Scholarly article on topic 'Portuguese wine ratings: An old product a new assessment'

Portuguese wine ratings: An old product a new assessment Academic research paper on "Agriculture, forestry, and fisheries"

CC BY-NC-ND
0
0
Share paper
Academic journal
Wine Economics and Policy
OECD Field of science
Keywords
{"Portuguese wines" / Ratings / Consistency}

Abstract of research paper on Agriculture, forestry, and fisheries, author of scientific article — José Caldas, João Rebelo

Abstract Typically, wine is a good experience, its quality being unknown before consumption, with wine drinkers tending to be risk-averse. This being so, expert and consumer opinions may help to fill this information void. Following the classic example of the Bordeaux region increasing numbers of wine producers submit their wines to the raters' evaluation, aware of its importance in consumer purchases and in the definition of wine prices. Portugal, particularly in the last decade, has been following this tendency, with an increasing number of Portuguese wines appearing on the ratings list of Robert Parker (RP) and Wine Spectator (WS) gurus. Using the ratings published in 2010, by RP, WS, João Paulo Martins (Portuguese) and cellartrack.com, this paper aims to assess the consistency between the ratings assigned by different experts and by consumers and, additionally, to determine if the score attained by a specific wine is influenced by colour attributes and/or wine region (geographic origin). A statistical analysis shows that, with minor differences, there is consistency between the different ratings. Furthermore, the results of the regression models indicate that red wines tend to have higher scores and, in geographical terms, Douro wines are in prime position.

Academic research paper on topic "Portuguese wine ratings: An old product a new assessment"

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Wine Economics and Policy 2 (2013) 102-110

www.elsevier.com/locate/wep

Portuguese wine ratings: An old product a new assessment

José Caldas*, Joao Rebelo

Department of Economics, Sociology and Management (DESG) and CETRAD, University of Trás-os-Montes and Alto Douro (UTAD),

5000-660 Vila Real, Portugal

Available online 5 December 2013

Abstract

Typically, wine is a good experience, its quality being unknown before consumption, with wine drinkers tending to be risk-averse. This being so, expert and consumer opinions may help to fill this information void. Following the classic example of the Bordeaux region increasing numbers of wine producers submit their wines to the raters' evaluation, aware of its importance in consumer purchases and in the definition of wine prices. Portugal, particularly in the last decade, has been following this tendency, with an increasing number of Portuguese wines appearing on the ratings list of Robert Parker (RP) and Wine Spectator (WS) gurus.

Using the ratings published in 2010, by RP, WS, Joao Paulo Martins (Portuguese) and cellartrack.com, this paper aims to assess the consistency between the ratings assigned by different experts and by consumers and, additionally, to determine if the score attained by a specific wine is influenced by colour attributes and/or wine region (geographic origin). A statistical analysis shows that, with minor differences, there is consistency between the different ratings. Furthermore, the results of the regression models indicate that red wines tend to have higher scores and, in geographical terms, Douro wines are in prime position.

© 2013 UniCeSV, University of Florence. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Portuguese wines; Ratings; Consistency

Contents

1. Introduction.................................................................................102

2. Literature review..............................................................................103

3. The Portuguese wine sector: an overview.............................................................105

4. Data and analysis.............................................................................106

4.1. Data collection, samples and descriptive statistics...................................................106

4.2. Analysis...............................................................................107

5. Conclusions.................................................................................109

References.....................................................................................109

1. Introduction

Corresponding author.

E-mail addresses: jcaldas@utad.pt (J. Caldas), jrebelo@utad.pt (J. Rebelo). Peer review under responsibility of UniCeSV, University of Florence.

During the last decade, the word rating has become increasingly familiar, whether in relation to sovereign debt, creditworthiness of firms and universities, or wines. In all these situations, it is assumed that ratings are a way of filling consumers' information failures, a question that belongs to the microeconomics topic of adverse selection market failures.

2212-9774 © 2013 UniCeSV, University of Florence. Production and hosting by Elsevier B.V. All rights reserved. http://dx.doi.Org/10.1016/j.wep.2013.11.004

Typically, wine is a good experience, where quality is not recognised before consumption, i.e., until one buys and opens a bottle its content and quality remain unknown. That is why wine consumers count on expert opinions expressed in wine ratings and critical reviews. Internationally, some of the best known experts are Robert Parker (RP) and his publication the Wine Advocate, first published in 1978, and Wine Spectator (WS),1 whose first number came out in 1976. These raters have an enormous influence on both producers and consumers of wine. Both wine raters score the wines from 50 to 100 and a high score (for RP and WS, an outstanding score is from 90 upwards), in general, leads to a higher price and increased sales. On the other hand, a low score can lead to decreased sales and often to a lower price (Thompson and Mutkoski, 2011).

In Portugal there have also appeared several raters and opinion makers, over the last decade, albeit with less renown than the international ones. One of the most prominent experts is Joao Paulo Martins (JPM), with his annual publication Guia dos Vinhos de Portugal, which, in addition to the rating, gives a detailed description of the characteristics of each wine tasted. This rater grades the wines from 14 to 20.

As a complement, and in some ways, as a counter power to raters' influence in consumer decisions, social networks have emerged recently, including the web, in which consumers express their opinions and ascribe their scores, the information included in http://www.cellartracks.com2 being particularly important.

Table 1 presents the scores and rating descriptions for each of the rating sources referred to above.

The aim of this paper is to assess the consistency between ratings assigned by different experts and by consumers and, additionally, to determine if the score attained by a specific wine is influenced by colour attributes and wine region. To achieve this objective the paper is organised as follows: Section 2 includes a literature review; Section 3 presents an overview of the Portuguese wine sector and Section 4 data and results; finally, in Section 5, concluding observations are given.

2. Literature review

In the wine economics field, three major research topics have deserved special attention: finance, climate changes and expert opinion. The research has been extended to subjects related to market regulation, quality signalling and consumer search (Storchmann, 2011).

Typically, wine is a good experience and wine drinkers tend to be risk-averse (Gocekus and Nottebaum, 2011). Hence, reputation is one of the most important factors in overcoming the informational asymmetry problem - adverse selection -associated with this type of products (San Martin et al., 2008).

:As with financial agencies, in the wine industry the best known gurus are also from the USA.

2This website follows a rating criterion identical to that of RP and WS. See Table 1.

Reputation can be influenced in three different ways: (a) producer and/or brand recognition associated with "objective" wine characteristics, such as colour, alcohol content, grape variety, and vintage year; (b) expert opinion, based on the wine's sensory characteristics; and (c) the umbrella brand/ collective reputation, such as region of origin.

The influence of reputation on price has been analysed mainly using the so-called hedonic price functions - following Lancaster's theory on consumer behaviour - where prices are regressed on a set of characteristics in order to determine which of these have a significant effect on prices. The price equation includes the objective characteristics of wine, where reputation is conveyed through the producer and/or brand name to consumers, and sensory characteristics, as judged by expert rankings and opinions (San Martin et al., 2008).

Different papers show the importance of the producer or individual brand on wine prices, namely, Oczkowski (1994) for Australian premium table wines; Combris et al. (1997, 2000) for Bordeaux and Burgundy wines; Schamel and Anderson (2003) for Australian and New-Zealand wines; Luppe and Angelo (2005) for Brazilian, Argentinean and Chilean wines; Haeger and Storchmann (2006) for American pinot noir; Lecocq and Visser (2006) for Bordeaux and Burgundy wines; Hadj Ali and Nauges (2007) for Bordeaux wines; San Martin et al (2008) for Argentinean wines in the US market; and Viana and Rodrigues (2007) for Porto wine. In the case of Portugal, Viana and Rodrigues, drawing on a sample of 14,000 observations from the largest Port wine firms, found that the brand/ type of Port and producer's reputation are significant determinants of wine prices.

The influence of wine expert raters, such as WA and WS, is, generically, seen to be of great importance. Several papers have addressed this influence via the estimation of the so-called hedonic price functions, comparing the effect on prices of official classifications with the impact of raters' classification, and analysing the marginal effect of rater marks. Jones and Storchmann (2001), using data from wine auction prices for 21 prestigious Bordeaux crus classés chateaux wines, conclude that experts' influence is significant, for each additional Parker's point score the price increased, on average, 7%. Schamel and Anderson (2003) show that, for Australian and New-Zealand wines, price premium associated with both James Halliday's and Winestate magazine's sensory quality ratings, and with Halliday's winery ratings and classic wine designations, are highly significant. Dubois and Nauges (2007), using a panel data set of 108 châteaux selling wines on the Bordeaux en primeur market, confirm that experts' grades affect the en primeur price more than the unobserved wine quality. Hadj Ali et al. (2010) estimate that the marginal effect of Parker's scores on en primeur Bordeaux wines, is, on average, 2.80 euro per bottle of wine.

Gocekus and Nottebaum (2011) attempt to shed light on the question of whose rating a regular buyer should pay attention to, comparing the taste of regular consumers with that of experts. From a sample of 120 vintage 2005 Bordeaux wines listed in cellartracker.com, they conclude that both average and median community scores are lower than expert scores, the

Table 1

Raters description of wine quality as related to wine score.

Source: Erobertparker.com (RP); winespectaror.com (WS) and JPM (2011).

Rater Score Description

RP 96-100

90-95 80-89 70-79 60-69

95-100

19-20 17.518.5 16-17 14-15.5 < 14

An extraordinary wine of profound and complex character displaying all the attributes expected of a classic wine of its variety. Wines of this calibre are worth special effort to find, purchase and consume.

An outstanding wine of exceptional complexity and character. In short these are terrific wines.

A barely above average to very good wine displaying various degrees of finesse and flavour as well as character with no noticeable flaws. An average wine with little distinction, except that it is a soundly made. In essence, a straightforward, innocuous wine.

A below average wine containing noticeable deficiencies, such as excessive acidity and/or tannin, an absence of flavour, or possibility dirt aromas or flavours.

A wine demand to be unacceptable. Classic: a great wine.

Outstanding: a wine of superior character and style. Very good: a wine with special qualities. Good: a solid well-made wine.

Mediocre: a drinkable wine that may have minor flaws. Not recommended.

Great wine, of world class. Excellent wine, of great refining.

Good, of strong personality.

Good, well done, very pleasant to drink.

Acceptable, without tasting note.

correlation between the expert ratings is higher than that between the community and expert ratings, and, most interestingly, compared to expert ratings, the average price paid for a bottle of wine is more highly correlated with median community score.

However, authors such Haeger and Storchmann (2006) and Lecocq and Visser (2006) found that expert opinion on prices is less important than the objective characteristics of wine. Haeger and Storchmann (2006), when analysing Pinot Noir price determinants, found that, after temperature and precipitation, producer/winemaker reputation is the most important determinant of wine prices, while expert opinions have little explanatory value. Furthermore, Lecocq and Visser (2006) conclude that the impact on price of expert opinions is very small compared with the objective characteristics appearing on the label, such as vintage and appellation. And Unwin (1999) stressed that the application of hedonic prices related to individual brands and expert ratings seems to be misguided and inappropriate, since most of the variables used are closely interdependent and there is insufficient knowledge on consumers' definitions of wine quality to enable valid conclusions to be drawn from regressions.

The umbrella brand/collective reputation literature provides additional and important insights and conclusions on the impact of reputation on wine prices and consumer's choice. Landon and Smith (1998): 629 state that in a market with a large number of firms, it may be very costly for consumers to acquire information on the past quality of goods produced by all firms [therefore] it is typically less costly for consumers to acquire information on a group/collective quality that can be used as an indicator of the quality of the goods produced by individual

firms in the group. Moreover, for Castriota and Delmastro (2009): 2 the use of a well-known group brand may enable (small) producers to reap the benefits of a reputation rent, without incurring all the costs that a company has to face when it has to establish the reputation of a commercial brand name.

The literature on collective reputation seems to be in its infancy, the topic having received more theoretical attention than empirical (Castriota and Delmastro, 2009; Gergaud and Livat, 2010). The issue of collective reputation is principally modelled using Tirole's concept of group reputation as an aggregate of individual reputations - considering that a group's reputation is only as good as that of its members or, at least, of its most famous members (Gergaud and Livat, 2004). Empirical analysis has been applied to the impact of collective reputation on individual reputation and/or prices, and the spillover effects of collective reputation as a quality signal.

Landon and Smith (1997, 1998), using data from the market for Bordeaux wines, showed that both individual and collective reputations explain a substantial part of price variation and that long-term reputation is considerably more important than short-term quality improvement.

Castriota and Delmastro (2008), using data from wineries located in four regions of North-West Italy with an established national reputation, test the determinants of the process of collective reputation, taking into account the interactions between individual and collective reputation, and the determinants of the "jump" from national to international reputation. They confirm the prediction of the theoretical literature, finding positive effects of collective reputation on the reputation of individual firms. Additional research by the same authors (2009) provides empirical evidence in favour of the positive

effects of minimum quality standards on group reputation. Furthermore, they show that the relationship between group size and collective reputation is non-linear: free entry may be not optimal, since, above a certain number of producers, group reputation can decline due to free-riding behaviour.

Frick (2010), using information published by the wine guide Gault Millau, also found statistically significant non-linear returns to individual reputation as well as statistically significant returns to collective reputation in the case of Mosel Valley wines. Gergaud and Livat (2010), from an application to Bordeaux wines, using detailed survey data collected in seven European countries, obtained positive and significant spillover effects from umbrella reputation (Bordeaux), which are found to increase with the reputation of individual wines. These spillover effects, when significantly positive, vary from a minimum of 5% to a maximum of 15% of additional favourable quality opinions. Schamel (2009), in a study based on wine prices from 27 regions around the world, concludes that wines from producers with a high quality reputation rely more on their own strengths and depend less on their region's reputation.

To sum up, this literature overview suggests that: (a) wine reputation is a means to overcome the informational asymmetry problem of wine consumers; (b) there is no unanimity among researchers on the most important group of variables that influence reputation: objective characteristics, expert opinion and region of origin; (c) reputation seems to be highly positively correlated with wines prices; (d) the judgement of experts, as a source of reputation, influences prices and buyer's decisions (e) the buyer is uncertain on the question of whose rating he/she should pay attention to.

3. The Portuguese wine sector: an overview

Like other southern European countries, Portugal is a traditional wine producer. Vineyards cover (IVV, 2011: 38) 237,786 ha, with almost 41.3% (98,210 ha) considered capable of producing higher quality wines, i.e., with production of denomination of origin (PDO) and/or production of geographic indication (PGI). The total vineyard area is occupied by 341 different varieties, most of them native, allowing Portugal to produce non-standard wines for market niches, which could be a strong point, according the unpublished document prepared by the Monitor Group of Michael Porter.

The wine industry is spread throughout the 11 Portuguese demarcated regions. Table 2 shows the average of wine production of the last five years for each of these regions and for the whole country. Relative to wine production, the Douro region occupies the first place, followed by Lisboa, Alentejo, Beiras3 and Minho. However, by area, the first place is occupied by Beiras, followed by Douro and Minho, Lisboa having highest per ha productivity.

Table 3 shows that, on average,4 about 57% is PDO and PGI, 14.4% fortified PDO and 28.8% table wine, figures that

3The region called Beiras includes the sub-regions of Dao, Bairrada, Beira Interior and others.

By wine colour, on average, 32% is white and 68% is red.

Table 2

Portuguese wine production (hl), by region and vineyards area (ha). Source: Instituto da Vinha e do Vinho - IVV (2011).

Wine region Average 2006-2011 % Area % hl/ha

Minho 841,984 13.02 31,010 13.04 27.15

Tras-os-Montes 132,779 2.05 21,730 9.14 6.11

Douro 1,509,768 23.35 47,035 19.78 32.10

Beiras 890,243 13.77 56,663 23.83 15.71

Tejo 599,874 9.28 24,799 10.43 24.19

Lisboa 1,070,845 16.56 18,743 7.88 57.13

Peninsula de Setubal 399,051 6.17 9210 3.87 43.33

Alentejo 940,873 14.55 23,490 9.88 40.05

Algarve 25,151 0.39 1983 0.83 12.68

Madeira 45,415 0.70 1700 0.71 26.71

Azores 10,122 0.16 1423 0.60 7.11

Total 6,466,105 100.00 237,786 100.00 27.19

are in line with the fact that most of the Portuguese wine production is carried out in demarcated regions. Mono-varietal wines have an insignificant share in Portuguese wine production. Furthermore, among the fortified wines, Port wine deserves special attention, representing (832374 hl), on average, i.e., around 90% of this category of wine.

In 2010, Portuguese (Instituto do Vinho e da Vinha, IVV, 2011) domestic consumption was 4695 thousand hectolitres (for a 2009 production campaign of 5894 thousand hectolitres), imports were roughly 1464 thousand hectolitres (67.7% in bulk) at an average price of 0.57 euro/l, and exports, by wine typology, were as follows: non-fortified wines, 1838 thousand hectolitres (64% in bottle), at an average price of 1.76 euro/l; Port wine, 862 thousand hectolitres (100% in bottle), at an average price of 4.30 euro/l; Madeira wine (fortified), 28 thousand hectolitres, with price identical to that of Port wine.

If both Port and Madeira wines have a long history of more than 200 years of export, with approximately 86% of their production being sold on the international markets, the situation is different for table wines. Until the late 1980s, both imports and exports of table wines were irrelevant, production being intended for domestic consumption and for the brandy5 used to fortify Port wine.

With Portugal's entry into the European Union (1986), a high number of grape-growers developed a strategy of forward vertical integration (Muhr and Rebelo, 2011), producing and bottling their quality table wines under their own labels, rather than selling the grapes to companies and co-operatives. While these new brands found heavy demand in Portugal, their entry into international markets was more difficult, as the Portuguese wine regions were not widely known, and therefore did not represent a category to be included in wine lists and on the shelves. Thus the market strategy followed by these new wine producers is based on differentiation for niche markets, where product promotion is made through marketing events, press releases and interactions with wine experts (such as media, trade and gastronomy) in the target markets. One of the consequences of this entrepreneurial behaviour is the

5550 l (a barrel) of Port wine incorporate, on average, 435 l of must and 115 l of brandy.

Table 3

Wine production by type (1000 hl).

Type of wine 2006-07 2007 08 2008-09 2009 10 2010 11 Average %

Wines from PDO 2360 1874 2013 2132 2454 2167 33.51

Fortified wines from PDO 961 966 964 886 867 929 14.36

Wines from PGI 1737 1516 1297 1261 1691 1500 23.20

Mon-varietal wines 0 0 0 4 26 6 0.09

Table wine 2484 1717 1415 1611 2094 1864 28.83

Total 7542 6073 5689 5894 7132 6466 100.00

emergence, in the last decade, of Portuguese wines in the ratings assigned by experts. One of the paradigmatic cases of this situation is that of Douro region quality table wines, which, from being unknown, have achieved, within a decade, high national and international recognition (Muhr and Rebelo, 2011).

4. Data and analysis

4.1. Data collection, samples and descriptive statistics

Data was collected on ratings from RP, WS, and Cellar via the websites of each rater and from JPM on his annual publication Guia dos Vinhos de Portugal, for the year of 2010.

The publication of JPM presents the scores of about 2000 wines tasted, the Wine Advocate of RP (December 2010) includes the scores of 180 Portuguese wines and the Wine Spectator (April 2010) rates 211 Portuguese wines. In Cellar 5167 wines (18,993 notes) are evaluated.

Comparing the wines classified by both RP and WS publications we observe that a large number of them do not coincide. There are wines classified by one rater but not by the other and vice-versa. In view of this fact, to establish a comparison, two samples were constructed, taking into account only the wines turning up simultaneously in all data sources: (a) the first sample includes 126 wines common to RP (international rating), JPM (national rating) and Cellar (consumers' rating); (b) the second sample includes 111 observations of wines common to WS (international rating), JPM (national rating) and Cellar (consumers' rating). In addition, since WS evaluations go beyond wine scores, also including sales prices, this sample also includes this variable.

Table 4 presents the descriptive statistical measures for the first sample, which includes wines from all 11 wine production regions of Portugal: Douro - 62; Alentejo - 33; other regions6 - 31 (Ribatejo - 11, Estremadura - 7; Dao - 6, Bairrada - 4, Tras-os-Montes - 3). Of the total, 33 (26.2% of the total) are white wines, 15 being from Douro (45.5%), 4 from Alentejo (12.1%) and 14 from other regions (42.4%); the remainder are red wines, 47 from Douro (50.5%), 29 from Alentejo (31.2%) and 17 from other regions (18.3%).

RP ranks 26 wines above 90 points, of which 23 are red (21 from Douro), and 3 are white (2 from Douro). JPM ranks 14 wines above 17 points, 12 being red and 2 white, all from

6Given the small number of observations in each of these regions they were aggregated as one.

Table 4

Descriptive statistics of RP, JPM and Cellar scores per colour and wine region.

RP JPM Cellar

White wines - no. of observations 33 33 33

Average 88.2 16.1 88.1

Standard deviation 1.9 0.8 2.2

Min. 85 14 84

Max. 93 17.5 93

Coefficient of variation (%) 2.1 4.9 2.5

Red wines - no. of observations 93 93 93

Average 88.7 16.4 87.9

Standard deviation 2.6 0.8 3.1

Min. 84 14.5 72

Max. 95 18 98

Coefficient of variation (%) 2.9 4.8 3.5

Total no. of observations 126 126 126

Average 88.5 16.3 87.9

Standard deviation 2.4 0.8 2.9

Min. 84 18 72

Max. 95 14 98

Coefficient of variation (%) 2.7 4.9 3.3

Douro - no. of observations 62 62 62

Average 89.5 16.5 88.8

Standard deviation 2.4 0.7 2.6

Min. 84 18 85

Max. 95 15 98

Coefficient of variation (%) 2.7 4.5 30

Alentejo - no. of observations 33 33 33

Average 87.2 16.1 86.5

Standard deviation 1.8 0.7 2.6

Min. 84 14.5 83

Max. 92 17.5 93

Coefficient of variation (%) 2 4.5 3,0

Other regions - no. of observations 31 31 31

Average 87.7 16.1 87

Standard Deviation 2.3 0.8 3.3

Min. 84 14 72

Max. 93 17 91

Coefficient of variation (%) 2.6 5 3.8

the Douro region. In Cellar 18 wines are classified above 90 points, 14 being red and 4 white, all from the Douro region, too. The values of the coefficients of variation show that scores tend to approach the average, being slightly higher for red and the Douro wines. JPM's scores present a higher coefficient of variation than that of RP and Cellar, but still relatively low (between 4.5% and 5%).

Table 5

Descriptive statistics of WS, JPM and Cellar scores and prices per colour and region.

WS JPM Cellar Price ($USA)

White wines - no. of observations 25 25 25 25

Average 86.2 15.9 86.7 18.9

Standard deviation 2.12 0.82 2.7 10.4

Min. 83 14.5 80 9

Max. 90 17.5 95 57

Coefficient of variation (%) 2.5 5.1 3.1 55.2

Red wines - no. of observations 86 86 86 86

Average 88.8 15.9 88.1 31.9

Standard deviation 3.7 0.82 3.6 29.8

Min. 83 14.5 14 7

Max. 96 17.5 18.5 159

Coefficient of variation (%) 4.1 5.1 6.9 93.4

Total - no. of observations 111 111 111 111

Average 88.2 16.1 87.8 29,0

Standard deviation 3.3 1 3 27.2

Min. 83 14 80 7

Max. 96 18.5 95 159

Coefficient of variation (%) 3.7 6.2 3.4 93.8

Douro - no. of observations 67 67 67 67

Average 89.8 16.4 88.7 38.2

Standard deviation 3.2 0.9 3 31

Min. 83 14 80 8

Max. 96 18.5 94 159

Coefficient of variation (%) 3.6 5.7 3.4 81.1

Alentejo - no. of observations 25 25 25 25

Average 86 15.8 86.2 15.9

Standard deviation 1.7 0.9 2.9 11.2

Min. 83 14.5 80 7

Max. 90 18 95 45

Coefficient of variation (%) 2 5.4 3.3 70.5

Other regions - no. of observations 19 19 19 19

Average 85.7 15.4 86.5 13.6

Standard deviation 1.7 0.6 1.6 4.5

Min. 83 14.5 83 7

Max. 88 16.5 89 24

Coefficient of variation (%) 2 3.7 1.7 33.9

The procedures used to analyse sample 2 are similar to those used for sample 1. Table 5 presents the descriptive statistical measures for the scores assigned by the raters and for prices in the following wine production regions: Douro - 67; Alentejo -25; other regions - 19 (Ribatejo - 3, Estremadura - 6, Dao - 6, Bairrada -1, Setubal - 3). Of these, 25 (22.5% of the total) are white - 12 from Douro (48%), 9 from Alentejo (36%) and 4 from other regions (16%) - and 86 are red, 52 from Douro (60.5%), 14 from Alentejo (16.3%) and 20 from other regions (23.3%).

In terms of reputation, WS awards scores above 90 points to 27 wines, all red and from Douro; JPM ranks 11 above 17 points, all red, 1 from Alentejo and the others from Douro; for Cellar there are 23 wines above 90 points, all red, 1 from Alentejo and the remainder from Douro.

Relative to price, there are 24 wines (23 red) priced above $40, one from Alentejo and 23 from Douro. These data

reinforce the findings of sample 1 that red wines and wines from Douro are in a prime position compared with the others.

The scores of this sample also present low levels of dispersion relative to their average (the coefficients of variation are low), though slightly higher in the scores assigned by JPM. The reverse is true in the case of prices, which present a relatively high dispersion, 93.8% (the average is $29, the minimum $7 and the maximum $159). Red wines have a higher average price ($31.90) than that of white wines ($18.90), and among regions, Douro is the one that displays a higher price: $38.70, versus $16.9' in Alentejo and $13.60 in other regions.

4.2. Analysis

To check if there are differences in scores in terms of wine colour (red vs. white) and between regions (Alentejo vs. others e Douro vs. others), for each rater and sample, the statistical analysis is made computing the differences in averages. To analyse the consistency between raters, correlation coefficients are calculated for each sample.

Additionally to test the mathematical relationships between the ratings and determinant factors, regression models were developed with rating scores (RP, JPM, WS, Cellars) and price, as the dependent variables, and the objective wine characteristics -colour (red/white) and collective reputation - wine region, as independent variables. The independent variables are defined using binary variables: for the wine typology assigning T — 1, if red wine and T — 0 if white wine; and for the wine regions considering two binary variables, Alen — 1 if the wine is from Alentejo region and Alen—0 otherwise and Dou — 1 if the wine is from Douro and Dou—0 otherwise.

Sample 1

Table 6 shows statistically the differences between the averages of RP, JPM and Cellar scores, in terms of colour attributes and geographic origins, as well as the respective linear correlation (Pearson) and order (Spearman) coefficients. On average, there is no statistically significant difference between the scores of red and white wines. By region, Douro wines are in a higher score position compared with the average of all other wines. On the other hand, those from Alentejo are at a disadvantage (negative values), although for Cellar this difference is not statistically significant.

The sign and significance of the linear correlation (Pearson) and order (Spearman) coefficients suggest that the scores of the experts and those of consumers follow the same pattern and that there is a strong association between them, both in linear and ordering terms.

The information presented above highlights the reputation of Douro wines. To check the consistency of this result, three regression models are estimated. Table 7 includes the OLS robust (corrected for heteroskedasticity) estimators.

The results of the estimations show that, although the three regression models are globally significant at a 5% level, only the constant and the parameter associated with the variable Dou

Table 6

Statistics of the differences between averages scores and correlation coefficients.

Variables RP JPM Cellar

Red-White 0.44 (0.91) 0.25 ( -1.59) - 0.24 ( - 0.41)

Alentejo - Other regions -1.34* (2.84) - 0.36**(2.26) - 0.91 ( - 1.56)

Douro - Others 1.79* (4.49) 0.54*(3.99) 1.61*(3.23)

Pearson correlation RP-JPM = 0.699* JPM-CELL = 0.497*

RP-CELL = 0.559*

Spearman correlation RP-JPM = 0.696* JPM-CELL = 0.435*

RP-CELL = 0.549*

The values in parenthesis are t student. nSignificant difference at 1% level. nnSignificant difference at 5% level.

Table 7

OLS robust estimators.

RP JPM Cellar

Constant 87.532* (206.53) 15.904* (89.79) 87.255* (153.08)

T = 1 if red wine and T = 0 if white wine 0.395 (0.97) 0.235 (1.57) - 0.364 ( - 0.71)

Alen = 1 if is Alentejo wine and AL=0 otherwise - 0.309 ( - 0.57) - 0.051 ( - 0.27) 0.315 (0.41)

Dou = 1 if is Douro wine and Dou=0 otherwise 1.620* (3.03) 0.498* (2.89) 1.779*(2.55)

R-squared 0.073 0.130 0.0817

Statistic F (level of significance) 6.24 (0.003) 5.72 (0.001) 3.50 (0.018)

nParameter statistically significant at 1% level. In parenthesis are statist t de Student.

Table 8

Statistics of the differences between averages scores and coefficients of correlation.

Variables WS JPM Cellar Price (USA)

Red - white 2.56 (3.57)* 0.26 (1.19) 1.36 (2.03)** 13.01 (2.14)**

Alentejo - Others - 2.81 ( - 3.83)* - 0.48 ( - 2.20)** -2.13 (-3.15)* - 18.81 ( - 3.10)*

Douro - Others 3.76 (7.07)* 0.80 (4.78)* 2.40 (4.50)* 23.87 (5.04)*

Pearson correlation WS-JPM=0.651* JPM-Cellar= 0.525* Cellar-Price = 0.610*

WS-Cellar = 0.641* JPM-Price = 0.75*

WS-Price = 0.691*

Spearman correlation WS-JPM=0.624* JPM-Cellar= 0.522* Cellar-Price = 0.697*

WS-Cellar = 0.633* JPM-Price = 0.777*

WS-Price = 0.766*

The values in parenthesis are t student. nSignificant difference at 1% level. nnSignificant difference at 5% level.

(Douro)7 are statistically significant. These findings demonstrate that, taking into account the value of the constant, on average, the minimum value of the RP, JPM and Cellar scores are 87.5, 15.9, and 87.3, respectively. Moreover the scores assigned to wines from the Douro region, when compared with those from other regions are higher in 1.6, 0.5 and 1.8, respectively for RP,

7Additional regressions models were estimated assuming as explanatory variables only one of the variables T, AL, Dou or, at the same time, T and one of the others. Taking into account the sign and significance level of their respective parameters, the results show that: the variable T has no influence on the scores; on average, the wines produced in Alentejo have lower scores than those from other regions of Portugal; the opposite occurs for Douro wines.

JPM and Cellar. The variation of the scores is neither influenced by colour attributes (red/white) nor by Alentejo. Sample 2

Table 8 presents statistically the differences between the averages of WS, JPM and Cellar scores and prices, in terms of colour attributes and geographic origin as well as the respective linear correlation (Pearson) and order (Spearman) coefficients. The results indicate that red wines have higher scores and that, when compared with those from other regions, Douro wines have both higher scores and prices. Moreover, the Alentejo wines, when compared with those from the set "Douro and other regions" have both lower scores and prices.

Table 9

OLS robust estimators.

WS JPM Cellar Price ($ USA)

Constant 84.28* (131.22) 15.39*(74.65) 85.82*(139.80) 7.75** (2.18)

T = 1 if red wine and T = 0 if white wine 2.13 *(4.08) 0.18(0.98) 1.00 (1.57) 9.93*(2.78)

Alen = 1 if is Alentejo wine and AL = 0 otherwise 0.42 (0.73) 0.22 (0.91) - 0.347( - 0,50) 0.29(0.10)

Dou = 1 if is Douro wine and Dou = 0 otherwise 3.76*(6.47) 0.89*(4.75) 2.13*(4.25) 23.04*(5.24)

R-squared 0.384 0.183 0.1794 0.212

Statistic F (level of significance) 24.75(0.00) 8.89 (0.00) 8.02 (0.00) 11.53(0.00)

*Parameter statistically significant at 1% level. **Parameter statistically significant at 5% level.

The values and significance of the linear correlation coefficients show that there is both a strong linear and order association between the different scores and between these scores and price. Prices are positively related with scores.

The results of the robust OLS estimators, presented in Table 9, show that: (a) the four regression models are globally significant at 1% level; (b) the constant is statistically significant in all models, meaning that on average the minimum scores for white wine produced out of Alentejo and out of Douro are 84.3 for WS, 15.4 for JPM and 85.8 for Cellar and the price 7.75 $US; (c) the variable Dou is statistically significant in all the models, indicating that, on average, the scores assigned to wines from the Douro region are higher (3.76 for WS, 0.89 for JPM, 2.13 Cellar and 23.04 for price) than those from other regions); (d) in the model in which the dependent variable is WS scores, there is a significant difference (2.13) between the scores of red vs. white wines; (e) the price of red wines is 9.9 $US higher than that for the white ones; (f) the non-significance of the variable Alen indicates that the wines produced in this region do not differ in prices and scores from those of other wine regions of Portugal.

5. Conclusions

In wine economics the subject of reputation, as a way to correct consumers' information failures, is a relevant area of research, since there are still doubts about which are the variables that influence reputation: objective characteristics, expert opinions or region of origin.

In the case of Portuguese wines, this paper analysed the differences and/or consistency between the ratings assigned by several experts and consumers and, additionally, assessed if the score attained by a specific wine is influenced by colour attributes and wine region.

From our results we concluded that: (a) the scores of experts and consumers follow the same pattern, with a strong association between them; (b) there is a positive correlation between the different scores and prices; (c) the region of origin positively influences both the score assigned and prices; (d) with the exception of WS, colour has no influence on the scores; and (e) prices for red wines are much higher than those for white ones.

The results achieved indicate that the marketing of regional origin, as a reputation attribute, may have a higher payoff for the regions primarily growing red wine. We are aware that this research is merely a first approach to the problem, the analysis of which is conditioned by data availability. Future research should verify the above findings with new data sets from other years and information sources. Moreover, the results would certainly be more robust if it were possible to incorporate in the estimated models objective characteristics appearing on the label as well as sensory characteristics for the wine tasted.

References

Castriota, S., Delmastro, M., 2008. Individual and Collective Reputation: Lessons from the Wine Market. Working Paper no. 30 Economics. American Association of Wine Economists.

Castriota, S., Delmastro, M., 2009. The Economics of Collective Reputation: Minimum Quality Standards. Vertical Differentiation and Optimal Group Size. Working Paper no. 50 Economics. American Association of Wine Economists.

Combris, P., Lecocq, S., Visser, M., 1997. Estimation of a hedonic price equation for bordeaux wine: does quality matter?. Econ. J. 107, 390-402.

Combris, P., Lecocq, S., Visser, M., 2000. Estimation of a Hedonic equation for burgundy wine. Appl. Econ. 32, 961-967.

Dubois, P., Nauges, C., 2007 Identifying the Effect of Unobserevd Quality and Expert Reviews in the Pricing of Experience Goods: An Application on Bordeaux Wine. Working Paper no. 10 Economics. American Association of Wine Economists.

Frick, B., 2010. The Impact of Individual and Collective Reputation on Wine Prices: Empirical Evidence from the Mosel Valley. Working Paper. University of Paderborn (Germany).

Gergaud, O., Livat, F., 2004. Team Versus Individual Reputations: a Model of Interaction and Some Empirical Evidence. Working Paper. Université de Reims Champagne-Ardenne (France).

Gergaud, O., Livat, F., 2010. Collective Reputation Effects: An Empirical Appraisal. Working Paper no. 73 Economics. American Association of Wine Economists.

Gocekus, O., Nottebaum, D., 2011. The Buyer's Dillema - Whose Rating Should a Wine Drinker Pay Attention To? AAWE Working Paper no. 91 Economics. American Association of Wine Economists.

Hadj Ali, Nauges, C., 2007. The pricing of experience goods: the example of en primeur wine. Am. J. Agric. Econ. 89 (1), 91-103.

Hadj Ali, Lecocq, S., Visser, M., 2010. The impact of gurus: parker grades and en primeur wine prices. J. Wine Econ. 5 (1), 22-39.

Haeger, J.W., Storchmann, K., 2006. Prices of American Pinot Noir wines: climate, craftsmanship, critics. Agric. Econ. 35 (1), 67-78.

Jones, G.V., Storchmann, K-H., 2001. Wime market prices and investment under uncertainty: an econometric model for Bordeaux Crus Classés. Agric. Econ. 26, 115-133.

Landon, S., Smith, C.E., 1997. The use of quality and reputation indicators by consumers: the case of Bordeaux wine. J. Consum. Policy 20, 289-323.

Landon, S., Smith, C.E., 1998. Quality Expections, reputations and price. South. Econ. J. 64 (3), 628-647.

Lecocq, S., Visser, M., 2006. What determines wine prices: objective vs. sensory characteristics. J. Wine Econ. 1 (1), 42-56.

Luppe, M.R., Angelo, C.F., 2005. Price components of Brazilian, Argentinean and Chilean wines: a hedonic price analysis. Rev. Gest. Univ. S. Paulo (Brazil) 12 (4), 1-10.

Muhr, D., Rebelo, J., 2011. Innovation in Wine SMEs: The Portuguese Douro Boys. Working Paper no. 84. American Association of Wine Economists.

Oczkowski, E., 1994. A hedonic price function for Australian premium table wine. Aust. J. Agric. Econ. 38 (1), 93-110.

San Martin, G.J., Troncoso, J.L., Brummer, B., 2008. Determinants of Argentinean wine prices in U.S. J. Wine Econ. 3 (1), 72-84.

Schamel, G., Anderson, K., 2003. Wine quality and varietal, regional and winery reputation: hedonic prices for Australia and New Zealand. Econ. Rec. 79, 357-369.

Schamel, G., 2009. Dynamic analysis of brand and regional reputation: the

case of wine. J. Wine Econ. 4 (1), 62-80. Storchmann, K., 2011. Wine Economics: Emergence, Developments, Topics.

Working Paper no. 85. American Association of Wine Economists. Thompson, Gary M., Mutkoski, Stephen A., 2011. Reconsidering the 1855 Bordeaux classification of the Médoc and Graves using wine ratings from 1970-2005. J. Wine Econ. 6 (1), 15-36. Unwin, T., 1999. Hedonic price indexes and the qualities of wine. J. Wine Res. 10 (2), 95-104.

Viana, R.C., Rodrigues., L.L., 2007. What determines port wine prices?. J. Wine Econ. 2 (2), 203-212.