Expert Opinion and Gastronomy: The Recipe for Success - Springer Link

16 downloads 0 Views 94KB Size Report
convention” if we refer to the French school of conventions (Favereau and ..... Very honoured to have been awarded the Presidents' Prize at the Rotterdam 2002.
Journal of Cultural Economics 27: 127–141, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.

127

Presidents’ Prize Paper

Expert Opinion and Gastronomy: The Recipe for Success VÉRONIQUE CHOSSAT 1 and OLIVIER GERGAUD 1, 2 1 CERAS-OMI, Université de Reims Champagne-Ardenne, Reims, France 2 TEAM, Université de Paris-I Panthéon-Sorbone, Paris, France

Abstract. Experts’ opinions play an important role in the gastronomic market for the following reasons: information is imperfect and very costly to acquire and quality is, in large part, subjective and consumers need experts to define it. The number of guidebooks currently issued, their success (Michelin: 650,000 sold; GaultMillau: 200,000 sold) and the strong level of correlation generally obtained between prices and ratings or rankings (0.63 on average) for this class of activity illustrate this influence. Without experts, supply and demand would find it difficult to meet. Therefore, identifying the determinants of these evaluations of quality and then estimating their respective impact become relevant. According to the experts, the art of cooking is the only determinant that they take into account when selecting and then evaluating the chefs. For the chefs, the setting also appears to be a determinant and not the least important one. What is the best strategy to become a “first rate” chef? Would Alain Ducasse, one of the most famous French chefs, get the same rating in a roadside café as in a luxury restaurant? To answer these questions, a quality equation is estimated using an original database concerning 185 leading French chefs who have been selected in one of the most famous French guidebooks: GaultMillau (2000 edition). The results show that there are two strategies to become a “first rate” chef but that the art of cooking prevails over setting. This is in line with the observation that some gourmet restaurants tend to over-invest in luxurious surroundings. Key words: expert opinion, gastronomy, quality

1. Introduction Most of the (few) theoretical contributions focusing on the demand for restaurants (Becker, 1991; Banerjee, 1992; Albrecht et al., 2001) either ignore or treat quality as an exogenous variable. Yet, as with many other cultural goods and services (ranging from dance to theatre and from music to wine) and non-cultural services (the stock market, and pari-mutuel horse racing, among others), expert opinion is a critical determinant of the demand for gourmet cuisine.1 The role of expert opinion, and the efficiency of the demand for it, has been questioned in studies of wine (Ashenfelter and Jones, 2000) and of music performances (Ginsburgh and van Ours, 2002).2 In all of these applications, a primary question is, to what extent

128

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

do expert opinions reflect the underlying fundamentals of value and to what extent do they reflect factors that are unrelated to fundamentals? In this paper, we study the determinants of the quality of French chefs as measured by guidebooks. The key question we ask is, are these ratings primarily determined by the quality of the food presented or is something else involved in the critic’s determinations? Evaluating the quality of a restaurant is indeed a difficult task that requires considerable effort for two reasons. First, the gastronomic market is by definition geographically scattered; you need to move around to consume, making information very costly to acquire. Second, evaluation is mainly subjective; the service we are considering here consists mostly of what economists call experience and credence characteristics, that is to say characteristics that are respectively discovered or not after the product has been consumed (e.g., quality of the ingredients, flavour of the preparation, etc.). On the other hand, service possesses few search characteristics, while other characteristics can be ascertained during the search process prior to purchase, for example the restaurant setting.3 Basically, information about restaurants and cooks is largely imperfect. Therefore, the main problem facing consumers of gastronomy is how to obtain useful information. In this article, we empirically analyse how experts rank restaurants, what their criteria are and what the relative impact of these criteria is on overall ranks or grades. This issue is of interest because experts or gastronomic guidebooks not only have a potentially major influence on price levels, but they also have an influence on consumer choice and even perhaps on the style of cuisine (classical, creative, regional, . . . ). Suppose that creation is in the air and classicism out of fashion. One can expect that chefs would tend to favour creativity instead of classicism in order to get good assessments and then charge more. The first empirical evidence of the major role played by critics in this market is certainly the strong and significant statistical link generally obtained between the price charged by a restaurant and the main assessments it receives from the most influential guidebooks.4 The second evidence is the fact that, in most developed countries, guidebooks are numerous and varied.5 The expert’s role is to select from among a wide range of chefs those who merit the “grand chef” denomination, i.e., the high-class cooks. According to Wijnberg (1995) “in expert selection, a special evaluative capacity is attributed to a relatively small group of selectors who are not members of the group in which the selection process takes place”.6 This procedure results from the application by each guidebook of a specific pattern of criteria of gastronomic quality: its “quality convention” if we refer to the French school of conventions (Favereau and Lazega, 2002). Because gastronomic guidebooks are private businesses in competition with one another and because of the necessity of assessing the creative dimension of the service, experts keep their way of rating secret for strategic purposes. Here the Red Guide (Michelin) distinguishes itself as the most discrete: no comment is given with its stars, contrary to its main competitors.7 Indeed, Bottin Gourmand,

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

129

Champérard and GaultMillau provide a comment on each restaurant but remain extremely evasive about the way restaurants are assessed (either ranked or rated). In these guides, it is generally declared that only the cuisine (creativity, technicality, quality of the ingredients, etc.) is evaluated and then awarded. In other words, elements of setting are claimed to have no influence on assessments even if the written comments often contain indications about the setting (ambience, service, cellar, etc.). For the chef, setting is an alternative strategy that can be implemented in the hope of rising in the hierarchy. Indeed, adjusting setting level to cuisine level seems to be particularly well perceived by the experts. Who is right? Chefs or experts? Do the chefs only have one strategy (cuisine) to obtain good ratings from the experts, or do they face a trade-off between two strategies (cuisine and setting)? Would Alain Ducasse, one of the most famous French chefs, obtain the same ranking in a roadside café as in a luxury restaurant? What are the incentives for a chef to invest in his setting? In his cuisine? Do chefs invest in setting to the detriment of their cuisine? Is it efficient to maintain a good cellar in a restaurant to supply good wines to customers? And what about convivial service? Is cuisine more, less, or as important as restaurant setting? For all of these reasons, wondering what the recipe for success is in French gastronomy becomes relevant. To answer these questions and find out this mysterious recipe, we estimate a quality equation using an original database that includes nearly 200 leading French chefs who have been selected by GaultMillau (2000 edition). This guidebook has been used because of the detailed comments it contains. This guide is Michelin’s main challenger (200,000 copies in 2000). Contrary to the situation of wines for which a precise glossary exists, in gourmet cuisine no official glossary is available. Moreover, in this analysis we have to take into account the fact that setting (cadre, service and cellar) is potentially correlated to cuisine. This endogeneity problem is common in economics, particularly in labour economics where the correlation between education and wages does not necessarily mean that the first factor determines the second (see Card, 1999, for a survey). This article is organised as follows. In Section 2, the data are presented. Section 3 contains the empirical results. Some concluding remarks follow. 2. The Data 8 The sample has been restricted to a particular category of head cooks (assistants are not considered); they are members of The Master Chefs of France Association (1999 edition) whose restaurants are located in metropolitan France and who work in the commercial fringe of the catering market. 185 cooks are included. The GaultMillau (GM) guidebook rates the restaurants on a scale ranging, in theory, from 0 to 20, but in practice the grades that are issued (GRADE) are between 11 and 19. It is useful to indicate that there is a strong and significant correlation between the grade attributed to a cook and the length of the comment (0.84) and that the structure of the comment is stable with three parts. The first part

130

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

concerns exclusively the cook and his cuisine. It begins with some remarks about geographical location and the history of the restaurant, of the cook and his family and continues with a presentation of the cuisine itself – quality of the dishes, technicality, know-how, professionalism, cooking style, quality of the ingredients, etc. The second part focuses on any extra supply of amenities, usually wine, whereas the last part is about service and setting. To sum up, grade dispersion seems to depend on three main criteria: 9 1. the cuisine, 2. the cellar, 3. the setting and the service. 2.1. CUISINE 2.1.1. General Comments The general comments introduce the chef and his way of cooking with more or less laudatory expressions as, for example, “he sits now on the top of the Mount Olympus of gastronomy” or “a megachef”. Comments may also mention one or several particularly appreciated courses.10 These two variables have been gathered into a unique count variable to avoid a potential multicolinearity problem. These superlatives give a global image of how a cook and his production are perceived by the GM experts. The number of superlatives is represented by 4 dummies: SUP1_5 when the total is greater than or equal to 1 and less than 5, SUP5_10 when the total is greater than or equal to 5 and less than 10, SUP10_15 when the total is greater than or equal to 10 and less than 15, and SUP15 when the total is 15 or more. The cook or his way of cooking may, in some circumstances, be criticised: “a lack of maturity” or “a disappointing experience”. In this case the dummy (CRIT) will take the value 1, otherwise 0. In most cases, the chef’s cooking style is emphasised. Three categories of cooking style have been used: regional cooking (terroir in French) (REG), as in “you can smell the bracing air of the rural Lozère region”; creative cooking (CREA) as in “reinvented seafood”; and classical cooking (CLAS) as in “classic dishes that built his success”. La cuisine de terroir (sic) is regional cooking: bouillabaisse in the South of France, quenelles in Lyon, choucroute in Alsace region, and so on. Creativity consists, for example, of introducing radical changes in the traditional way of cooking, or experimenting with new combinations of foodstuffs like fish and meat in the same course. Classical or traditional cooking signals the reproduction by a chef of the French gourmet cuisine standards or canons (volaille en vessie, or bœuf bourguignon, for example). When no style is suggested or perceptible, the dummy NSTYL takes the value 1, otherwise 0. Because a chef often combines more than one of the styles of cuisine we have been listing above, dummies CLCR, CLRE, RECR, CLRECR will also be included in the expert’s grade equation. These dummies take the value 1 when combinations – classical-creative, classical-

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

131

regional, regional-creative, or all three styles – are mentioned; otherwise, they are all zero. 2.1.2. Technical Comments The guidebook also mentions the chef’s technicality or know-how, referring, for instance, to the quality of sauces and/or of preparations. When the content of the comments is rather positive, as for example in “chicken breast in a delicate chervil sauce”, the dummy TECH+ takes the value 1, otherwise 0. On the other hand, when this attribute is perceived negatively, as in “a quail-breast puff pastry was on the dry side”, TECH– takes the value 1, otherwise 0. 2.1.3. Other Comments Dummies FLAV+ and FLAV– take the value 1, respectively, if flavour combinations prove to be a success, as in “a wonderful combination of flavours and tastes”, or fail, as in “the red peppers overwhelmed the lobster salad”, otherwise 0. Some remarks concern the delicacy of preparations (DELIC): “a light salad of bay scallops and green asparagus”; their simplicity (SIMPL): “sublimely simple pan-roasted langoustines”; their subtlety (SUB): “a model of finesse”; or the chef’s artistic skill (ART): “proves his great artistry” or “his creations are staggering”. Moreover, certain comments often suggest a positive sophistication (SOPH+): “a wonderful sophisticated dish”, or a negative one (SOPH–), that is useless, even harmful to the preparation, as in “too much sophistication in this dish”. Generosity has an ambivalent dimension: it can be positive (GEN+) as in “he serves in generous portions” or negative (GEN–) as in a “lack of generosity”. Finally, sometimes the guidebook emphasises the perfect harmony between all the dishes composing the menu (MENU). All these variables are dummies, which take the value 1 when the expert considers that the cuisine has the so-called characteristic, otherwise 0. 2.1.4. Quality of the Ingredients Insofar as a chef is also evaluated on his capacity to select good ingredients, the GM points out the use of good ingredients (ING+): “all ingredients are selected with the same eagle eye” as well as bad ingredients (ING–): “spoiled by indigestible beans”. Finally, we created several dummies to account for the use by the chef of rare foodstuffs such as noble fish (salmon, turbot, . . . ) (FISH), crustaceans (crawfish, . . . ) (CRUS), caviar (CAV), foie gras (FOIE) or truffles (TRUF). For some restaurants, it may happen that the guidebook emphasises “the very good cheeses”. In this case, the dummy CHEESE will take the value 1, otherwise 0. 2.2. CELLAR The GM gives very general comments about the wines a customer may find in a particular restaurant. For this reason, we prefer another indicator for cellar quality,

132

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

the one contained in the Champérard guidebook. In this guide, a restaurant can get from 1 to 4 wine-taster symbols (WINE = 1 to 4), 4 corresponding to an exceptional cellar with many great wines.

2.3. SETTING AND SERVICE Our analysis of setting and service is based on three main categories of indicators contained in three different guidebooks: GM, Michelin and Champérard. GM dispatches the restaurants in three different categories: “Bistrot, neobistrot, brasserie and auberge (inn)” (SETGM = 1), “traditional, comfortable or even elegant” (SETGM = 2), and “prestige, luxury restaurant” (SETGM = 3). Michelin distinguishes more categories for setting by giving a restaurant from 1 to 5 black or red “forks and spoons”. In this context, red is better than black and signals a more pleasant setting within each category. In the experts’ grade equation, the variable F&S will indicate the number of “forks and spoons” attributed by this guide to a particular restaurant and the dummy RED will take the value 1 if the colour of these “forks and spoons” is red, otherwise 0. Champérard describes the setting and service of the restaurants by the device of a number of candlesticks (CAND) and trays (TRAY), which vary from 1 to 4 according to, respectively, the quality of the setting and of the service.

2.4. OTHER VARIABLES 2.4.1. New Restaurant The dummy NEW indicates that the restaurant appears for the first time in the GM guidebook (2000 edition). With this variable we test whether newly referenced restaurants are evaluated in the same way as restaurants that entered the guide earlier. 2.4.2. History of Gastronomy Sometimes, GM mentions whether restaurants or chefs belong to the history of gastronomy. Because this characteristic is bound to influence the grade, the dummy HIST will take the value 1 when comments include remarks such as “this restaurant is a sanctuary for the Landes tradition”, otherwise 0. 2.4.3. The Name of the Chef In most cases (81.1%), comments explicitly include the cook’s name. We have decided to take into account this disparity with the dummy NAME, which takes the value 1 when the name is mentioned, otherwise 0.

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

133

3. Estimation Procedure Our initial idea was to regress GRADE against CuisGM, a vector representing the 34 variables found for Cuisine (see sub-section 2.1); SET, a two-variable vector including WINE and SETGM; and X, a vector of exogenous variables (NAME, HIST and NEW). GRADEi = CuisGMi β + SETi γ + Xi δ + ui

i = 1, . . . , n.

It was tempting first to reduce SET to its single principal component and CuisGM to a less important vector by using a principal component program. But the results obtained for CuisGM were disappointing. However, the first principal component for SET took into account nearly 78% of the total variation. Therefore, for easier computation we decided to pursue the procedure with SET as an indicator.11 In this estimation procedure, we assume implicitly that the causal direction is from cuisine criteria to the grade and not the contrary. In other words, CuisGM is considered as exogenous because criteria and ratings are determined first by the expert. This is not a strong assumption if we refer to the example of wine in which grades are generally determined following the Davis method, which consists of weighting four families of criteria: visual examination (12%), olfactory examination (24%), gustatory examination (40%) and general remarks (24%). However, we assume that setting assessments (SET) are endogenous with respect to CuisGM. Indeed, it is likely that the chefs choose the setting non-randomly, and we can expect ex-ante a significant correlation between SET and CuisGM. To avoid the potential pitfalls associated with this typical endogeneity problem and to obtain the true effect of setting, we use instruments to predict SET. The instruments we have retained are the assessments given by Michelin (F&S and RED) for the setting. The main motivation for this choice is that these variables affect statistically the setting but have no (or little) effect on cuisine criteria.12 On the contrary, assessments made by Champérard (CAND and TRAY) have been excluded from the regression because of a significant level of correlation with CuisGM. The R-squared obtained for the OLS regression 13 of SET against F&S and RED is 0.257.14 To see whether OLS or 2SLS is the appropriated estimator here, we apply the Hausman (1978) test, which consists of adding a variable SETH – a prediction of SET thanks to the auxiliary regression described just above – to the experts’ grade equation and testing whether the corresponding parameter is significant or not – this is what we call the Wu statistic after Greene (2000, pp. 383– 384). A simple Wald test proves that it is significant at a 5% level (χ 2 calc = 4.02 for 1 df.) and therefore that 2SLS is the appropriate estimation procedure.15 Finally, because grades are not really continuous (11.5 is not a choice for example) but ordered, we run the regression following the ordered logit method (Greene, 2000, pp. 875–879). The results are presented in Table I. With a Pseudo R2 of 0.3756, the quality of the fit is quite satisfactory for this kind of regression.

134

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

Table I. Ordered logit parameter estimates for factors affecting GaultMillau grade 16 Variables

Abbreviations

Coefficient

z-stat.

Cuisine General comments: – Number of superlatives: [1; 5[ [5; 10[ [10; 15[ ≥15 – Number of general criticisms – Style of cuisine: Not indicated Only regional Only classical Only creative Classical and regional Classical and creative Regional and creative Classical, regional and creative

SUP1_5 SUP5_10 SUP10_15 SUP15 CRIT

– 0.78 3.29 5.34 –0.36

– 2.01 b 4.69 a 4.09 a –0.71

NSTYL REG CLAS CREA CLRE CLCR RECR CLRECR

– –0.46 0.37 0.65 0.98 0.68 0.00 1.86

–0.79 0.69 1.2 1.61 1.17 0.001 2.68 a

Technical comments: – Good technicality (cooking, taste, . . . ) – Criticism of technicality

TECH+ TECH–

1.63 –0.05

4.44 a –0.1

FLAV+ FLAV– DELIC SIMPL ART SUB

0.94 1.78 1.21 1.11 0.14 0.14

2.08b 2.12 b 1.96 b 2.49 a 0.15 0.3

SOPH+ SOPH–

1.95 –

1.81 c

GEN+ GEN– MENU

–1.07 – 0.86

–1.77 c

ING+ ING–

–0.02 0.73

–0.05 0.82

FISH CRUS CAV FOIE TRUF CHEESE

0.17 –0.05 1.02 0.45 0.72 0.67

0.52 –0.14 1.20 1.25 1.62 0.68

Various comments: – Flavour combination: Good Bad – Delicacy of courses – Simplicity of courses – Artistic design of courses – Subtlety of courses – Sophistication: Useful Useless – Generosity of courses: Emphasised Lacking – Nice menu (rich, diversified, . . . ) Ingredients: – Outstanding quality – Bad quality – Rare foodstuffs: Fish (salmon, turbot, . . . ) Crustaceans (crawfish, . . . ) Caviar Foie gras Truffles – Cheese

1.82 c

135

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

Table I. (Continued) Variables

Abbreviations

Setting (cellar, service and cadre)

SETH

1.45

5.02 a

Other variables – Newly referenced restaurant – History of gastronomy – Chef’s name

NEW HIST NAME

–2.65 1.22 0.9

–3.78 a 2.06 b 2.17 b

Number of observations χ 2 (34 df.) Log likelihood Pseudo-R 2 Threshold parameters: 12 13 14 15 16 17 18 19

Coefficient

z-stat.

187 268.77 –223.39 0.38

–2.69 –0.91 0.9 3.26 5.17 8.73 10.82 12.74

–3.58 a –1.53 1.55 5.11 a 7.38 a 9.66 a 10.19 a 9.89 a

a Significantly different from 0 at the 1% level. b Significantly different from 0 at the 5% level. c Significantly different from 0 at the 10% level.

In Table II, we report the results of two distinct LR tests to see whether the two principal criteria we have focused on are of influence or not. Two factors are detected: cuisine is, fortunately, the most influential, but the grades of the GM’s experts do also depend on setting (cellar, service and cadre). Thus, chefs have not one but two strategies to rise in the gourmet cuisine hierarchy. Looking more closely at the details, unsurprisingly, we see that the “superlatives” (laudatory expressions plus appreciated courses) are positively and significantly correlated with the grade. Assessments also vary according to the chef’s style(s) of cuisine. Diversification is preferred to specialisation, as suggested by the coefficient of CLRECR, which is positive and significant at the 1% level. The GM guidebook seems to appreciate, as well, the chef’s technicality. Good technical skill (TECH+) contributes, ceteris paribus, to a significant improvement in the grade. Strangely, criticisms of technicality (TECH–) as well as of quality of courses (CRIT, TECH–, ING–) have no apparent impact on the grade level. In this context, criticisms seem to be only a warning for the chef about the shortcomings in his cuisine. GM experts seem to react positively to flavour combinations (FLAV+), delicacy (DELIC), simplicity (SIMPL) and positive sophistication in preparation (SOPH+). Unexpected signs are obtained for the coefficient of GEN+ and FLAV–. Indeed,

136

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

Table II. Log Likelihood Ratio tests for the degree of significance of the two alternative strategies Full regression Log likelihood value

–223.39

No cuisine Log likelihood value χ 2 test (33 d.f.) χ 2 tabulated value (5% level)

–336.4 226.03 47.39

No setting Log likelihood value χ 2 test (1 d.f.) χ 2 tabulated value (5% level)

–236.88 26.98 3.84

generosity of courses acts negatively but only at a 10% level, whereas bad flavour combination has a positive impact on the grade at a 5% level. These results are probably due to the fact that these criteria appear infrequently in comments (15 and 9 observations respectively). The quality of ingredients and the use of rare foodstuffs have no influence on the grade. On the other hand, a remarkable menu seems to be appreciated by the GM experts. Finally, other elements (those included in the vector X) appear to discriminate between lower and higher grades. First, new entrants (NEW) obtain, ceteris paribus, a lower grade as if GM’s experts are waiting for them to consolidate their gains in the future. Second, GM seems to reward cuisine that is “personified”, as indicated by the level of significance obtained for NAME, i.e. for the fact that the name of the chef is mentioned in the comment. Finally, for either a cook or a restaurant, being part of gourmet cuisine history has a positive and significant impact on GM grades. 4. Conclusion According to GM, only cuisine is taken into account when evaluating restaurants: “Rating ranging from 11 to 19 evaluates the quality of the cuisine only” (GM guidebook, 2000, p. 42). Our results suggest the contrary: chefs have two strategies to get a good grade in this guide – a “cuisine” strategy and a “setting” strategy. It is worth noting, however, that the former does prevail over the latter. In the context of the first strategy, the chef’s know-how and creativity are crucial elements, whereas in the context of the second strategy it is only his capacity to invest in setting and service that matters. The setting (cellar, cadre, service) obviously carries weight but less than art of cooking (the chef’s technical skill, flavours combination skill, his various cooking styles, his attractive menus, the delicacy,

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

137

sophistication and simplicity of his preparations). These results are similar to those of Gergaud and Vignes (2000) for the champagne market, in which they showed that firms invest in reputation either by producing a wine of high quality or by means of public recognition through advertising. This result might explain why in recent years chefs have tended to invest more and more in their setting. Many examples are illustrative: Georges Blanc (23 million euros invested in Vonnas), Marc Veyrat (10 million euros invested in Annecy and Megève), Bernard Loiseau (8 million euros invested in Saulieu). The most extreme case is perhaps Pierre Gagnaire in Saint Etienne; he was compelled to cease his business because of debts that were too heavy. In this paradoxical context, a restaurant offering first-rank cuisine will not reach the top of the ladder if its employees do not have a professional attitude, if the cellar is not well-provided, or if tablecloths are spotted and not ironed. But, a chef investing only in setting and atmosphere while neglecting his cuisine has little chance to be recognised as a high-class cook by experts or peers. In economic terms, it may suggest that a mixed strategy is more efficient than any pure strategy.17 All this could indicate that the relationship between these two strategies is nonlinear: the payoff to non-food strategies probably depends on the level to which the chef aspires. A second step in this analysis would be to check this point by estimating a Translog production function to see whether the two strategies we have detected are substitutable or complementary. Acknowledgements Very honoured to have been awarded the Presidents’ Prize at the Rotterdam 2002 ACEI Conference, the authors would like to thank the “gurus” of the Scientific Committee for this decision. They would also like to express their acknowledgements to C. Barrère, T. Coupé, S. Creigh-Tyte, V. Ginsburgh, J.M. Schuster and to the anonymous referee for their helpful comments. The authors alone are responsible for the remaining errors.

138

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

Appendix: Descriptive Statistics Table A.I. Variables

Abbreviations

Modalities

Descriptive statistics

Experts grade (GaultMillau)

GRADE

Continuous

Mean = 14.88 Std. deviation = 1.72 Min. = 11; Max. = 19

SUP1_5 SUP5_10 SUP10_15 SUP15

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

32.97% (61 obs) 50.27% (93 obs.) 12.43% (23 obs.) 4.32% (8 obs.)

– Number of criticisms:

CRIT

Yes = 1; No = 0

16.76% (31 obs.)

– Style of cuisine: Not indicated Only regional Only classical Only creative Classical and regional Classical and creative Regional and creative Classical, regional and creative

NSTYL REG CLAS CREA CLRE CLCR RECR CLRECR

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

20.54% (38 obs.) 9.19% (17 obs.) 14.05% (26 obs.) 15.68% (29 obs.) 10.27% (19 obs.) 14.05% (26 obs.) 8.65% (16 obs.) 7.57% (14 obs.)

Technical comments: – Good technicality (cooking, taste, . . . ) – Criticisms of technicality

TECH+ TECH–

Yes = 1; No = 0 Yes = 1; No = 0

57.84% (107 obs.) 14.59% (27 obs.)

FLAV+ FLAV– DELIC SIMPL ART SUB

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

24.32% (45 obs.) 4.86% (9 obs.) 7.57% (14 obs.) 20% (37 obs.) 4.86% (9 obs.) 11.89% (22 obs.)

SOPH+ SOPH–

Yes = 1; No = 0 Yes = 1; No = 0

3.24% (6 obs.) 1.62% (3 obs.)

GEN+ GEN– MENU

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

8.11% (15 obs.) 1.08% (2 obs.) 15.68% (29 obs.)

ING+ ING–

Yes = 1; No = 0 Yes = 1; No = 0

40% (74 obs.) 3.24% (6 obs.)

Cuisine General comments: – Number of superlatives: [5; 10[ [10; 15[ ≥15

[1; 5[

Various comments: – Flavour combination: Good Bad – Delicacy of courses – Simplicity of courses – Artistic design of courses – Subtlety of courses – Sophistication: Useful Useless – Generosity of courses: Emphasised Lacking – Nice menu (rich, diversified, . . . ) Ingredients: – Outstanding quality – Bad quality

139

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

Table A.I. (Continued) Variables

Abbreviations

Modalities

Descriptive statistics

– Rare foodstuffs: Fish (wolf fish, . . . ) Crustaceans (crawfish, . . . ) Caviar Foie gras Truffles – Cheese

FISH CRUS CAV FOIE TRUF CHEESE

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

55.14% (102 obs.) 46.49% (86 obs.) 4.86% (9 obs.) 33.51% (62 obs.) 17.3% (32 obs.) 3.24% (6 obs.)

Setting and Serving Michelin: – Number of forks and spoons

F&S

Cardinal scale

– Red forks and spoons

RED

Yes = 1; No = 0

Mean = 2.97 Std. deviation = 0.94 Min. = 1; Max. = 5 25.95% (48 obs.)

GaultMillau:

SETGM

Cardinal scale

Mean = 2.15 Std. deviation = 0.38 Min. = 1; Max. = 3

Champ´erard: – Number of candles

CAND

Cardinal scale

– Number of trays

TRAY

Cardinal scale

Mean = 2.47 Std. deviation = 0.68 Min. = 1; Max. = 4 Mean = 2.40 Std. deviation = 0.64 Min. = 1; Max. = 4

Other Variables – Newly referenced restaurant – History of gastronomy – Chef’s name

NEW HIST NAME

Yes = 1; No = 0 Yes = 1; No = 0 Yes = 1; No = 0

4.86% (9 obs.) 9.19% (17 obs.) 81.08% (150 obs.)

Notes 1. Gourmet cuisine can be viewed as the highest segment of the catering market. According to the French National Institute of Economic and Statistical Information (Insee, 1999), the catering market consists of more than 100,000 businesses, has a turnover of 17 billion euros, and annually provides more than 2.5 billion meals (see Chossat, 2001 for more details). 2. See also Ginsburgh (2002) for a more global view. 3. The two concepts of search good and experience good were introduced by Nelson (1970, 1974). The notion of credence good was introduced later by Darby and Karni (1973). Previously, Lancaster (1966) considered a simple meal as a bundle of two kinds of characteristics – nutritional and aesthetic – and a dinner party as a mix of these two characteristics plus an additional intellectual characteristic, connected to the social setting. 4. On average, the simple correlation coefficient between price and ratings is 0.63.

140

VÉRONIQUE CHOSSAT AND OLIVIER GERGAUD

5. In France, a distinction can be made between the leader – Michelin (650,000 copies sold in 2000), the most famous, and the oldest – and its main followers, namely GaultMillau, Champérard, and Bottin Gourmand. 6. Initiated between 1803 and 1812 by the “Almanach des Gourmands” by Grimod de la Reynière, the principles of gastronomic selection waited for the publication of the Michelin guide to perpetuate the “étoiles de bonne table” [Michelin stars] (in force since 1933). 7. Since its 2000 edition, three (short) lines of comments that mainly focus on the restaurant setting and history have been added for each restaurant. 8. See the appendix for the descriptive statistics for our data. 9. In doing so, we implicitly assume that the GM experts always assess restaurants by referring to a specific and stable set of criteria. 10. Criticised courses are accounted for in another category, see below. 11. SET has been generated from the following equation: 0.70711 WINE + 0.70711 SETGM. SET is an ordered variable given that WINE and SETGM are ordered. 12. This statistical link suggests a general convergence of the main experts’ assessments on the look of the venue. This is probably due to the fact that the setting is assessed mainly from objective criteria like the number of waiters, etc. 13. According to Angrist and Krueger (2001), “Using a linear regression for the first-stage estimates generates consistent second-stage estimates even with a dummy endogenous variable”. Therefore, it is not necessary here, and may even do some harm, to use ordered logit to generate first-stage predicted values for SET. 14. Even if R-squared is higher (0.569) when the regression includes CAND and TRAY. 15. However, the results are quite comparable using SET instead of SETH and when SETH is fitted from an auxiliary regression including CAND and TRAY in addition to F&S and RED. 16. Similar results are obtained with a stepwise ordered logit regression (whether the procedure is backward or forward). 17. The costs associated with each strategy (“cuisine”, “setting” and “cuisine/setting”) are crucial elements that should be taken into account to carefully determine the best performing strategy.

References Albrecht, J., Lang, H., and Vroman, S. (2002) “The Effect of Information on the Well-Being of the Uninformed: What’s the Chance of Getting a Decent Meal in an Unfamiliar City?”. International Journal of Industrial Organization 20 (2): 139–162. Angrist, J.D. and Krueger, A.B. (2001) “Instrumental Variables and the Search for Identification: From Supply and Demand to Natural Experiments”. Journal of Economic Perspectives 15 (4): 69–85. Ashenfelter, O. and Jones, G. (2000) “The Demand for Expert Opinion: Bordeaux Wine”. Cahier Scientifique de l’Observatoire des Conjonctures Vinicoles Européennes, 3. Banerjee, A.V. (1992) “A Simple Model of Herd Behavior”. Quarterly Journal of Economics 107 (3): 797–817. Becker, G.S. (1991) “A Note on Restaurant Pricing and Other Examples of Social Influences on Price”. Journal of Political Economy 99 (5): 1109–1116. Card, D. (1999) “The Causal Effect of Education on Earnings”, in O. Ashenfelter and D. Card (eds.), Handbook of Labor Economics. North Holland, Amsterdam, pp. 1801–1863. Chossat, V. (2001) Les processus de sélection sur un marché, le cas de la grande cuisine. Thèse pour le Doctorat en Sciences Economiques, Université de Reims Champagne-Ardenne. Darby, M. and Karni, E. (1973) “Free Competition and the Optimal Amount of Fraud”. Journal of Law and Economics 16: 67–88.

EXPERT OPINION AND GASTRONOMY: THE RECIPE FOR SUCCESS

141

Favereau, O. and Lazega, E. (2002) Conventions and Structures in Economic Organization: Markets, Networks and Hierarchies. Edward Elgar Press. Gergaud, O. and Vignes, A. (2000) “Emergence et dynamique du phénomène de réputation. Le vin de Champagne: entre savoir-faire et faire savoir”. Revue d’Economie Industrielle 91 (1): 55–74. Ginsburgh, V. (2002) “Award, Success and Aesthetic Quality in the Arts”. Manuscript, ECARES, Université Libre de Bruxelles. Ginsburgh, V. and van Ours, J.C. (2002) “Expert Opinion and Compensation: Evidence from a Musical Competition”. American Economic Review, forthcoming. Greene, W.H. (2000) Econometric Analysis, 4th edn. Prentice Hall International Editions. Hausman, J.A. (1978) “Specification Tests in Econometrics”. Econometrica 46: 1254–1270. Institut National de la Statistique et des Etudes Economiques (1999) La France des Services, Département des activités tertiaires. Lancaster, K.J. (1966) “A New Approach to Consumer Theory”. Journal of Political Economy 74: 132–157. Nelson, P. (1970) “Information and Consumer Behavior”. Journal of Political Economy 78: 311–329. Nelson, P. (1974) “Advertising as Information”. Journal of Political Economy 82: 728–754. Wijnberg, N.M. (1995) “Selection Processes and Appropriability in Art, Science and Technology”. Journal of Cultural Economics 19 (3): 221–235.