The Economic Journal, 107 (March), 390-402. © Royal Economic Society 1997. Published by Biackwell
Publishers, 108 Cowley Road, Oxford OX4 iJF, UK and 238 Main Street, Cambridge, MA 02142, USA.
ESTIMATION OF A HEDONIC PRICE EQUATION
FOR BORDEAUX WINE: DOES QUALITY
MATTER?*
Pierre Combris, Sebastien Lecocq and Michael Visser
In this paper the hedonic price technique is applied to Bordeaux wine. In the hedonic price
function we include not only the 'objective' characteristics appearing on the label ofthe bottle, but
also the sensory characteristics of the wine. Our data come from an experimental study in which
juries have evaluated and graded a sample of Bordeaux wines. The estimation ofthe hedonic price
equation shows that the market price is essentially determined by the objective characteristics. The
estimation of a jury grade equation shows that quality, unlike the market price, is essentially
determined by the sensory characteristics.
Covering an area of more than 100,000 acres, the Bordeaux wine region is
undoubtedly the world's largest wine region. It produces more than 660 million
bottles per year, varying from simple and plain table wines to the most
prestigious wines such as Petrus, Margaux, Lafite-Rothschild, Latour, Yquem,
Haut-Brion and others. The Bordeaux wine region produces bottles that can
cost just FFrio but also more than ten thousand francs, young wines that are
ready to drink as well as wines that improve even after decades of keeping,
wines that are made from different combinations and proportions of grape
varieties, and wines that originate from various sorts of soils. Clearly, Bordeaux
wines are not the same, but possess different characteristics and vary in quality.
This makes Bordeaux wine a suitable product for a hedonic price analysis.
The hedonic price method is a useful approach to study the price-quality
relationship of a product. The method amounts to a (possibly nonlinear)
regression analysis of the price on the characteristics of the product. The
implicit price of a characteristic is defined as the derivative of the price with
respect to the product attribute. Rosen (1974) has shown under which market
conditions the implicit price can be interpreted as the value consumers place on
an additional unit ofthe characteristic. If the estimated implicit price turns out
to be not significantly different from zero, then the characteristic is not valued
by consumers, or the characteristic is not considered important or relevant in
connection with the product.
The earliest reference to this literature seems to be Waugh (1928). He
observed that prices of certain fresh vegetables varied considerably on the
Boston wholesale market, and regressed the price on various physical
characteristics of the vegetables. The notion 'hedonic pricing method' is
generally attributed to Court (1939). He applied the method to automobiles,
and included several technical characteristics of the car in the hedonic price
equation. It was not until Griliches (1961) that Court's work was really
* We would like to thank O. Ashenfelter, F. Gardes, V. Lechene, L. Levy-Garboua, T. Magnac,
S. Rosen and a referee for their comments.
[ 390 ]

[MARCH 1997] HEDONIC PRICE EQUATION FOR WINE 39I
revived. Griliches' work on the hedonic price method had an important impact
and immediately initiated a vast body of empirical work, with applications to
the automobile industry (some recent references are Atkinson and Halverson
(1984), Arguea and Hsiao (1993), and Couton et al. (1996)), the computer
industry (Chow, 1967; Triplett, 1989), and the housing market (Witte et al.,
1979; Palmquist, 1984; Bartik, 1987).
As this list of applications shows, the hedonic price method has been applied
mainly to durable goods. The application that we consider in this paper is
different in two respects. The first obvious difference is that Bordeaux wine is
not a durable good, but a food product.^ The second difference is that, unlike
durable goods, not all the relevant characteristics of wine are easy to identify
and to measure.^ Although the label on the wine bottle gives important
information about the vintage year, the name of the chateau, the eventual
ranking of the wine, etc..., this is only part of the story. Indeed, labels do not
in general provide information on the sensory characteristics (taste, texture, or
odour) of a wine. The purpose of this paper is to estimate a hedonic price
function for Bordeaux wine that includes both the label characteristics and the
sensory characteristics. This will allow us to assess how much value consumers
implicitly place on the label and sensory characteristics. It will also allow us to
answer the intriguing question posed in the title of this paper: does quality
matter? As will be shown in the paper, there is strong evidence that, at least in
our data, quality is mainly determined by the sensory characteristics of a wine.
By examining the estimates of the parameters associated with the sensory
characteristics, we are able to tell to what extent quality matters in explaining
market prices.
As far as we know, there are two other studies on the hedonic pricing of wine.
In Nerlove (1995), which came to our attention while the present paper was
being revised, a hedonic price function is estimated using Swedish data. He
argues that the Swedish wine market is not competitive (it is argued that the
market prices and the product characteristics can be taken as exogenous to the
Swedish consumer), and that a standard hedonic regression is not appropriate.
The author proposes therefore an alternative method to draw inferences about
consumer preferences. In Oczkowski (1994) the method is applied to Australian
table wine. The main difference with the present study is that he essentially
considers objective and easily observable characteristics (vintage year, vineyard
region, grape variety, etc...). The sensory characteristics are lacking in his data
set. Our data source, in contrast, contains detailed information about the
sensory characteristics of wines, and has a number of other advantages over the
sort of information one can find in wine guides.
The paper is organised as follows. In Section I the data are presented. In
Section II we present the empirical results. Section III contains some
concluding remarks.
' Other applications of the hedonic price technique to food products are Ladd and Suvannunt (1976),
Stanley and Tschirhart (1991), and the references in these two papers.
* Waugh also considered food products, but the relevant product characteristics were measurable physical
properties that were readily available to all consumers by simple observation.
© Royal Economic Society 1997