Content uploaded by Larry Lockshin
Author content
All content in this area was uploaded by Larry Lockshin on Nov 17, 2020
Content may be subject to copyright.
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 19
1. Introduction
The vibrant nature of the Australian wine sector is
typified by the compound annual growth rate of exports
of over 18% from 1988 to 2001. This compares to a
world growth rate of just 3.6% over the same period
(AWBC, 2001). Australian wines have increased from
just over one percent of world wine export value to 5.5%
in the same period. The growth in volume and value of
exports has been accompanied domestically with an
increase in the consumption of bottled wine compared to
the consumption of cask (or bag in the box) wine
(Anderson et al., 2001). Both of these trends have raised
the price of the average bottle of Australian wine over
the same 13 year period.
Australian wine production practices have differed in
many key ways from those of the traditional European
wine producing countries (Halliday, 1999). Wines are
identified by the varietal names of the grapes rather than
the regions as in France and other traditional wine
producing countries. Also, wines are often blended
across regions. Brands are important identifiers for the
wines, and like in Europe, some producers have better
reputations than others for wine quality. But there is a
trend to single region wines and the establishment of the
Australian-EU trade agreement for wine has resulted in
the formal identification of regions, now called
geographic indicators (AWBC, 1994; Ryan, 1994). This
has awakened interest in the value of regions compared
to wine brands (Tustin and Lockshin, 2001). Also wine
companies have merged and created multiple wines
under specific brand names, so the reputation that
accrues to a brand may be related to a number of
different wines at different price points. This vertical
ranging is similar to how automobile brands seek to
establish a halo effect from the higher priced models to
the lower end of the range.
Hedonic price analysis has been used to measure the
marginal value or contribution to the price for a number
of different products. Wine is one product well suited to
this type of analysis, due to its highly differentiated
forms and the difficulty in objectively assessing quality
(Oczkowski 2001). This paper uses hedonic price
analysis to measure the impact on price of winery brand
name, wine quality, region, size of the winery
Components of Wine Prices for Australian Wine: How Winery Reputation,
Wine Quality, Region,Vintage, and Winery Size Contribute to
the Price of Varietal Wines
Bith-Hong Ling & Larry Lockshin
Abstract
Australian wines are identified by the varietal names of the grapes rather than the regions as in France and other
traditional wine producing countries. This paper uses the concept of hedonic price theory to investigate a range of
extrinsic characteristics’ ability to predict prices for different climate regions (warm and cool) and four major wine
varieties of Australian wines, two reds (shiraz and cabernet) and two whites (chardonnay and riesling). The effects of
winery reputation (wine company/brand), winery size (production scale), age of the wine, and region of origin (wine
grape source) contributing to the relationship between price and quality attributes of Australian wines are investigated,
based on 1880 observations of bottled wines. Wine quality rating and winery/brand reputation have major effects on the
price, while region and size of winery have differential effects depending on the variety of grape. Vintage has only a
minor effect.
Keywords: Australian wine, Wine attributes, Wine brand, Wine price, Hedonic price analysis
2. Review of Literature
Hedonic price analysis is based on the hypothesis that
any product represents a bundle of characteristics that
define its quality and therefore price. The theoretical
foundation for this type of research was first examined
by Rosen (1974). The observed price for a good is the
sum of the implicit prices for the attributes. Various
independent variables, often quality indicators, are
regressed on the retail price. The major issue with this
type of analysis is that supply and demand factors are
part of the price as well as the quality attributes (Unwin,
1999). The solution proposed by Arguea and Hsiao
(1993) is to pool cross sectional data specific to a
particular side of the market rather than measure a single
product. We follow their recommendation.
Wine is a product that varies with vintage year, producer,
region, and production technique among other factors.
Consumers often find it difficult to judge the quality of
wine before actually drinking the wine (Charters et al.,
1999). They use label information to help make an
informed choice. These extrinsic factors (Olson and
Jacoby, 1972) are typically the major source of quality
determination before purchase. The reputation of the
winery (brand), the region of origin, the variety, and even
third party quality ratings can be used by the consumer
to help make the purchase decision. Shapiro (1983)
modeled the reputation effects for high quality products.
He showed that reputation allows some producers to sell
their products for a premium, and he based this on the
interpretation that these reputation effects are the
outcome of investments in building that reputation. He
also stated that it is costly for consumers to gain
information in some environments and that learning
about reputations is an effective way to reduce their
decision making costs.
Oczkowski (1994) did the first hedonic pricing study of
Australian wine. He showed that the log linear form was
the best function to model retail prices for six wine
attributes. Reputation effects were significant, but
quality effects were not. Nerlove (1995) examined the
Swedish wine market, a government monopoly, which
allowed him to assume totally exogenous prices. His
form of the equation was similar to Oczkowski with
similar results. Combris et al. (1997) estimated a hedonic
price equation for Bordeaux wines using intrinsic wine
quality judgments of acid, fruit, palate by expert tasters,
as well as the extrinsic factors of reputation and vintage.
There most of the variance in price was explained by
extrinsic factors, especially the reputation as measured
Components of Wine Prices for Australian Wine
20 Australasian Marketing Journal 11 (3), 2003
(processing capacity), and vintage date. These objective
elements are available to consumers when making a
choice of wines. The quality rating of the wine is
available to consumers through ratings published by
various wine writers and critics. Often these are
displayed on the retail shelf. Understanding the value of
each of these elements is useful for wine marketing
managers. Knowledge of the relative importance of
brand or region to consumers would allow marketing
managers to decide how much each should be
emphasized in promotion or even how to organize
products on the shelves. Small producers need to know
whether region is an important component, because with
limited funds to spend, they could band together to
promote regions rather than their own brands. Though
the quality of the wine is a subjective judgment, wine
writers are believed to influence consumer perceptions.
Hedonic price research, using wine critics’ evaluations,
can estimate the value of these judgments. The results
then are useful for allocation of scarce resources to the
most efficient factors in informing consumers and
maintaining or increasing prices.
The major difference between this paper and the
previous use of the hedonic price function for Australian
wines (Combris et al., 1997, 2000; Oczkowski, 1994,
2001; Schamel and Anderson, 2001) is the specification
of an equation for each variety. This allows us to
measure the value of the independent variables
separately for each type of wine. This makes sense, as
the reputations of different regions for wine quality are
linked to specific varieties, and this is usually related to
the climate as either cool or warm (White, 1995). One
can add regions and varieties to a single equation to find
the partial value of each, but this method does not allow
differential values to be estimated for each variety, and
therefore, we don’t know if all varieties behave similarly.
Following Oczkowski (2001), we also used two stage
least squares (2LS) in order to first estimate the quality
rating before including it in the hedonic price equation.
This methodology takes explicit account of the fact that
quality itself is subjective and even in the mind of the
quality rater (James Halliday here) it is often influenced
by the reputation (brand) and price of the wine.
We present our work in several sections. First we review
the literature on hedonic pricing and follow that with the
model specification and data. We present the results next
along with the discussion. We conclude with
implications and some ideas for future research.
by the classification of the winery (first growth, second
growth, etc.). The same authors recently published a
paper using the same methodology for Burgundian wines
rather than Bordeaux wines (Combris et al 2000). The
same effects were noted, with reputation as measured by
ranking (rather than classification) and vintage having
the largest effect, but some of the sensory measures
‘boosted’ these objective quality effects, because the
sensory characteristics were correlated with the ranking
of the wineries. Landon and Smith (1997, 1998) added
extrinsic factors and intrinsic factors to their hedonic
price analysis for Bordeaux wines. They found that
reputation factors were most significant in explaining the
price consumers were willing to pay for Bordeaux wines,
but the estimated coefficients varied over vintages.
Schamel and Anderson (2001) used two separate wine
quality rating sources to analyse Australian and New
Zealand wines. Their method pooled all the wines and
regions in a series of equations for each year. They
showed that winery reputation, region, variety, and
quality estimation were all significant in some of the
equations using Barossa shiraz as the relative
comparison point in the dummy variable estimation. The
trends were similar in both data sets. They also showed
that region has become more important over the 10 year
period of their data.
One of the major issues in using hedonic price
techniques with wine is the subjective measure of quality
(Oczkowski 2001). The objective characteristics on the
label are easy to define and easy for consumers to view.
However, ratings or reputations of wineries and ratings
of individual wines by wine writers are purported to
carry weight with the consumer, but little empirical
evidence exists as to their effects. The two papers by
Combris et al. (1997, 2000) demonstrate that the
reputation as measured by the classification of Bordeaux
wineries, or the ranking of Burgundian wineries has a
major effect on the price, while quality assessments by
tasters have little effect. Oczkowski’s paper (2001) uses
multiple reputation ratings by different wine writers to
demonstrate this factor has the largest effect on wine
prices as well.
Oczkowski (2001) incorporated the error rates in his
analysis by using multiple wine writer ratings in a
confirmatory factor analysis prior to the regression. He
then goes on to use 2SLS to measure both quality and
reputation as attributes containing measurement error.
The fit statistics and error rates show that 2SLS is
superior to OLS using this specific data set. He also
shows that when combined reputation and quality
measures are used in the same equation, there is little
price variance explained by quality added to reputation.
One of the major issues with the analysis is that in order
to utilize multiple indicators of quality and reputation,
Oczkowski could only find 276 wines in common across
the four datasets, as compared to over 1200 in his
previous paper (1994). This small sample size made
measuring the effects of regions or varieties with
separate equations impossible. This also may have
impacted the measurement of reputation and quality. We
use reputation (brand name) to help predict quality and
then do not use reputation in the second stage.
We are most interested in the effect of different regions
on the prices of selected varieties. We use the same
Halliday (1999) data set as Schamel and Anderson
(2001) to investigate a range of extrinsic characteristics’
ability to predict retail prices for four major varieties of
grapes. We understand the criticism of using only a
single wine writer’s subjective measures or wine quality
(Oczkowski 2001), however we strongly believe that
consumers are unlikely to compare different wine
writer’s evaluations before buying wine. Our experience
also shows that wine stores in Australia may use a single
score from one writer on the edge of the shelf below a
bottle of wine (shelf talker), but never use multiple
writer’s scores. We run equations for each variety
separately in order to be able to see the effect of region
and winery reputation by variety. Previous hedonic
research has used interaction terms to measure the
impact of region and variety (Schamel and Anderson
2001; Oczkowski 1994, 2001). However, we feel that
both regions and wineries have gained reputation for
specializing in a few varieties, so our method separates
the effects along the lines of four major grape varieties.
We also use winery size, age of winery and vintage to see
how these influence price for each variety.
3. Model Specification
Several studies have attempted to estimate the
relationship between the price of a wine and its various
characteristics and hedonic wine functions by using the
ordinary least squares (OLS) technique in recent years.
Oczkowski (2001) first addressed the measurement error
issue on the previous hedonic wine studies when the
OLS regression was used to estimate the single measures
of wine quality and reputation given the presence of
regressors which contain measurement error. According
to Johnson (1987) and Greene (1993), when explanatory
variables are endogenous in the equation, applying OLS
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 21
may lead to biased estimates of the coefficients, which
are correlated with the error term. The essence of two
stage least squares (2SLS) is the replacement of
endogenous variables by the predicted values, which
could be purged of the stochastic elements. The
consistent 2SLS estimator is obtained by OLS regression
of the dependent variable on predicted endogenous
variables and predetermined variables in the equations.
Due to the concern that the reputations for wine quality
of different wine-growing regions are linked to specific
varieties, we estimate a separate equation for each type
of climate region (warm vs. cool) and wine variety (red
vs. white) and then make a comparison of price-quality
relationships between the different equations. Our
model, to be estimated by 2SLS regression for the
hedonic price function for Australian wines, assumes
that wine price is influenced by quality and the objective
characteristics of attributes as follows in Equation 1.
The hedonic wine price equations for different climate
(warm vs. cold) regions and wine varietals (two whites
vs. two reds) are postulated in Equations 2 and 3,
respectively.
where:
PRICE: the natural logarithm of wine retail price,
recommended in the Halliday’s wine guide (1999), and
sourced from each wine producer.
QUALITYPRED: the predicted values of the quality rating
score of the wine (by Halliday on a 100 point scale),
which is obtained from the 1st stage of 2SLS by the OLS
regression of wine quality rating score (QUALITY) on the
recommended retail price (PRICE) and the subjective
characteristics - its associated producing winery
(including wine group) reputation rating score
(WINERY). These variable values are all subjectively
judged by wine expert James Halliday.
Objective characteristics of wines:
VARIETALK: the dummy variable for wine varietal k,
including two whites (chardonnay and riesling) and two
reds (shiraz and cabernet sauvignon).
SIZE1:the dummy variable for the wine producer size
(tonnes of wine grapes crushed) class l, including very
small size (under 100 tonnes); small size (100~499
tonnes); medium size (500~2,499 tonnes); large size
(2,500~9,999 tonnes); and very large size (over 10,000
tonnes).
Components of Wine Prices for Australian Wine
22 Australasian Marketing Journal 11 (3), 2003
AGEm:the dummy variable for the wine age class m
where m = under 3 years (vintage year after 1997); m =
3~4 years (1996 and 1997); m = 5~6 years (1994 and
1995); m = 7~8 years (1992 and 1993); and m = over 8
years (before 1992). Although previous research
(Combris et al., 1997; Landon and Smith, 1997 and
1998) used vintage date as a variable, according to the
Australasian Wine Exchange (2000), vintage is not as
important in Australia as brand and specific wine.
ORIGINALn: the dummy variable for the wine region of
origin class n, which is classified into two groups - warm
climate regions (including Barossa Valley, Clare Valley,
Great Southern, Hunter Valley, Margaret River and
McLaren Vale,) and cool climate regions (including
Adelaide Hills, Coonawarra, Mornington Peninsula,
Tasmania and Yarra Valley).
The parameters ( ßq, ßk, ßl, ßm and ßn) are to be estimated
for forming the price contribution of the product
characteristics of Australian wines. The structure of
hedonic price (PRICE) is hypothesized to be positively
related to the wine quality rating score (ßq > 0).
According to the different objective characteristics of the
wines, the positive/negative values of coefficients for
dummy variables reflect relative price
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 23
Classification
All samples
Climate of region
Warm regions
Cool regions
Variety of wine
Chardonnay
Riesling
Shiraz
Cabernet Sauvignon
Rating of winery reputation
Typically good (3 star)
Good (3.5 star)
Very good (4 star)
Extremely good (4.5 star)
Outstanding (5 star)
Size of wine producer
< 100 tonnes (very small)
100 ~ 499 tonnes (small)
500 ~ 2,499 tonnes (medium)
2,500 ~ 9,999 tonnes (large)
> 10,000 tonnes (very large)
Sample
Size
1880
1262
618
637
326
531
386
90
257
590
501
442
369
454
449
188
420
Mean
25.96
26.47
24.92
23.25
16.17
32.94
29.08
17.40
20.68
20.62
28.79
34.68
21.73
24.52
30.97
22.67
27.34
S.D.
20.05
23.18
11.17
9.66
4.19
31.50
16.05
5.47
8.33
7.16
17.85
32.91
8.96
12.02
23.67
12.42
29.28
Mean
87.99
87.86
88.26
87.31
88.21
88.57
88.14
82.50
85.53
87.19
88.94
90.53
88.00
87.59
88.40
86.92
88.45
S.D.
4.75
4.84
4.55
4.95
4.47
4.64
4.69
5.02
4.30
4.30
4.29
4.10
4.45
4.73
4.87
4.84
4.76
Correl.
P and Q
0.338
0.358
0.321
0.501
0.454
0.352
0.428
0.302
0.329
0.162
0.312
0.347
0.292
0.296
0.388
0.288
0.397
Table 1:
Means, standard deviations (S.D.) and correlations of retail price (PRICE) and quality rating score (QUALITY) for
Australian wine
PRICE QUALITY
premiums/discounts, as compared with the dummy base
variable. Moreover, the degrees of ßk, ßl, ßm and ßnand
values respectively, represent the effects of wine varietal,
production scale (producer size), age of the wine, and
region of origin (wine grape source) contributing to the
relationship between retail price and quality attributes of
Australian wines.
The econometric package PcGive 9.0 (Doornik &
Hendry, 1996) is used to generate the properties of the
data and all 2SLS estimations for the study.
Multicollinerity might be a potential problem when
several linear relationships exist between a set of dummy
variables in Equations 2 and 3. Using the Wald test
statistics, the linear restrictions test on the variance-
covariance matrix for the estimated parameters is
performed and points to no serious problem of
multicollinerity in the equation.
4. Data
The total final sample covered 1880 observations of
bottled wines, including 1262 warm climate wines and
618 cool climate wines; 637 chardonnay wines, 326
riesling wines, 531 shiraz wines, and 386 cabernet
sauvignon wines. Only wines from single regions and
single varieties were used in our analysis; multi-region
and multi-varietal blends were excluded in order to
measure the impact of region on each chosen variety of
wine. All basic data on technical specifications and retail
prices of bottled wines selected were obtained from the
database set provided by well-known Australian wine
expert James Halliday (1999). The exception is that data
on the production capacity of each wine company/group
is collected from The Australian & New Zealand Wine
Industry Directory (Winetitles, 2001).
Summary statistics of data on price of selected
Australian wines categorized by the climate of region
and wine variety are provided in Table 1. We can see that
by separating the wines into varieties, quite different
average prices are charged for each. Riesling has the
lowest average price at AU$16.17 per 750ml bottle,
Components of Wine Prices for Australian Wine
24 Australasian Marketing Journal 11 (3), 2003
Figure 1: The impact of winery reputation on wine quality rating score
Notes: The dummy base is the five-star outstanding winery.
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 25
while the highest is for shiraz at AU$32.94. In addition,
the correlation values of price and quality for
chardonnay and riesling are 0.50 and 0.45 respectively,
which are higher than other types of wine varieties.
Overall, price does not count for a major portion of the
variation in quality, even though it is positively
correlated. The relatively high of standard deviation
(S.D.) of wine price can be found for warm region wines,
shiraz wines, and wine produced by outstanding wineries
(5 star) and very large size wine producers (> 10,000
tonnes).
We can also see that even though there are fewer large
wineries than small ones in Australia, the number of
wines made by each size class is relatively similar. Large
wineries produce many types of wines in each variety
under different brand names. We should note here that
the winery rating (reputation) is for the brand names, but
the size class is for the parent company. For example,
Southcorp is Australia’s largest wine group and all
brands made by that company are given a dummy
variable for the largest size, but each brand has a
different rating based on its long term reputation:
Penfolds, Lindemans, Wynns, Seppelt, etc. We feel that
the production practices (type of equipment, expertise of
the wine maker, etc.) are best identified with the parent
company, while the reputation effects accrue to the name
the consumer sees on the bottle.
5. Results and Discussion
5.1 Winery Reputation
According to the winery star rating system of Halliday
(1999), the Australian winery reputation is classified into
five stars for an outstanding winery regularly producing
exemplary wines; four and a half stars for an extremely
good winery virtually on a par with outstanding one; four
stars for a winery consistently producing high-quality
wines; three and a half stars for a solid, reliable producer
of good wine; and three stars for typically good
producer, but may have a few lesser wines. Winery
ratings are for the overall reputation of the winery, not
for individual types of wines made. The impact of
winery reputation on wine quality rating score, obtained
from the 1st stage of 2SLS regression is shown in Figure
1. The five-star outstanding winery is used as the
comparison for other four classifications of winery
reputation.
As expected, each rating below five stars results in a
decrease in wine quality rating score, except for riesling.
The degrees of sensitivities are different for each of the
regions and wine varieties. In terms of the climate for
growing wine grapes, cool region wines have a relatively
higher change in wine quality for each change in winery
reputation than warm region wines. A one half star
reduction in reputation from five to four and a half would
result in a 2.5 point decrease in quality rating points (100
point scale) for riesling, 1.4 for cabernet, and only slight
quality change for chardonnay (0.82) and shiraz (0.07).
The effect of being a lower than four star winery
becomes larger. The estimated coefficients across all
four varieties are similar for being a three and one half
star winery versus a five star. As the winery rating
decreases to three stars, however, greater variation in the
coefficients can be found, ranging from -8.2 (for
cabernet), -7.0 (for shiraz), -5.7 (for chardonnay) to -1.9
(for riesling). Clearly, having a good winery reputation is
relatively important for cabernet and shiraz but less so
for riesling. Also, the quality effects of region (greater
for cool regions than warm regions) seem to concur with
the general viticultural situation; it is more difficult to
achieve good quality in cool regions, due to the potential
for frosts and poor ripening, but the results may warrant
the difficulty.
5.2 Region:Warm vs. Cool
Table 2 presents the results of 2SLS regressions
conducted based on the model in Equation 2, using the
above definition for dependent variable (PRICE) and
independent variables (QUALITYPRED ) and objective
characteristics of wine. The coefficients for independent
variables are directly interpreted as a percentage price
effect. The regression for all regions together is shown in
first column and then warm and cool region equations
are reported in second and third columns. We first can
look at the effect of the wine quality rating from
Halliday’s book (1999). The effect of the quality score
out of 100 points represents an increase of 12.5% in price
for each point increase in quality score for all regions
and 14.5% for warm and 8.6% for cool climate regions.
A larger quality effect on wine price is found for the
warm climate regions. If a warm region wine increases
in quality from 85 points to 90 points, we could expect a
price increase of 72.5%. For a cool region wine the price
increase would be only 43%. This may occur because
Australian consumers prefer wines from warmer regions
or perhaps because warmer regions were planted earlier
and consumers have more knowledge of them and are
therefore willing to pay higher prices for their wines.
Taking chardonnay as the comparative benchmark for
wine variety, the coefficients for are highly significant in
Components of Wine Prices for Australian Wine
26 Australasian Marketing Journal 11 (3), 2003
PRICE CLIMATE (dependent variable)
QUALITY PRED
VARIETY k
Shiraz
Cabernet
Riesling
Chardonnay
SIZE 1a
< 100 tonnes (very small)
100 ~ 499 tonnes (small)
500 ~ 2,499 tonnes (medium)
2,500 ~ 9,999 tonnes (large)
> 10,000 tonnes (very large)
AGE m(vintage year)
1 ~ 2 years (after 1997)
3 ~ 4 years (1996 and 1997)
5 ~ 6 years (1994 and 1995)
7 ~ 8 years (1992 and 1993)
Over 8 years (before 1992)
ORIGIN nb
Barossa Valley (W)
Clare Valley (W)
Great Southern (W)
Hunter Valley (W)
Margaret River (W)
McLaren Vale (W)
Adelaide Hills (C)
Coonawarra (C)
Mornington Penins (C)
Tasmania (C)
Yarra Valley (C)
Constant
R2
Wald test c
Coeff.
0.125**
0.181**
0.152**
-0.245**
Base
0.168**
0.167**
0.167**
-0.001
Base
-0.136**
-0.093*
-0.078*
-0.066
Base
-0.053
-0.109**
-0.127**
-0.149**
Base
-0.109**
-0.054
-0.105**
-0.002
-0.026
-0.070*
-7.899**
0.592
8853.2**
t-value
(39.00)
(9.35)
(6.95)
(-10.56)
(6.27)
(7.17)
(7.51)
(-0.04)
(-3.10)
(-2.34)
(-1.97)
(-1.60)
(-1.57)
(-3.02)
(-3.43)
(-4.20)
(-3.32)
(-1.36)
(-2.88)
(-0.05)
(-0.63)
(-1.89)
(-26.86)
(0.000)
Coeff.
0.145**
0.163**
0.112**
-0.263**
Base
0.182**
0.163**
0.149**
0.019
Base
-0.095*
-0.063
-0.045
-0.032
Base
-0.040
-0.098**
-0.122**
-0.142**
Base
-0.090**
-9.592**
0.640
7227.4**
t-value
(35.91)
(6.86)
(4.10)
(-8.81)
(5.46)
(5.72)
(5.93)
(0.57)
(-1.77)
(-1.36)
(-0.95)
(-0.64)
(-1.14)
(-2.58)
(-3.14)
(-3.85)
(-2.64)
(-26.19)
(0.000)
Coeff.
0.086**
0.183**
0.234**
-0.209**
Base
0.104*
0.144**
0.186**
-0.091*
Base
-0.164*
-0.096
-0.089
-0.082
Base
0.113**
Base
0.178**
0.122**
0.114**
-4.597**
0.524
5450.1**
t-value
(19.03)
(5.72)
(7.01)
(-6.21)
(2.21)
(3.49)
(4.06)
(-1.60)
(-2.34)
(-1.47)
(-1.37)
(-1.22)
(2.82)
(4.04)
(2.91)
(2.95)
(-11.15)
(0.000)
Table 2:
Hedonic price equation for Australian wines by climate of region
Notes: * Significantly different from zero at the 5% level and ** at the 1% level.
aMeasured by the tonnes of wine grapes crushed of wine producers.
bW referring for warm climate region and C for cool climate region.
cWald statistics show the linear restrictions test (Doornik and Hendry 1996, pp.241).
All regions Warm regions Cool regions
all three equations. The positive sign on the coefficient
for shiraz and cabernet reflects that these two red wine
varieties have price premiums as compared to
chardonnay. However, a negative sign indicating a price
discount is found for riesling. We also investigate the
price effect of winery size, which has not been part of
other hedonic price research papers for wine. Our results
indicate that the largest size wineries (greater than
10,000 tonnes) have no effect on price compared to the
2,500 to 9,999 tonne wineries across all four varieties.
As we move to the small and medium size (less than
2,500 tonnes) wineries, we find significant and positive
coefficients for , indicating that smaller wineries charge
more (and presumably consumers are willing to pay
more) for their wines than the largest wineries. Most
interestingly, the greatest price premiums are received by
the smallest wineries (less than 100 tonnes) for
producing warm region wine, while the medium
wineries (500~2,499 tonnes) receive a premium for cool
region wine.
As compared to wines from Bordeaux (Combris et al.,
1997; Landon and Smith, 1997 and 1998), where older
vintages often increase in price, we find little price effect
of older vintages for Australian wines. Our results show
that the values of coefficients on nearly approach to zero
(except for less than 3-year-old young wine), indicating
not much price change for different vintages of
Australian wines for two reasons. First, there are not
many older wines sold in Australia. Wines are mainly
made for early drinking and only a small percentage are
cellared commercially and resold (Halliday, 1999).
Second, the prices of Australian wines have been rising
over the last 10 years (Anderson et al., 2001) and any
increase in price of older wines might be hidden by
overall price rises.
For the past decade, many Australian wine companies
have used region along with their brand names as a wine
branding and marketing tool to differentiate the quality
and type of wine from a specific region and generate
awareness of the region itself. (Lockshin, 1997;
Rasmussen and Lockshin, 1999) The coefficients of
ORIGINnreported in Table 2 represent the regional
reputation effects on the price of the wines. Compared
with Margaret River as a warm region, wines from the
other warm regions receive a significant price discount,
ranging from 9.0% for McLaren Vale and 9.8% for Clare
Valley to 12.2% for Great Southern and 14.2% for
Hunter Valley. On the other hand, while Coonawarra is
thought of as the comparative benchmark for cool region
red wines, our results show that all other cool regions
have a significant price premium, ranging from 11.3%
for Adelaide Hill to 17.8% for Mornington Peninsula.
These summary effects may mask effects on price for
specific wine varieties.
5.3 Wine Variety: Red vs. White
Table 3 details the results of separate hedonic price
equations for each of four wine varieties. Our results
show that the effect of the wine quality rating for shiraz
and cabernet is a 14.1% and 13.1% price rise per rating
point, whereas that of chardonnay and riesling reaches
10.9% and 9.3%. Furthermore, the largest size wineries
have no effect on price for cabernet and riesling as
compared to all other size of wineries. However, the
coefficients for the small size (100~499 tonnes) and
medium size (500~2,499 tonnes) winery dummy
variables for shiraz and chardonnay are significantly
positive, implying that the small and medium size
wineries received a premium price compared to the
largest sized wineries. In particular, the premium price
(26.6% and 21.7%) in the case of chardonnay is greater
than shiraz (11.5% and 18.3%). In comparison to the
oldest vintage for wine, the younger chardonnay
commands a decreasing price premium, but the younger
shiraz commands a rising price. The differences in the
vintage effect suggest the existence of unequal price
behaviours according to the wine variety. As stated
earlier, the prices of shiraz have increased dramatically
over the last decade (Schamel and Anderson 2001),
while those for chardonnay have not. The increases in
the prices of shiraz may mask vintage effects.
Overall, for each variety there are regions that command
a higher price than other regions, given the same quality
score. In Table 3 we can easily see that the magnitude of
the coefficient on the wine region effect (ORIGINn) for
the red varieties (shiraz and cabernet) is statistically
significant and larger than for white varieties
(chardonnay and riesling). Barossa is often thought of as
the comparative benchmark for shiraz. In general our
results show that all the warm regions, which are
significantly different from Barossa, have a negative
coefficient; they all have a negative effect on the price of
shiraz compared to Barossa. Barossa is classed as a
warm region, but the regions which do not differ
significantly from Barossa are all cool (Adelaide Hills,
Mornington Peninsula, and Yarra), with the exception of
Coonawarra. Because shiraz doesn’t ripen as well in cool
areas, the lack of a significant difference could be due to
small samples from each of those regions or possibly
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 27
Components of Wine Prices for Australian Wine
28 Australasian Marketing Journal 11 (3), 2003
PRICE CLIMATE (dependent variable)
QUALITY PRED
SIZE 1a
< 100 tonnes (very small)
100 ~ 499 tonnes (small)
500 ~ 2,499 tonnes (medium)
2,500 ~ 9,999 tonnes (large)
> 10,000 tonnes (very large)
AGE m(vintage year)
1 ~ 2 years (after 1997)
3 ~ 4 years (1996 and 1997)
5 ~ 6 years (1994 and 1995)
7 ~ 8 years (1992 and 1993)
Over 8 years (before 1992)
ORIGIN nb
Barossa Valley (W)
Clare Valley (W)
Great Southern (W)
Hunter Valley (W)
Margaret River (W)
McLaren Vale (W)
Adelaide Hills (C)
Coonawarra (C)
Mornington Penins (C)
Tasmania (C)
Yarra Valley (C)
Constant
R2
Wald test c
Coeff
0.141**
0.114
0.115*
0.183**
0.089
Base
-0.282*
-0.187
-0.151
-0.160
Base
Base
-0.172**
-0.275**
-0.281**
-0.192**
-0.134**
0.049
-0.261**
-0.064
n.a.
0.068
-6.855**
0.507
2045.3**
t-value
(18.55)
(1.64)
(2.25)
(3.82)
(1.32)
(-2.17)
(-2.09)
(-1.69)
(-1.75)
(-2.75)
(-3.37)
(-4.51)
(-2.13)
(-2.60)
(0.49)
(-3.67)
(-0.50)
n.a.
(0.76)
(-24.62)
(0.000)
Coeff
0.131**
0.047
0.065**
0.042**
0.024
Base
-0.221
-0.103
-0.092
-0.117*
Base
-0.116**
-0.174**
-0.214**
-0.427**
-0.018
-0.114**
-0.155
Base
n.a.
-0.115
-0.119
-5.511**
0.659
3091.7**
t-value
(23.07)
(0.85)
(3.13)
(3.14)
(1.61)
(-1.80)
(-1.91)
(-1.76)
(-2.18)
(-2.71)
(-3.02)
(-2.84)
(-4.78)
(-0.35)
(-2.52)
(-1.34)
n.a.
(1.31)
(-1.73)
(-23.97)
(0.000)
Table 3:
Hedonic price equation for Australian wines by wine varietal
Notes: * Significantly different from zero at the 5% level and ** at the 1% level.
aMeasured by the tonnes of wine grapes crushed of wine producers.
bW referring for warm climate region and C for cool climate region.
cWald statistics show the linear restrictions test (Doornik and Hendry 1996, pp.241).
Shiraz Cabernet
higher demand due to lower availability. McLaren Vale
and Hunter Valley are both thought to be good regions
for shiraz, but both have a significant price discount in
comparison to Barossa, 13.4% and 28.1% respectively.
The benchmark region for cabernet is Coonawarra. Our
results reinforce this with all significant comparisons for
cabernet from warm regions being negative, with the
exception of Margaret River, which has a good
reputation for cabernet wine. For instance, the price
discounts range from 42.7% for Hunter Valley to 11.4%
for Mclaren Vale and 11.6% for Barossa Valley. The
regions that do not significantly differ in its effect on
price with Coonawarra, like with shiraz and Barossa, are
cool climate regions such as Adelaide Hills, Tasmania
and Yarra Valley, probably for the same reason of small
samples or limited availability.
Chardonnay is priced the highest from Margaret River,
which is classed as a warm region, especially in
comparison to Adelaide Hills, Yarra Valley and Tasmania.
Surprisingly, we find that the coefficient on the regional
effect (ORIGINn) for almost all other wine regions as
compared to Margaret River is statistically insignificant
(except for Clare Valley) and approach to zero, which can
not support the existence of the regional reputation effect
on the price of the Australian chardonnay wines. The
story with riesling is not so clear. Clare Valley is thought
by some to be Australia’s premiere riesling region. But
our analysis shows that Clare only receives a 13.9%
premium by compared to Tasmania and is actually
discounted by 10.6% ~ 12.7% in relation to riesling from
Coonawarra, Margaret River and Yarra Valley.
6. Implications and Future Research
Our research has added to the use of hedonic pricing for
wine by adding several new parameters: winery
reputation and size and by calculating quality as a
function of the other characteristics before using it in the
hedonic pricing equation. We also derive equations for
two climate regions and four major wine varieties
separately, rather than treat all the regions or varieties as
one. Keeping the varieties in one equation probably
makes sense for European regions, where there are
restrictions on which varieties can be grown, but it still
mixes up white and red varieties. When a range of
varieties are put into one equation for a country like
Australia, we only can see the price effect of the varieties
in general; we cannot untangle the effects of different
varieties in each region.
Our first table shows that across three of the four
varieties, Halliday’s rating of winery reputation is highly
correlated with quality. This gives credence to his rating
system, that wineries with higher ratings do indeed
produce better quality wine. This could be an artifact of
his system, where he gives higher quality scores to his
higher rated wineries, but the fact that reputation is
significant in the hedonic price equations shows that
wineries that have invested in better quality over the
years command higher prices. Shapiro’s (1983)
observation of reputation effects does hold in the
Australian wine industry. This should give heart and
direction to wine marketing managers to keep investing
in long term brand building at the same time as their
wine making colleagues strive to make better quality
wines. The analysis shows that wines from wineries with
the highest (five star) reputations charge an average of
$6.00 more per bottle than a comparable four and a half
star winery, and an amazing $14 more than a four or three
and a half star winery. This price differential when
multiplied over the production capacity of the winery
shows the value of a top reputation and gives some
indication of how much a wine company could invest in
both production and promotion to gain this level of status.
There is also a return to quality above winery reputation
(Table 2). Each quality point increases the price of the
wine between 9% and 14%, which again gives some
measure of how much a winery could invest to improve
quality and make a reasonable return for that investment.
The word ‘quality’ is mentioned too often in the wine
industry without regard to its costs or to its benefits. Our
analysis puts some real numbers on quality as rated by
James Halliday. For example, if a winery had done some
small batch experiments and shown that by reducing
yields on a certain shiraz vineyard from 6 tonnes per
hectare to 3 tonnes would result in the quality rising from
88 points to 91 points, they should be able to raise their
prices by 14% X 3 (points) or 42%. If their bottle price
for shiraz was $18, they might be able to raise their price
to $25. This would mean about $3.50 more per bottle at
the winery wholesale price. If they made about 3000
cases of this wine, this would mean an increase in gross
margin of $105,000, which could be compared to the
cost of the reduced yields in the vineyard to see if this is
a feasible investment.
Our research has shown that winery size can have an
effect on the price of the wines sold. Wineries crushing
greater than 2,500 tonnes of grapes get substantially less
for each bottle of wine than smaller sized wineries, and
the effect is fairly constant down to the smallest size
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 29
(Table 2). Large wineries are more efficient and probably
charge lower prices as a competitive measure (Wollan,
1998). Smaller wineries are not as efficient, but may
charge prices. Small and medium sized wineries most
likely must charge a higher price for their wines in order
to make a profit, and yet they also must use distributors
rather than sell direct. We have no evidence that the
wines from small and medium sized wineries are
perceived to be better than the wines from large wineries,
but it may be that these sized wineries are large enough
to develop a national reputation and yet are perceived as
relatively small or as boutique by consumers. If this is
the case, then wineries within these size groups can
benchmark their prices against others and some of them
may find that they are undercharging and can raise prices
and profits. The price premiums are substantial at
between 10 and 18% across the various sizes compared
to the largest two categories of wineries. Also, wineries
that are larger cold take advantage of consumer’s
perceptions by positioning individual brands as coming
from boutique-style wineries.
One of the key areas of interest is the effect of region on
wine price. We decided to separate the varieties and run
an equation for each, so that we could see what effect
regional reputation has on the wines by variety. Even
though there are no restrictions of what can be grown
where in Australia, anecdotally, different regions have
better reputations for different varieties of wines. Our
research gives weight to that observation. Barossa for
shiraz, Coonawarra for cabernet and Clare Valley for
riesling all show increased value compared to other
regions producing that variety. Along with Schamel and
Anderson’s observation (2001) that region is gaining in
importance in Australia, we find that different regions
certainly are valued more highly for different grape
varieties than others.
From a marketing manager’s point of view our results
provide some direction for both promotion and pricing.
There certainly is evidence that region can add to the
price consumers are willing to pay for wine. Wineries
from regions with a comparably positive price impact
should utilize that fact on the label and in promotions.
Probably the most important advice is for wineries in
each region to keep working together. This collaborative
effort is relatively unique in Australia, but seems to have
borne fruit and can continue to do so. Wineries with
strong reputations for specific varieties can emphasize
this fact in group promotions.
More importantly, regions where the wine quality is
good, but suffer from lower prices need to engage in
activities to build the reputation for their regions wines,
and consequently be able to raise prices. McLaren Vale
is a good example. Here the shiraz wines have won many
awards and are highly demanded within Australia.
However, the reputation for the region is lower than for
Barossa, and this is especially true in export markets
(Murphy 2001). Some idea of the price increase can be
predicted, which would help decide what a regional
promotional budget might be over a period of years.
McLaren Vale wineries have begun to promote their
wines in Australia with festivals, dinners, and a new
visitor’s center. However, the prices paid domestically
for Australian wines are partly determined by their
international value, and obviously McLaren Vale needs
to do more international promotion. At the beginning of
the 1990s, Barossa Valley wineries made several trips as
a group to the UK market to conduct tastings, dinners,
music and wine events, and Barossa food and wine
events. McLaren Vale should follow in this same mode.
With prices 13% lower than Barossa, but similar climate
and quality, wineries in McLaren Vale can calculate the
returns to improving their reputation through
promotional activities. Although these are not as high as
those returns to winery reputation, they are substantial.
The same advice can be given to the other newer regions
with good quality, but relatively lower prices. Regional
promotions are less expensive per winery than individual
wineries conducting the same events and over a period of
time, raising the regional reputation will help all the
wineries achieve higher prices.
Of course hedonic price analysis does have some
shortcomings. We must assume that the quality and
reputation ratings have validity, since they are subjective
judgments. We used James Halliday (1999), one of the
foremost authorities on Australian wine. We could use
the same logic as Schamel and Anderson’s (2001) and
accord Halliday influence status, so that his wine and
winery ratings actually help create the prices charged. In
our opinion, the quality of the wine and the winery
reputation do have a relationship with price, regardless
of which is cause and which is effect. We also must
assume a balance of supply and demand for the
equations to have meaning.
There is scope for future research. In the area of hedonic
price analysis, a simple but powerful confirmation would
be to use a different set of quality and reputation
indicators, like Schamel and Anderson (2001) did, which
showed similar relationships. One might also use
Components of Wine Prices for Australian Wine
30 Australasian Marketing Journal 11 (3), 2003
separate equations for each variety over a historical
period in order to confirm Schamel and Anderson’s
observation that overall region is gaining in importance
in Australia. Outside of hedonic price analysis,
conclusions on the various indicators of value, like
winery reputation, wine quality, and region could be
tested with different techniques. Preliminary research by
Tustin and Lockshin (2001) showed that region and
brand did have a significant impact on consumer choice,
but the pattern differed for low and high involvement
consumers. Awell known wine region added about $5.00
to the price consumers were willing to pay for a bottle in
a choice-based conjoint experiment. More research using
actual choice situation experiments could confirm the
effects of the variable found significant in this analysis.
This is important for wineries, so management can
decide whether focusing on the brand or the region has
the largest payback in price received.
It is also important to know what consumers actually have
in their minds when choosing wine. Do they have
Halliday or other quality ratings in their mind when
buying wine? What is the effect of putting those ratings on
the shelf? This is done more regularly in the US with Wine
Spectator ratings than in Australia. How many consumers
use winery reputation as part of their decision process?
Looking at the process a different way, does price act as a
signal, rather than as an outcome of the other factors?
Does merely setting a higher price result in greater sales or
a higher reputation? These are just a few of the useful
questions remaining to be answered in future research.
7. References
Anderson, K., Norman, D., Wittwer, G., 2001.
Globalization and the world’s wine markets: An overview,
CIES Workshop Paper 01, Centre for International
Economic Studies, University of Adelaide, Adelaide.
Arguea, N., Hsiao, C., 1993. Econometric issues of
estimating hedonic price functions. Journal of
Econometrics 56, 243-67.
Ashenfelter, O., Ashmore, D., Lalonde, R., 1995.
Bordeaux wine vintage quality and the weather. Chance
8 (4), 7-14.
Australasian Wine Exchange, 2000.
http://www.aawx.com.au/newsletters/news001.html,
Website accessed on 11 December 2002.
Australian Wine and Brandy Corporation (AWBC),
2001. Wine Export Approval Report, Winetitles
Publisher, Adelaide.
Australian Wine and Brandy Corporation (AWBC),
1994. Guidelines for applicants for the determination of
Australian geographical indications, Winetitles
Publisher, Adelaide.
Charters, S., Lockshin, L., Unwin, T., 1999. Consumer
responses to wine bottle back labels. Journal of Wine
Research 10 (3), 183-196.
Combris, P., Lecocq, S., Visser, M., 2000. Estimation of
a hedonic equation for Burgundy wine. Applied
Economics 32, 961-967.
Combris, P., Lecocq, S., Visser, M., 1997. Estimation of
a hedonic price equation for Bordeaux wine: Does
quality matter? The Economic Journal 107, 390-402.
Doornik, J.A. and Hendry, D.F., 1996. Empirical
Econometric Modelling Using PcGive 9.0 for Windows.
London, International Thomson Business Press.
Greene, W.H., 1993. Econometric Analysis, Macmillan
Publishing Company, New York.
Halliday, J., 1999. Australia and New Zealand Wine
Companion 2000, Harper Collins, Sydney.
Johnson, J., 1972. Econometric Methods, McGraw-Hill
Book Company, New York.
Landon, S., Smith, C.E., 1997. The use of quality and
reputation indicators by consumers: The case of
Bordeaux wine. Journal of Consumer Policy 20, 289-323.
Landon, S., Smith, C.E., 1998. Quality expectations,
reputation, and price. Southern Economic Journal 64 (3),
628-647.
Lockshin, L., 1997. Branding and brand management in
the wine industry. Australian and New Zealand Wine
Industry Journal 12 (4), 386-387.
Murphy, H., 2001. Personal Communication of the
results of a survey of wine consumers at Australian
tastings in the UK.
Nerlove, M., 1995. Hedonic price functions and the
measurement of preference: The case of the Swedish
wine consumers. European Economic Review 36, 1697-
1716.
Oczkowski, E. 2001. Hedonic Wine Price Functions and
Measurement Error. The Economic Record 77 (239),
374-82.
Oczkowski, E., 1994. A hedonic price function for
Australian premium table wine. Australian Journal of
Agricultural Economics 38, 93-110.
Components of Wine Prices for Australian Wine
Australasian Marketing Journal 11 (3), 2003 31
Olson, J.C. and Jacoby, J., 1972. Cue utilization in the
quality perception process. Proceedings of the Third
Annual Conference, Association for Consumer
Research, Ann Arbour, pp. 167-179.
Rasmussen, M. and Lockshin, L., 1999. Wine choice
behaviour: The effect of regional branding. International
Journal of Wine Marketing 11 (1), 36-46.
Rosen, S., 1974. Hedonic prices and implicit markets:
Product differentiation in pure competition. Journal of
Political Economy 82, 34-55.
Ryan, D., 1994. The protection of geographical
indicators in Australia under the EC/Australia wine
agreement. European Intellectual Property Review 16
(2), 521-524.
Schamel, G. and Anderson, K., 2001. Wine Quality and
Varietal, Regional, and Winery Reputations: Hedonic
Prices for Australia and New Zealand. CIES Workshop
Paper 20, Centre for International Economic Studies,
University of Adelaide, Adelaide.
Shapiro, C., 1983. Premiums for high quality products as
returns to reputations. Quarterly Journal of Economics
98 (4), 659-679.
Tustin, M., Lockshin, L., 2001. Region of origin: Does it
really count? The Australia and New Zealand Wine
Industry Journal 16 (5), 139-143.
Unwin, T., 1999. Hedonic price indexes and the qualities
of wine. The Journal of Wine Research 10 (2), 95-104.
White, T., 1995. Fine lines in Coonawarra. Financial
Review, Friday, 17th March, 15.
Winetitles, 2001. The Australian & New Zealand Wine
Industry Directory. Winetitles Publsiher, Adelaide.
Wollan, D., 1998. The Australian Wine Industry- a
Sceptic’s view? The Australia and New Zealand Wine
Industry Journal 13 (1), 22-24.
Biographies
Bith-Hong Ling is an associate professor in the
Department of Marketing at the National Chung Hsing
University in Taiwan. She received her Masters in
Agribusiness from the Arizona State University and a
Ph.D. in Agricultural and Resource Economics from the
University of Hawaii. Her research interests are in the
areas of wine economics, marketing in the Asia-Pacific,
global marketing strategy, international business and
brand management. Bith-Hong holds a 2003 visiting
fellow of Dentsu Ikueikai and Yoshida Hideo Memorial
Foundation (Japan) in the Center of Business and Market
Studies of the Faculty of Economics at the University of
Tokyo and works on the brand choice decision and
strategic brand marketing.
Larry Lockshin is Professor of Wine Marketing at the
University of South Australia. He has spent more than 20
years working with the wine industry, first as a
viticulturist and now as a marketing academic and
consultant. He received his Masters in Viticulture from
Cornell University and a PhD in Marketing from Ohio
State University. His research interests are consumer
choice behaviour for wine and wine industry strategy.
Dr. Lockshin is the Associate Editor for Wine Marketing
for the Australia and New Zealand Wine Industry Journal
and a regular marketing columnist for the last 5 years. He
writes monthly for Harpers, The Wine and Spirit Weekly,
in the UK and he is also the Associate Editor for
Business and Economics for the Journal of Wine
Research.
Correspondence Addresses
Dr. Bith-Hong Ling, Associate Professor, Department of
Marketing, National Chung-Hsing University, 250 Kuo Kuang
Road, Taichung, Taiwan, R.O.C., Telephone: +886 (4) 2284
0392 ext 43, Facsimile: 886 (4) 2285 6691, Email:
bhling@dragon.nchu.edu.tw; Dr. Larry Lockshin, Professor
and Director, Wine Marketing Research Group, University of
South Australia, GPO Box 2471, Adelaide, South Australia
5001, Telephone: +61 (08) 8302 0261, Facsimile: +61 (8) 8302
0442, Email: larry.lockshin@unisa.edu.au
Components of Wine Prices for Australian Wine
32 Australasian Marketing Journal 11 (3), 2003
