeg, 88k
Fichier image/jpeg, 112k
Fichier image/jpeg, 24k
Fichier image/jpeg, 24k
Fichier image/jpeg, 72k
Fichier image/jpeg, 52k
Titre Figure 2.5.c: Club versus firm profits maximization
Fichier image/jpeg, 604k
Titre Table 2.6.a: Industry concentration and preferred differentiation device
Fichier image/jpeg, 124k
Titre Table 2.7.a: Champagne market concentration 2009
Crédits Source:
Fichier image/jpeg, 88k
Titre Table 2.8.a: Case study on Champagne brands
Fichier image/jpeg, 134k


Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search