By Carly Shin, The George Washington University.
This paper investigates the relationship between expert ratings and the restaurant market. Specifically, this paper aims to broaden our understanding of the experience goods markets by looking at the effects of both having an expert-awarded Michelin star and earning or losing a Michelin star on New York City restaurant prices. The Michelin Guide is one of the world’s leading expert restaurant review sources. Therefore, it is important to investigate the influence that the Guide can have on consumer behavior in order to gain a better understanding of consumption decisions in the context of asymmetric markets. By using a hedonic pricing model, the results show that there is a significant effect of having a Michelin star on market price. In other words, restaurants that are distinguished by a Michelin star correspond to a higher willingness to pay from consumers. Therefore, these restaurants can afford to raise their prices above those of their competitors. A second hedonic model is designed to test for the effect of earning or losing a Michelin star. However, more data is needed to demonstrate if earning or losing a Michelin star can have a similar effect.
1. Introduction
When studying consumer behavior, one common concern is information asymmetry. Information asymmetry exists when one party in a transaction has more information than the other party. This condition becomes especially problematic when pertaining to the quality of a product, such as with experience goods. In these cases, the consumer does not know the true quality of a product until after the transaction has taken place (Akerlof, 1970; Cardebat et al., 2014; Nelson, 1970). As a consequence, when information asymmetries increase, consumers are often guided by expert opinions in their consumption decisions (Flanagin & Metzger, 2013).
This paper will attempt to broaden our understanding of the experience goods market by examining the effect of expert opinion on restaurant pricing. By using the Michelin Guide, one of the world’s leading expert restaurant review sources, this paper will examine if having an award in the Michelin Guide influences the restaurant’s menu prices. This is an important question to investigate as it can allow us to have a better understanding of asymmetric markets and the considerable influence that expert reviews can have on consumer behavior. The restaurant market exhibits a high level of variation in prices. Also, the evaluation of a restaurant involves a high level of subjectivity. Due to these uncertainties, an expert’s opinion may have substantial influence on the consumer.
The present study investigates two research questions. First, what is the effect of having a Michelin star on restaurant prices? Second, what is the effect of earning or losing a Michelin star on restaurant prices? This study hypothesizes that restaurants that are distinguished by experts will charge higher prices than restaurants that have similar characteristics but do not have the same expert recognition. Furthermore, this study hypothesizes that earning or losing this expert recognition will have a similar effect on market prices.
2. Literature Review
There are a few previous studies that look specifically at the effects of earning a Michelin star on the restaurant market. One of these studies serves as the main inspiration for this paper: Gergaud et al. (2006) studies the effect of expert reviews on restaurant prices in Paris, France. This study concludes that earning a Michelin star in Paris not only generates a price premium for the rewarded restaurants of about 25%, but also generates a price premium for surrounding restaurants in the area of about 5% to 13%. This paper expands upon these results to investigate if there is a similar price premium in the New York City market.
The only study to look at the effect of the Michelin Guide in New York is Gergaud et al. (2015), which examines the effects of expert reviews on perceived product quality. This study looks at the introduction of the Michelin Guide in New York City in 2005 to examine whether the availability of expert reviews affect consumers’ opinions along with pricing structure. The authors concluded that the introduction of the Michelin Guide not only generated a price premium effect of about 30% for the Michelin-starred restaurants but also a surge in food, decor, and service consumer reviews.
Although this paper and the Gergaud et al. (2015) study appear to share the same research question, there are a few key differences. First, the Gergaud et al. (2015) study uses a sample of all restaurants in Zagat’s New York City Guide, while this study restricts the sample to restaurants that have been reviewed only in the Michelin Guide. Therefore, the results of the Gergaud et al. (2015) study find increased prices for restaurants that have earned an award in the Michelin Guide as compared to all restaurants in Zagat’s New York City Guide. The current study presents a better controlled experiment because this study hypothesizes an increase in prices for restaurants that have earned an award in the Michelin Guide compared to all other restaurants that are already reviewed in the Michelin Guide. Restricting the sample to only Michelin Guide-reviewed restaurants eliminates any biases that can be attributed to Michelin Guide experts, such as preferences for non-food attributes (Chossat & Gergaud, 2003). Therefore, all characteristics confounded with being included in the Michelin Guide are held constant.
The current study also uses a different testing method while examining the effect of earning a Michelin star. The 2015 Gergaud et al. study uses an ordered variable treatment. Specifically, this is a variable that takes on the value of 0 if the restaurant is not in the Michelin Guide, 1 if the restaurant is in the Guide but has zero stars, 2 if the restaurant is in the Guide and has one star, 3 if two stars, and 4 if three stars. Therefore, this variable assumes that there is an equivalent effect on price regardless of if the restaurant holds one star or three stars. The current study utilizes three separate dummy variables to test for the individual effects of having one star, two stars, and three stars. Because it is considered to be a greater honor to have three stars than it is to have one star, it is possible that there is an unequally larger price premium effect for restaurants with a greater number of stars.
The present study also provides a better control for neighborhood. The Gergaud et al. (2015) study controls for neighborhood by using a spatial measure based on the distance between each Michelin Guide-reviewed restaurant. Specifically, two restaurants are considered to be in the same neighborhood if the distance between the two restaurants is less than ten kilometers. In contrast, the current study controls for neighborhood by using the commonly held definitions of New York City neighborhoods. Using this definition of neighborhood will serve as a better control because the neighborhoods have been naturally divided based on common characteristics. New York City holds a large variety of environmental attributes. Therefore, even though a restaurant is within ten kilometers of another restaurant, it is possible that these two restaurants are located in neighborhoods with significantly different demographics. Using the commonly accepted neighborhood definitions will better control for the environment in which a restaurant is located.
Other studies on the restaurant market have proven that critic ratings have a significant effect on meal prices. For example, Fogarty’s 2011 study on the Australian restaurant market shows that expert reviews from The Age of Good Food Guide and The Sydney Morning Herald Good Food Guide are significant determinants of restaurant prices. The influence of expert reviews has also been studied in markets beyond the culinary field. Examples include studies on the wine market (Ashton, 2016; Cardebat et al., 2014; Dubois & Nauges, 2010; Hadj Ali et al., 2008; Lester et al., 2011), the hotel market (Schamel, 2012), the Cuban cigar market (Livat & Vaillant, 2006), and art auctions (Bauwens & Ginsburgh, 2000; Ekelund et al., 1998). All of these studies support a price premium effect for products that have been distinguished by experts.
3A. Data: The Effect of Having a Michelin Star
To test for the effect of having a Michelin star, this paper uses data from two restaurant review guides, the 2011 Michelin Red Guide and the 2012 Zagat Guide. Both of these guides are released in October of the previous year; the 2011 Michelin Guide was released in October 2010 and the 2012 Zagat Guide was released in October 2011. Therefore, in order to capture any effects of the 2011 Michelin reviews, it is necessary to use the 2012 Zagat Guide. This is because the 2012 Zagat Guide will collect their data within the time frame of October 2010 to October 2011, which is the time period right after the 2011 Michelin Guide reviews are released (See Figure 1).
The Michelin Red Guide is an annually published hotel and restaurant reference guide. The first edition was published in France in 1900 by brothers Édouard and André Michelin. While the original publications were intended as a road guide for motorists with the aim of increasing the demand for cars, the Michelin Guide gained popularity throughout Europe and eventually began reviewing fine-dining restaurants in 1926 (Michelin Guide, 2017). Today, the Michelin Guide’s star ranking system is regarded as one of the most distinguished and selective determinants of fine-dining status. With publications in over 90 countries, many restaurateurs are mindful of the effect that earning a place in the renown guide can have. Some consider earning a Michelin star to be the highest level of achievement that a chef can earn. In addition, earning a Michelin star is almost always accompanied by a significant boost in business.
The Michelin Guide uses a star system to signal quality. A restaurant can be rewarded zero, one, two, or three stars. This system is described as follows (The Michelin Guide, 2010):
The Michelin Guide also gives a Bib Gourmand award to restaurants that are considered to be “inspectors’ favorites for good value,” (The Michelin Guide, 2010). Restaurants that are awarded the Bib Gourmand status generally have a lower price range than Michelin-starred restaurants. Therefore, no restaurant holds a Bib Gourmand award and a Michelin star at the same time. Michelin inspectors, all of whom are considered culinary experts, remain anonymous and pay their own bills when reviewing restaurants. Furthermore, the star reviews are based on food quality alone. Michelin claims that no other factors, such as decor or service, are considered when awarding stars.
The Michelin Guide also indicates other characteristics of restaurants, such as comfort level (on a scale of 1 to 5), whether there is outdoor dining, and whether restaurants take cash only. This paper includes the following controls: cash only, brunch, outdoor dining, notable wine list, notable sake list, notable cocktail list, small plates, and a “more pleasant” symbol that represents “a particularly charming spot with unique décor and ambiance,” (The Michelin Guide, 2010).
This study uses 486 restaurants from the Michelin Guide. These restaurants represent 41 different types of cuisine in 16 neighborhoods of New York City (See Table A1 and Table A2). 51 of the 486 restaurants in this sample have a Michelin star. Specifically, 38 restaurants have one star, 8 restaurants have two stars, and 5 restaurants have three stars. Furthermore, 64 restaurants hold the Bib Gourmand award.
The Zagat Guide is a consumer-based restaurant review guide. It was founded in New York in 1979 by Nina and Tim Zagat. The reviews are established “based on public opinion surveys of regular restaurant goers,” with the quantifiable scores representing an average of all consumer reviews (Zagat, 2011). In the 2012 Zagat Guide, 41,604 respondents rated 2,111 restaurants. Furthermore, the restaurants with low response rates are indicated in order to signal less reliable information.
This paper uses four measures of interest from Zagat. First, Zagat lists the estimated cost of a single dinner with one drink and tip included. In addition, consumers rate the restaurants’ food, decor, and service quality on a scale of 0 to 30. The scale is described as follows (Zagat, 2011):
This sample includes all restaurants in the 2011 Michelin Guide that have a corresponding review in the 2012 Zagat Guide and have not been indicated as a restaurant with a low response rate (the Guide does not specify what is considered to be a low response rate). The range of sample restaurant prices is very large, from $16 to $585. Also, it is important to note that none of the restaurants in the sample has a food rating lower than 17, a decor rating lower than 5, or a service rating lower than 12. No restaurant has ever received a 30 in any of these categories (See Table 1).
3B. Data: The Effect of Earning or Losing a Michelin Star
When testing for the effect of earning or losing a Michelin star, this study uses data from the 2009, 2010, and 2011 Michelin Guides. The pricing data that correspond to these Michelin Guides are from the 2010, 2011, and 2012 Zagat Guides respectively. The data include 417 restaurants from the 2009 Michelin Guide and 482 restaurants from the 2010 Michelin Guide. When added to the data from the 2011 Guide, there are a total of 1,385 observations.
Although the data consist of a large sample size, there are very few restaurants that exhibit a change in the number of Michelin stars over this time period. Specifically, there are 9 restaurants with available data that have either earned or lost a Michelin star between 2009 and 2011. 4 of these restaurants have gained a Michelin star (3 restaurants went from zero to one star; 1 restaurant went from two to three stars) and 5 of these restaurants have lost a Michelin star (2 restaurants went from two to one star; 3 restaurants went from one to zero stars). Due to the small sample, the results for this model are expected to be insignificant.
4. Empirical Models
This paper uses a model based on Rosen’s 1974 hedonic pricing model. The hedonic structure assumes that the price of a product is the sum of all individual attributes of that product (Rosen, 1974). Therefore, the price should change as the attributes of that product change. The log-linear model used in this paper is as follows:
Model 1: ln(cost)i= α0 + α1onestari+ α2twostarsi+ α3threestarsi+ βXi + εi,
where ln(cost) is the logarithm of the cost of a dinner, as taken from the 2012 Zagat Guide, onestar is a binary variable that takes on the value of 1 for all restaurants that have one Michelin star in the 2011 Michelin Guide, 0 if otherwise, twostars is a binary variable that takes on the value of 1 for all restaurants that have two Michelin stars in the 2011 Michelin Guide, 0 if otherwise, threestars is a binary variable that takes on the value of 1 for all restaurants that have three Michelin stars in the 2011 Michelin Guide, 0 if otherwise, X is a variable that represents all other restaurant characteristics that are controlled for.
The dependent variable in this model is ln(cost). The three variables of primary focus are onestar, twostars, and threestars. Furthermore, the restaurant characteristics that are controlled for in this model include, food rating, decor rating, service rating, Bib Gourmand status, neighborhood, type of cuisine, brunch, outdoor dining, notable wine list, notable sake list, notable cocktail list, and small plates.
Based on the hypothesis that Michelin-starred restaurants charge a price premium, the coefficients for onestar, twostars, and threestars are expected to be positive. Furthermore, this study runs three regressions based on Model (1). The first regression excludes all neighborhood and cuisine characteristics. The second regression includes neighborhood data only, and the third controls for both neighborhood and type of cuisine. These separate regressions are run in order to see if controlling for neighborhood and type of cuisine are significant determinants of the expected price premium.
The model used to test for the effects of earning and losing a Michelin star is as follows:
Model 2: ln(cost)i,t+1= α0 + α1onestarit+ α2twostarsit+ α3threestarsit+ α4starincreaseit+ α5stardecreaseit + α6(onestarit x starincreaseit)+ α7(onestarit x stardecreaseit)+ α8(twostarsit x starincreaseit)+ α9(twostarsit x stardecreaseit)+ α10(threestarsit x starincreaseit)+ βXit +εit
where starincrease is a binary variable that takes on the value of 1 for all restaurants that have gained one star from the previous year, 0 if otherwise, stardecrease is a binary variable that takes on the value of 1 for all restaurants that have lost one star from the previous year, 0 if otherwise, X is a variable that represents all other restaurant characteristics that are controlled for.
This model uses the same critical variables as the first model. However, this model includes time fixed effects in order to control for any changes within the three-year time period. Furthermore, this model includes five interaction variables to individually test for the effects of earning and losing a Michelin star for every star-change combination. For example, the coefficient of (onestar x starincrease) will represent the effect of going from 0 to 1 star on restaurant prices. Due to the small sample size, only (onestar x starincrease), (onestar x stardecrease), and (threestars x starincrease) are used in this study.
5. Results
Table 2 presents the results of Model (1). As seen in the table, there is a significant effect of having a Michelin star on price. This effect is apparent without controlling for neighborhood or type of cuisine. However, once these two characteristics are controlled for, the results for onestar, twostars, and threestars are all significant at the 99% level. Specifically, these results show that having one Michelin star increases price by 14.8%, having two Michelin stars increases price by 55.1%, and having three Michelin stars increases price by 80.2%. Furthermore, the results for onestar, twostars, and threestars support the hypothesized non-linearity of the effect of earning a Michelin star, as the effects of having one star, two stars, and three stars are not equivalent. In addition, the test statistics for these three variables are statistically significant.
The results of the third regression imply that neighborhood and cuisine characteristics do play a significant role in how restaurant prices are determined. This model shows that some of the other control variables are significant as well. Food rating, décor rating, and service rating all prove to be significant determinants of price. All three of these variables have a positive coefficient, indicating that a one-point increase in rating leads to a positive increase in restaurant price. However, these three effects are relatively small, ranging from a 2.2% to 3.3% price increase. Finally, the Bib Gourmand status is also a significant determinant of price point. Specifically, having a Bib Gourmand award corresponds to a 7.9% decrease in restaurant prices. This finding supports that being recognized by the Bib Gourmand award does signal that the restaurant prices are a “good value,” as claimed by the Michelin Guide. The other control variables, including brunch, outdoor dining, notable wine list, notable sake list, notable cocktail list, and small plates, all showed insignificant results.
The results for Model (2) were insignificant (See Table A3).
6. Summary and Conclusions
The results of this study support that expert opinion does have an influence on market prices. Restaurants that are distinguished by a Michelin star correspond to a higher willingness to pay from consumers. Therefore, these restaurants can afford to raise their prices above those of their competitors. Furthermore, this price premium effect varies depending on the number of Michelin stars that a restaurant has. The price increase for a restaurant with three Michelin stars is expected to be 65.4% higher than the increase for a restaurant with one Michelin star. These results are analogous to those of both the Gergaud et al. (2006) study and the Gergaud et al. (2015) study.
As expected, the results for Model (2) were insignificant. This is due to the data’s small sample size. In order to achieve a significant estimate for this model, more data is necessary. For example, it is possible that including more Michelin Guides in the data set would yield a significant result. This study only uses three Michelin Guides over a three-year time period. Therefore, if the time period were expanded, there would be more restaurants included with observable changes in their number of Michelin stars. Alternatively, this model could be estimated for data from a different location, such as France. There are more Michelin stars awarded annually in France compared to New York City. Therefore, this change in location could potentially produce a significant estimate. These are important considerations to take into account when studying the influence of expert opinion in the future.
Appendix
References
Akerlof, George A. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics, 84(3): 488-500.
Ashton, Robert H. 2016. “The Value of Expert Opinion in the Pricing of Bordeaux Wine Futures.” Journal of Wine Economics, 11(2): 261-288.
Bauwens, Luc and Victor A. Ginsburgh. 2000. “Art Experts and Auctions: Are Pre-Sale Estimates Unviased and Fully Informative?” Louvain Economic Review, 66: 131-144.
Cardebat, Jean-Marie, Jean-Marc Figuet, and Emmanuel Paroissien. 2014. “Expert Opinion and Bordeaux Wine Prices: An Attempt to Correct Biases in Subjective Judgments.” Journal of Wine Economics, 9(3): 282-303.
Chossat, Véronique, and Olivier Gergaud. 2003. “Expert Opinion and Gastronomy: The Recipe for Success.” Journal of Cultural Economics, 27: 127-141.
Dubois, Pierre and Celine Nauges. 2010. “Identifying the Effect of Unobserved Quality and Expert Reviews in the Pricing of Experience Goods: Empirical Application on Bordeaux Wine.” International Journal of Industrial Organization, 28(3): 205-212.
Ekelund, Robert B., Rand W. Ressler, and John Keith Watson. 1998. “Estimates, Bias and ‘No Sales’ in Latin-American Art Auctions, 1977-1996.” Journal of Cultural Economics, 22(1): 33-42.
Flanagin, Andrew J., and Miriam J. Metzger. 2013. “Trusting expert- versus user-generated ratings online: The role of information volume, valence, and consumer characteristics.” Computers in Human Behavior, 29(4): 1626-1634.
Fogarty, James Joseph. 2011. “Expert opinion and cuisine reputation in the market for restaurant meals.” Applied Economics, 44: 4115-4123.
Gergaud, Olivier, Linett Montano Guzman, and Vincenzo Verardi. 2006. “Stardust over Paris Gastronomic Restaurants.” Journal of Wine Economics, 2(1): 24-39.
Gergaud, Olivier, Karl Storchmann, and Vincenzo Verardi. 2015. “Expert Opinion and Product Quality: Evidence from New York City Restaurants.” Economic Inquiry, 53(2): 812-35.
Hadj Ali, Héla, Sébastien Lecocq, and Michael Visser. 2008. “The Impact of Gurus: Parker Grades and En Primeur Wine Prices.” The Economic Journal, 118(529): F158-F173.
Lester, M. K. Kwong, Don Cyr, Joseph Kushner, and Tomson Ogwang. 2011. “A Semiparametric Hedonic Pricing Model of Ontario Wines.” Canadian Journal of Agricultural Economics, 59: 361-381.
Livat, Florine and Nicolas G. Vaillant. 2006. “Expert Opinion and Brand Reputation: An Analysis from a French Cuban Cigars Guidebook.” Applied Economics, 13(2): 97-100.
Michelin Guide. 2017. Michelin. Retrieved from https://www.guide.michelin.com (accessed September 2017).
The Michelin Guide: New York City Restaurants 2011. 2010. Clermont-Ferrand, France: Michelin.
Nelson, Phillip. 1970. “Information and Consumer Behavior.” Journal of Political Economy, 78(2): 311-329.
Rosen, Sherwin. 1974. “Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition.” The Journal of Political Economy, 82(1): 34-55.
Schamel, Guenter. 2012. “Weekend vs. midweek stays: Modelling hotel room rates in a small market.” International Journal of Hospitality Management, 31: 1113-1118.
Zagat: New York City Restaurants 2012. 2011. New York, NY: Zagat Survey.