Literature Review On Correlation Analysis Of Wine Ratings With Wholesale Pricing, Vintage, Variety, And Region

Abstract

Wine reviews, such as those from Wine Spectator and other consumer publications, help drive wine sales. Standardized wholesale “line pricing” from a major wholesale distributor in the Southwest to compare pricing to the ratings published by Wine Spectator were utilized by the researchers in this study and to determine whether there were any correlations among other key attributes of the wine. Interesting results were produced, including that the wholesale price and vintage of a wine are significant in the prediction of the wine’s rating. Consumers have varieties of choice when it comes to the purchase of wine. Wines vary across varieties, regions, years of vintage, alcohol content, price and expert ratings. The relationship between those variables and whether a combination of explanatory variables can predict outcome variables will be exploring in this study.

Introduction

It has been known that there are a huge variety of wine which are produced throughout the world. In 2003, a wine taster was being insured based on her olfactory systems ability to taste wine for £10 million for a wine buyer for a supermarket chain in the UK (Cave, 2003). There is also an owner of a wine yard from the Bordeaux region which insured his nose for $8 million (Seattle Times, 2008). It is therefore required to understand the ways to determine the quality of wine based on its rating which are done by wine testers also known as wine reviewers to understand the fundamentals of the quality of the wine. This is due to the wine reviews greatly impacts on the reputation of the wine manufactured as there are some wine reviewers that are well renowned such as Robert Parker Jr and James Suckling which acts as validators and they are also known to break new records before it comes to market. Therefore, it is important for the manufacturers and the public to understand the ways taken by the wine reviewers to validate the wines which are being rated.

The scope of this literature review will cover the relationship between the types, price, region and the manufacturer of the wines. As an experienced consumable good, the quality of a wine is only known after its consumption. In contrast to consumers, wine producers are informed about their products’ quality. This information asymmetry has led to the emergence of wine experts providing information on wine quality. The contingent information market is particularly well developed in the wine sectors where numerous experts coexist. The subjectivity of the wine quality assessment, the regional segmentations, or their (supposed) preferences (Storchmann, 2012) partly justify a large number of experts. Moreover, the grading systems and habits could differ from one expert to another. In particular, the European experts are used to rating wine on a 20-point scale whilst US experts use 100 points (e. g. , Masset and Weisskopf, 2015). The heterogeneity of the rating systems can increase the consumer’s perceived uncertainty. The question of rating homogenization on the same scale of preferences is therefore at the heart of the uncertainty debate about wine quality. The uncertainty about wine quality is particularly high during the en primeur campaign in the Bordeaux Region.

The primeur market can be seen as a forward market dedicated to fine Bordeaux wines. The en primeur campaign takes place during the spring, starting with a huge multi-day tasting organized by the chateaux in the first week of April. Wine merchants, wine enthusiasts, and of course, wine experts are involved in this event. They all taste the wine from the latest harvest. Therefore, the wine is not yet vinified and the quality assessment is particularly difficult and uncertain. The aim of this campaign is to sell (chateaux) and buy (wine merchants) before the wine is effectively released in bottles, which will happen about 18 months later. The prices and quantities exchanged are determined during the en primeur campaign and the wine will be delivered once it is bottled. Many studies produce data consisting of rankings or ratings of n objects by m judges. Almost every award and admission to a competitive program is based on such data.

In this paper, we are concerned with measuring the quality of rankings and ratings specifically for the judging of wines. Because both rankings and ratings are typically produced using the assessments of two or more judges, their consistency is one measure of the quality of the resulting ranking or rating. We discuss advantages and disadvantages of a variety of measures that have been or could be used for evaluating wine judge consistency. Measuring the consistency of ratings has been studied in different disciplines for decades, and there is a long bibliography on methods. Some of the key descriptive term that apply are: methods of concordance, measures of agreement, multirater data, ranking methods, interrater reliability, paired comparisons, and rater agreement (or judge agreement). Judge agreement, which we use as a generic term, can refer to exact matching for nominal ratings or consistency of the ranking. There is little consensus about which statistical methods are best for measuring judge agreement. The literature has considered wine’s quality, reputation, and price in marketing and economic research.

For example, Landon and Smith (1998) found empirical evidence indicating that a wine’s reputation has a greater impact than its quality on the price of the wine. Roberts and Reagans (2007) studied how critical exposure and ratings affect producer pricing strategies. However, few have researched wine in relation to price and ranking by wine critics. Horowitz and Lockshin (2002) used previously developed wine-quality ratings to predict retail pricing. They found no strong correlation between price and wine-quality rating. However, the influence of the other factors--vintage, variety, and region-- influenced the findings significantly. Hedonic price influences were explored by Combris, Lecocq, and Visser (1997) and Landon and Smith (1998), both of which studies used sensory techniques to attempt to determine the quality of Bordeaux wines. This culminated in their trying to develop a pricing scheme based on these influences. Yet this, too, failed to address the simpler question of whether rating systems currently utilized in popular consumer culture correlate with price.

In economics, an experience good is a product or service whose characteristics, such as quality, are difficult to observe prior to consumption (Nelson, 1970). For certain experience goods, research is ineffective. In assessing a product or service, consumers must instead rely on quality determinants, which include evaluations by product experts. Unlike other products, for which Consumer Reports provides quality assessments, wine is rated for quality by the Wine Spectator, the Wine Advocate, the Wine Enthusiast and the Connoisseur's Guide.

According to research by Roberts and Reagans (2007) on critical exposure and price-quality relationships, when consumers value quality and rely on expert opinion, the price for a product is positively related to its reported quality rating. In their study of the price-quality relationship in a sample of Bordeaux wines, Landon and Smith (1998) found a positive relationship between the quality scores reported in the Wine Spectator and the wines’ prices. A similar result was obtained using Connoisseur’s Guide, by Benjamin and Podolny (1999) in their study of California wines, and by Schamel and Anderson (2003) in their study of wines from Australia and New Zealand using quality ratings from James Halliday and Winestate. In 2007, however, Miller, Genc, and Driscoll examined the wide variance in pricing as related to Wine Spectator ratings of 2001 California Cabernet Sauvignon but did not systematically compare wholesale pricing with Wine Spectator ratings. Presumably, wines have been evaluated since they were first consumed. Over the past 100 years, these evaluations have developed into formal ranking systems that have impacted how wines have been priced and how consumers have accepted these wines. Despite the multitude of informal wine ratings performed by internet bloggers, cooking magazines, mail-order retailers, and other sources, none has been as influential or controversial as the rating systems developed by wine critic Robert Parker and the Wine Spectator. According to Arrow (1974), the strength of the relationship between a wine’s price and its quality rating depends upon the extent to which individuals pay attention to the product and its rating. Recent research found that consumers, particularly at lower retail-price points, do not fully understand how price and ratings can equate to a “quality/value” wine product.

This is particularly so given the disparity in ratings from critique to critique and from brand to brand (Barber, Almanza, & Dodd, 2008). Information about higher quality may increase demand and translate into a higher price. For lower-quality products, greater accessibility reveals negative information, which reduces demand and lowers price (Roberts & Reagans, 2007). The question then becomes whether there are any significant discrepancies between a wine’s price and its rating. A wine’s retail price can vary from one retailer to another, and from one geographic area to the next. A good rating from a major publication can affect the retail price almost instantly. Conversely, wholesale prices are stable within a large distributor network and are less likely to be affected by wine ratings until the winery increases prices to the distributor on future deliveries. Ranking means that the objects are arranged in order of some attribute, from largest to smallest or vice versa. Rating means that objects are sorted into categories with similarly defined attributes. Rankings and ratings are usually based on an underlying score assigned by a judge or judges. The scores can be interval, meaning that the differences in scores are comparable, or ordinal, in which case they are not.

Whether the score is interval or ordinal, and whether ranking or rating is the goal, determines the data analysis needed to assess consistency of judging. Neither rankings nor ratings are meaningful unless there is a minimum level of consistency (or concordance) among judges. A maximum ranking consistency ignores the closeness of the scores and occurs when all judges rank the wines in the same order. It can be achieved even if the judges do not have perfectly consistent rankings but requires the additional condition that judges’ assigned scores are close enough that they fall in the same category. If a set of judges produces rankings that are independent of one another, then the ranks provide no information of the merits of the wines. An exception to this occurs if information on the relative competency of the judges is available or can be discerned from the rankings themselves. Restaurants and other hospitality providers typically purchase wines from a wholesale distributor rather than from a retailer. Therefore, it is more relevant to address the wholesale price in this study. Furthermore, as wholesale purchasing from the producer takes place prior to the release and distribution of the wine, ratings on a particular wine are not typically made; nor do they impact pricing at the wholesale level. It is important to understand that wine producers and wholesalers generally have formed an opinion about product quality when they set their list prices, and there is a connection between their own and the critics’ assessments of quality (Roberts & Reagans, 2007).

According to Lockshin (1993), when a new vintage is released, its wholesale price is generally determined not by its critical rating, but by its taste and/or a negotiation process between the producer and the wholesaler. However, Roberts and Reagans (2007) found that prior reputation--particularly if critic ratings have been issued--can influence the pricing decisions of a current release. This viewpoint was confirmed through discussions with several large wholesale distribution houses, where expert tasting and evaluations are part of the numerous services provided to producers. These wholesalers noted that they have a very good idea of a wine’s quality before they assist a producer in setting its list price.

Following Roberts and Reagans (2007), this concept--that experienced wine tasters tend to reach similar conclusions about quality--was confirmed by examining the quality ratings of a random sample of 50 wines evaluated by both the Wine Spectator and the Wine Advocate in 2005, each using a 100-point rating system. The results showed that 87% of the wines were within two points of one another. This leads to the purpose of this research study, which is to use regression to determine the impact that price has on predicting a wine’s rating. Other variables used in prior research studies, such as vintage, varietal, and country of origin, were also included in the regression model. This study employed available wholesale prices of wines sold in the state of Texas and the scores that those wines received from the Wine Spectator.

Methodology

The relationship between wholesale price and wine-critique ratings was examined. Wholesale pricing data was obtained using the spring and summer 2007 Texas wholesale catalog from the Domaines and Estates division of Glazer’s Wholesale, based in Dallas, Texas. This wholesale catalogue contains hundreds of different wine brands, varieties, and vintages. The names of each of the 853 wines from the catalog were individually entered into the Wine Spectator online wine- rating database (www. winespectator. com) in order to search for its published ratings. When an exact match of the wine brand, varietal, and vintage was found, the corresponding rating and price were recorded, thereby creating an expanded database of the wine’s price, rating, vintage, varietal, and country of origin. For this study, a total of 197 exact matches were found and analyzed. The data were analyzed using statistical procedures, including descriptive statistics, correlation analysis, and multiple regression (SPSS, release 14. 0 and SAS 9. 0 TS level 02M0). The descriptive analysis focused on the wine’s price, rating, vintage, varietal, and country of origin. Correlations were calculated to measure association between the price, vintage, and rating variables.

Finally, multiple regression was used to gain insight into the rating as the dependent variable in relation to price, vintage, varietal, and country of origin in predicting a wine’s rating. The general purpose of multiple regression (Cohen & Cohen, 1983) is to learn more about the relationship between several independent or predictor variables, and a dependent or criterion variable. For the multiple regression, a set of variables, either continuous or categorical, were used as the independent variables. The general regression model for the dependent variable wine rating is written as follows, where Y is the linear combination: Y = β0 + β 1X1 + β 2X2 + β 3X 3 + ··· + β nX n In this symbolic representation Y = wine rating β0 = a constant that can be interpreted as the value of Y when X1 = X2 = ··· = Xk = 0 β1 -β n = estimates of the parameters X1 - Xn = the independent variables of interest. Multiple regression requires a large number of observations. The number of cases (participants) must substantially exceed the number of predictor variables used in regression.

The absolute minimum suggested is five times more cases than predictor variables. A more acceptable ratio is 10: 1(Hair, Anderson, Tatham, & Black (1998). Four main assumptions about the relationships between the dependent and independent variables involved in the multiple regression analysis are as follows: Linearity (using a bivariate scatterplot), Equality of variance or homoscedasticity (scatterplot between each independent variable and the dependent variables), and Independence and Normality (construction of a normal probability plot). These assumptions were tested prior to running the regression analysis (Hair et al. , 1998). Standard multiple regression can accurately estimate the relationship between dependent and independent variables only if the relationships are linear. Regression assumes that variables have normal distributions. Non-normally-distributed variables (highly skewed or kurtosis variables, or variables with substantial outliers) can distort relationships and significance tests. Overall, the products in our sample were from producers who never before had received a single review, and from producers who had had several prior critical reviews. This variance allowed a test of whether the producers’ different histories of ratings influenced the extent to which their product-quality ratings and their reputations were reflected in the prices charged.

This time-rating variable is the number of prior reviews received by the producer up to the year of release, divided by 10, and dates from the first year in which a producer received a review. Other variables used in this study included the country of origin, the vintage, the varietal, the price per bottle, the producer’s age (the current year minus the focal producer’s founding year), the exchange rate (percentage change, relative to 1987, of the value of the focal country’s currency relative to the U. S. dollar), and the log quantity (the natural log of the total number of cases produced). The regression model also included a set of indicator variables for each country, for each varietal, and for each year within the sample period. Because the wines sampled were produced in different countries, exchange rate fluctuations may have affected the prices charged in the U. S. In particular, local currency appreciations relative to the U. S. dollar should be associated with higher local production costs (measured in U. S. dollars) and therefore higher U. S. prices. This was controlled for with a variable that accounted for the percentage change (relative to 2000) of the value of each country’s currency relative to the U. S. dollar. This study considered that producers and distributors have a good idea about product quality, which produces a strong relationship between producers and critics’ assessments of product quality and suggests whether producers tend to become better over time in predicting the reactions of critics to their wines. We dealt with this issue by controlling for total producer experience. Here, we consulted numerous sources to compile information on the founding dates of wine producers. To determine the extent to which the producer’s learning had an impact on the sensitivity of price-to-quality, we included the producer’s age variable and its interaction with the quality variables in the model tested in this study. The age of each wine at release was also controlled for, as wines aged longer tend to command higher prices. For wines produced in different quantities, the price-quantity trade-offs may affect the results (Roberts & Reagans, 2007). The Wine Spectator reported quantity data for 47% of the observations in our sample. We used the log of the number of cases produced (divided by 1,000) for the quantity available for sale in the U. S.