Are geographical indications a worthy quality label? A framework with endogenous quality choice

1. Introduction

Our focus is on a food quality standard, the European protected designation of origin (PDO), which signals an immaterial dimension of quality, namely ‘terroir’ – meaning some tradition, authenticity and typicality derived from a combination of natural conditions (soil, climate etc.) and accumulated know-how in a given place of origin. European Union (EU) regulation 510/2006 defines the PDO standard as designating a product that originates from a specific geographical area where it is produced, processed and prepared, and ‘the quality or characteristics of which are essentially or exclusively due to a particular geographical environment with its inherent natural and human factors’ (Council of the European Union, 2006). This EU regulation does not stipulate how these standards should be met in practice. Its detailed implementation is defined in product specifications, which are different for each of the European PDOs (there are currently over 500). Each of these product specifications reflects a local balance of power, as the decision process for PDO creation or evolution devotes a central place to professional organisations (Ansaloni and Fouilleux, 2008). The legal framework, therefore, allows for heterogeneity in production and processing rules and in the link between quality and geographical origin. A recurring element of the general debate surrounding geographical indications (GIs) is whether they are a genuine quality signal or a rent for producers (see, in particular, Josling, 2006).

The preamble to the EU Regulation offers the political view of Europe on this matter by stating that PDO protection is beneficial to both consumers and producers: for consumers, by giving them information regarding the origin of products and by making available high-quality products with guarantees on the method of production and origin; for producers, by helping to secure higher incomes in return for a genuine effort to improve quality, and by retaining rural population in less-favoured or remote areas. However, the way the PDO system is applied in practice does not appear to guarantee a link between a geographical name and high quality (Lucatelli, 2000). Benefits to consumers depend on the extent to which natural factors and accumulated know-how in the place of origin actually confer some specific quality attributes on products, the extent to which the GI label encourages producers to improve their quality and the extent to which consumers intrinsically value the fact that production took place in a given area with given methods of production, independent of the quality attained.

In the theoretical economic literature, GIs are usually modelled as a signal of high quality in a vertical differentiation context. In some articles, the high quality of the GI good is exogenous and its existence precedes that of the label. The GI label aims at informing consumers on this quality level, which they cannot assess in its absence (Marette, Crespi and Schiavina, 1999; Marette and Crespi, 2003; Zago and Pick, 2004). Alternatively, some authors assume that GI producers have the ability to produce an exogenously high level of quality only if they invest fixed costs of quality improvement. Given imperfect information, the GI standard gives them the incentive to invest in quality (Lence et al., 2007; Mérel, 2009). In other papers, GI producers face a competitive market of low-quality products and strategically choose the high-quality level of the GI product (Moschini, Menapace and Pick, 2008; Bouamra-Mechemache and Chaaban, 2010; Mérel and Sexton, 2012). Finally, Menapace and Moschini (2012) use a dynamic model of reputation with endogenous quality choice in competitive markets. Producers within a region with distinctive ‘terroir’ features may use GI certification and/or private trademarks in order to establish a reputation of high quality. They compete with producers who use a standard technology and who may therefore only use private trademarks.

This literature addresses the welfare implications of GI labelling in a context of asymmetric information (for a survey, see Teuber, Anders and Langinier, 2011). Some studies investigate GIs as a way of direct or indirect supply control. They find that allowing GI producers to collude may improve general welfare by enabling these producers to cover the fixed costs of quality development and certification (Marette, Crespi and Schiavina, 1999; Marette and Crespi, 2003; Zago and Pick, 2004; Lence et al., 2007; Mérel, 2009). Bouamra-Mechemache and Chaaban (2010) investigate whether producers with a quality advantage should collectively choose a GI certification (modelled as entailing variable certification costs) or a private common label (entailing fixed certification costs). Moschini, Menapace and Pick (2008) differ in their approach, by showing that GI labelling is compatible with a competitive provision of quality with free entry. In this competitive context, they also find that GI labelling may improve welfare. Mérel and Sexton (2012) investigate a situation in which GI producers are able to collude on quality choice, but not on quantities. They find that GI producers may end up supplying a higher level of quality than the welfare-maximising one. Menapace and Moschini (2012) show that GI certification, by acting as a tool for collective reputation, improves welfare compared with a situation in which only private trademarks would be available to firms.

Here, we aim to provide a theoretical framework that studies welfare effects of GIs depending on the extent to which product specifications are binding. In order to do so, we step back from the sophisticated level on which some issues, such as direct versus indirect supply control or dynamic reputation, have been addressed in this literature.

With this in mind, our framework borrows some insights from that of Crampes and Hollander (1995a) on denomination or appellation standards. Their article builds on a stylised fact about the harmonisation of the standard for gold in the EU, which has led France to reduce the number of karats that a metal has to contain in order to be called ‘gold’. The authors examine how relaxing this standard affects producers and consumers. They model the appellation standard as certifying that quality is at least equal to a given threshold. They adopt a vertical differentiation framework in which consumers value the intrinsic quality level of the good they consume, as well as the presence of a label on this good, and in which a firm that has adopted the standard competes with a low-quality firm that has not. The authors show that relaxing the denomination standard benefits high-quality producers. As in their model, we adopt some assumptions on how consumers derive utility both from the existence of the PDO label that signals where the product originates from, and from the actual taste of the product. Our paper also relates to the analysis of Chambolle and Giraud-Héraud (2005), who model quality choices by two firms, one of which can adopt a label of origin, while the other cannot. Their analysis differs from ours in that they assume that the labelled firm must commit itself to quantity restrictions in order to benefit from the label (we do not model any capacity constraint linked to PDO labelling). More generally, our article builds on the industrial organisation literature on endogenous quality choice (notably, Motta, 1993; Cremer and Thisse, 1994; Lambertini, 1996).

In our model, we derive endogenous quality choice both by a firm whose product bears a PDO label and by one that cannot put forward the necessary traditional know-how to benefit from PDO labelling, with a Bertrand or a Cournot duopoly. We examine the equilibrium in which the PDO firm produces the high-quality good. We compare two extreme implementations of the PDO label: either this label only guarantees the origin of the product or it guarantees both the origin and the final quality of the product. We also analyse an extension of this duopoly model in which producer organisations choose the quality of each good and in which a competitive equilibrium with atomistic producers arises once qualities are chosen.

2. Are PDO product specifications stringent? The examples of the Comté and Cantal French cheeses

We illustrate the heterogeneity of PDO product specifications, both through time and between products, with the example of Comté and Cantal, two French AOC cheeses with ancient origins (AOC, which stands for ‘Appellation d'Origine Contrôlée’, is the French equivalent of the European PDO).¹ Both cheeses were recognised as AOC by courts in the 1950s after their local craft unions, respectively, initiated lawsuits against grocers selling cheeses, which originated from outside the traditional production areas, as ‘Comté’ and ‘Cantal’.

The 1952 court decision for Comté, which put forward evidence of the long tradition of Comté production in the region, delimitated the area of authorised production and specified that AOC Comté cheese should be produced in accordance with local, loyal and consistent customs and notably come from local cow breeds, fed according to customs codified in the statutes of the local cheese dairies. The initial decree of the Comté AOC, published in 1958, imposed only the area of authorised production and the condition that it should be produced ‘in accordance with local, loyal and consistent customs’. Gradually, other rules have been introduced on milk production and processing, as summarised in the Appendix (Table A1). Currently, the main rules for milk production are the obligation to use local cow breeds, the prohibition of the use of silage, a limitation on livestock and milk production per hectare and a limitation on feed concentrate. For processing, the main rules are the obligation to collect milk near the production point and the duration of cheese ripening (Colinet et al., 2006).

The first decree of the Cantal AOC was published in 1980, long after a 1956 court decision recognised the AOC. This decree only instituted a minimum of 45 days of cheese ripening (decreased to 30 days in 1986) and a few details concerning the temperature and length of certain manufacturing stages (INAO, 1980 and 1986a). Yet, the 1956 court decision that ruled in favour of the Cantal AOC insisted on the significance of every constitutive detail of the Cantal terroir, notably the volcanic ground, the special local flora, the rustic cow breeds, the transhumance and mountain summer pasture and the traditional manufacturing (see Appendix Table A2). A new specification, adopted in 2007 after a lengthy process and under pressure from the French institute for AOCs, instituted a series of precise rules at the production, processing and ripening stages. These specifications are not completely in line with the initial description of the Cantal terroir in the 1956 court ruling. Dairy and cheese production practices in the area have evolved a lot in the meantime, and no longer match the initial description of the Cantal terroir. Cantal milk production, initially limited to the mountain area during summer, has developed in a peneplain area contained in the AOC area with more intensive milk production methods based on corn silage, intensive Holstein cattle has partly replaced the initial Salers and Aubrac rustic breeds and processing has been industrialised (Colinet, Desquilbet and Hassan, 2007).

For the purpose of our analysis, two elements of these examples are noteworthy. First, even though the EU PDO Regulation requires that the quality or characteristics of the PDO product must be ‘essentially or exclusively due to a particular geographical environment with its inherent natural and human factors’, our examples show that the extent to which actual PDO specifications convey a specific terroir varies in practice. Second, the overall production/processing specifications are more restrictive for Comté than for Cantal. At least some of these specifications (e.g. lower stocking rate, specific cow breeds) impact the quality of the product. These conclusions are consistent with those of Barjolle and Sylvander (2000), who report the results of case studies that qualitatively analyse the specificity of 16 PDOs in six European countries, with respect to the characteristics of the product, its perceived quality, its technological factors and its denomination. These case studies indicate that PDOs are heterogeneous and that the presence of a PDO label does not necessarily imply a strong specificity of the product.

3. Theoretical framework and benchmark equilibrium with no labelling

We adopt a model that draws on standard features of the vertical differentiation literature. We consider two goods: a good with PDO labelling, denoted a, and a good for which PDO labelling is not possible, denoted b. The economic literature assumes either that both the high- and low-quality levels are exogenous (Marette, Crespi and Schiavina, 1999; Zago and Pick, 2004), or that only the high-quality level (enjoying a GI) is endogenous (Moschini, Menapace and Pick, 2008; Menapace and Moschini, 2012; Mérel and Sexton, 2012). In our framework, we assume that the quality levels of both the PDO and the non-PDO goods are endogenous. We study the equilibrium in which the PDO good is of higher quality than the non-PDO good. We consider two extreme alternative implementations of the PDO label. We call the PDO label a ‘minimum quality requirement’ when PDO specifications impose both local production and binding quality requirements, so that the final product meets at least some given quality level.² Alternatively, we call the PDO label a ‘denomination standard’ if PDO specifications require that every production step takes place within a given area, but do not induce any effective quality constraint.

In models on GIs, the quality level is typically one-dimensional and the high-quality good is a credence good (or an experience good in the dynamic framework of Menapace and Moschini, 2012). We distinguish two attributes of goods: gustatory quality and geographic origin. Gustatory quality of good i is a one-dimensional continuous vertical attribute labelled q_i (the higher q_i, the better the taste of good i). Within the usual classification of search, experience or credence attributes (Nelson, 1970; Darby and Karni, 1973), taste is typically an experience attribute, known only after purchase. However, for a product purchased repeatedly, which is usually the case with food products, consumers have an idea of its taste before purchase (as long as the product is identifiable, for example, by a trademark, and its taste is stable enough over time). In our one-period model, we handle this gustatory quality as a search attribute (detected before purchase). We leave out any informational problems related to this gustatory quality, assuming, as in Menapace and Moschini (2012), that firms can identify their products by trademarks at no cost. In contrast, we assume that consumers are not able to discern whether the good originates from a particular area and, therefore, the attribute of geographic origin is a credence attribute, which can be revealed only by public labelling.

The marginal consumer who is indifferent between the two goods is characterised by $θ~=(pa−pb−g)/(qa−qb)$⁠. As long as $qa>qb$ (which is the case studied in this paper), the demand function for the PDO good is given by $xa(pa,pb,qa,qb)=θ¯−θ~$⁠. The demand function for the non-PDO good $xb$ depends on whether the market is covered (each consumer buys one of the two goods) or uncovered (some consumers, with a low, do not buy anything). If the market is covered, it is given by $xb(pa,pb,qa,qb)=θ~−θ_$⁠. If the market is uncovered, the consumer who is indifferent between consuming the non-PDO good or nothing is characterised by $θ0=pb/qb$ and the demand function for the non-PDO good is $xb(pa,pb,qa,qb)=θ~−θ0$⁠.

Our assumption that consumers derive a positive utility from the origin of the good conveys the idea that the PDO label has a positive image, which may confer some ‘status’ characteristic on a good bearing this label.³ Alternatively, it may reflect the idea that consumers are altruistic and consider the purchase of PDOs as contributing to a public good, such as the maintenance of rural activity and traditional production (although we do not model such a public good in our welfare function). This assumption is consistent with the results of empirical studies that evaluate the willingness to pay (WTP) for PDO. In order to disentangle the intrinsic quality (the gustatory taste) and the extrinsic quality (the geographical origin) of PDO products, Loureiro and McCluskey (2000), Hassan and Monier-Dilhan (2006), Fandos and Flavián (2006) and Menapace et al., (2011) take some cues for gustatory taste, such as colour, freshness, aroma or fat content, into account. The main result of these studies is that, after controlling for the effects of available variables that describe the intrinsic quality, a positive consumer WTP for PDO products remains. (However, it is not possible to determine to what extent this positive WTP is attributable to an attachment to the geographical origin per se or to a higher gustatory quality that results from immeasurable characteristics of PDO products.)

Analogously to Crampes and Hollander (1995a), we assume [equation (1)] that all consumers derive the same utility from the origin but, for the purpose of simplification and contrary to these authors, we do not assume that the utility attached to the origin depends on the gustatory quality of the good (although we report simulation results with the assumption that the preference for origin increases with the quality level, with $g(qa)=γqa$⁠).

Our assumption of one firm per product, with strategic behaviour both with and without labelling, applies to the processing stage of PDOs, in which there is often market power, rather than to the farm stage. In this sense, our model can be viewed as an extreme case in which processors have integrated farmers vertically. In the last section, we also study a game in which producer organisations set quality in the first stage and farmers behave competitively in the second stage.

4. PDO labelling: denomination standard versus minimum quality requirement

In the case of the denomination standard, the PDO label only indicates the geographical origin, while product specifications stating how the PDO good must be produced do not impose any effective production constraints. The minimum quality requirement represents the extreme case in which production practices of the PDO product have to ensure at least a given level of product quality.

Proposition 1.A analyses the effects of the introduction of a minimum quality requirement in a Bertrand duopoly with a covered market (proof is given in Appendix). Table 1 gives equilibrium values with a denomination standard and the comparative statics of the introduction of a minimum quality requirement. Proposition 1.B analyses these effects for an uncovered market, when firms compete either in Bertrand or in Cournot, and when the non-PDO firm either bears a cost disadvantage or not. It relies on simulations performed for different values of the cost parameter c, the cost disadvantage for the non-PDO firm f and the preference for geographical origin g.⁸ Table 2 gives simulation results for some specific parameter values. 

The introduction of a minimum quality requirement leads to an increase in the quality level of both the PDO firm and the non-PDO firm, compared with the case of the denomination standard. A main result for our purpose is that the PDO firm has no interest in this strengthening of the quality requirement. This result reinforces the findings of Crampes and Hollander (1995a). From the consumer's point of view, they find that all consumers lose from a strengthening in the standard. We obtain the analogous result that consumer surplus is smaller if production requirements for the PDO label are binding when the market is covered. When the market is uncovered, consumers may be worse off or better off with a minimum quality requirement than with a denomination standard. The global effect on consumer welfare depends on the relative sizes of two opposing effects. In all simulations, consumers who keep consuming the PDO good are better off, because its quality increase more than compensates for its price increase in their utility, while all other categories of consumers are worse off: those who stop consuming the more expensive non-PDO good; those who keep consuming the non-PDO good (given their lower valuation of quality than PDO consumers, for them the quality increase does not compensate for the price increase); those who turn from the more expensive PDO good to the non-PDO good. The effect on global welfare is determined by the relative sizes of the decrease in the profit of the PDO firm, the increase in the profit of the non-PDO firm and the change in consumer surplus.

Our result that the PDO firm obtains a lower profit level with a minimum quality requirement than with a denomination standard is robust to the assumption that the preference for origin, g, either does or does not depend on the quality level of the PDO good (see simulation results with $g(qa)=γqa$ in Table 3). The intuition for this result is the following. In the situation in which the PDO label acts as a denomination standard and the utility derived from the origin depends on the quality level, the PDO firm has an incentive to take advantage of this structure of preferences by setting a higher quality level than it would have if the utility from the origin was independent of the quality level. If the regulator sets a minimum quality requirement that imposes a further increase in this quality level, it is detrimental to the PDO firm. This is what occurs in our simulations. With an invariant g = 0.05, in Table 2, the equilibrium value of the PDO quality is 0.719 with a Bertrand duopoly and a denomination standard. With $g(qa)$⁠, setting$γ=7/100$⁠, we have that $g(0.719)≈0.05$⁠: that is, if the PDO firm chose to set its quality at 0.719, consumers' utility from the origin would be 0.05 as in the simulation with an invariant g. Yet, the PDO firm has an interest in setting its quality to a higher level, equal to 0.817, in order to benefit from a higher valuation of origin by consumers, $g(0.817)≈0.057$⁠. But it is not profitable for this firm to set its quality to the even higher level imposed by the minimum quality requirement.