67 OMURA EBIHARA SAKURAI - Soyons futiles

ID=67 COIMBRA & VISEU Introduction
This paper attempts to understand the mechanism of this upward trend by analysing the trends in
2012 - PORTUGAL
wine consumption and analysing the correlations between wine consumption and other household level
An Analysis of Wine Consumption Trends and Food-Related Expenditures in Japan expenditures. For the wine consumption trends, we briefly look at the trends for domestic, imported and
OMURA Makiko, EBIHARA Kensuke, SAKURAI Yuka total wine consumptions from 1970 to 2009. In the next section, we look at the aggregate data on household
Meiji Gakuin University - JP expenditure patterns since 1970, followed by the analyses on the possible correlations between wine
consumption and other food-related-expenditures, applying several different estimation models. Then we
[email protected], [email protected], [email protected] utilise the data on food service industry to explore the inter-linkages between wine consumption and food
service industry. A concluding remark is given at the end, offering a hypothesis derived from this paper.
Although Japanese economy has been experiencing recession since early 1990, wine consumption
has been remaining relatively stable. In fact, looking at the growth rate, wine consumption in Japanese A Brief Background on Wine Consumption
market has a fairly steady upward trend, with significantly higher average growth rate compared to the Wine consumption in Japan has a general upward trend for the period between 1970 and 2010,
GDP or many other consumption goods. This is particularly so for imported wines. Although Japanese
consumers do not drink as much wine as their Westerns counterparts, such a growing trend may suggest a although the trends differences are seen between domestic and imported wine. The trends of total wine
possibility of wine steadily gaining its place in Japanese life. In particular, wine consumption may be related consumption and wine consumption per capita, for those of drinking age of 20 years and up, are very similar.
to the westernisation of cuisine in Japan. Whilst the analyses do not give strong evidence of correlation While the domestic wine consumption has more steady increase, the imported wine consumption has a
between wine consumption and other food item consumption at a household level, wine consumption is sharper increase, with a peak point in 1998 caused by the red wine polyphenol boom in Japan, when a large
found to be significantly positively correlated with expenditures outside the household, in particular, quantity of wine was imported in Japan.
expenditure on eating-out and social-expenses. The results from the analyses on food service industry sales
may suggest younger generations, reasonably-priced wines, and casual eating occasions outside household
being keys to the future wine consumption in Japan, contrary to the trends recently experienced by France.

Graph1a: Wine Consumption Trend in Japan: 1970-2009

Domestic & Imported Wine

consumption in HKL Wine Consumption Wine Consumption Per Capita
100 200 300
consumption in L
123

0 0

1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
year year

Consumption Domestic W ine Consumption Imported Wine
Consumption Sum

Note: Per Capita concerns age of 20~79
Source: Japan National Taxation Agency; Population Census

©www.vdqs.net/2012Coimbra
1

©www.vdqs.net/2012Coimbra

Graph1b: Wine Consumption (log quantity) in Japan: 1970-2009 Graph2: Wine Consumption (log quantity) Trends in Japan: 1970-2009
Domestic & Imported Wine Domestic & Imported Wine

Wine Consumption (log) Wine Consumption Per Capita (log) Domestic Wine Consumption Imported Wine Consumption

log quantity 10 12 14 log quantity log quantity 10 12 14 log quantity 10 12 14
-4 -2 0 2
8

8 8

6

1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 log dom. wine cons.= -109.3802 + 0.0604 year log imp. wine cons.= -237.9044 + 0.1248 year
year year (0.0055) (0.0066)

6

lncdome lncimp 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
lncsum year year

Note: Per Capita concerns age of 20~79 bubble_econ lncdome bubble_econ lncimp
Source: Japan National Taxation Agency; Population Census Fitted values lo/hi Fitted values lo/hi

Data Source: Mercian Corp, with original data form the Japanese National Tax Agency

Plotting the regression of wine consumption on year, the growth of wine consumption has different Graph3: GDP (log) Trends in Japan: 1970-2009
trend for domestic and imported wine (Graph2). Both have high and increasing growth rates especially over Real GDP & Real GDP Per Capita
the 1970s, with its growth rate becoming fairly stable with some bumps in certain years, especially for
domestic wine. After 1988, the consumption growth of domestic wine has decreased, while that of imported Real GDP (log) Real GDP Per Capita (log)
wine remained stable. The regression coefficient of year, which is significant at the 1% level for both cases, is
the estimated annual percentage growth rate of wine consumption, which is 6% and 12.5% for domestic and 12 12.5 13 13.5 14 1 1.5 2
imported wine, respectively (Graph2 box). This growth rate is much higher than the country’s real GDP and
real GDP per capita growth rate during the same period, which is 2.8% and 2.3%, respectively (Graph3 box). ln GDP = -42.1775 + 0.0277 year log GDPpc.= -44.4157 + 0.0229 year
The bump in the late 1980s to early 1990s highlighted in blue shade indicates the period of Bubble Economy, (0.0013) (0.0011)
which does coincide with one of the bumps in the wine consumption trends. These observations suggest that
the increase in wine consumption is likely to be positively affected by the economic growth, but there are
likely to be other factors accelerating higher growth rate of wine consumption during these period. Although
the economic growth has been stagnant after the Bubble burst, from around 1993, wine consumption
growth did not seem to be much affected, at least in terms of quantity.

.5

1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
Year Year

bubble_econ lngdprl bubble_econ lngdprlpc
Fitted values lo/hi Fitted values lo/hi

Data Source: Japan Statistics Bureau

©www.vdqs.net/2012Coimbra ©www.vdqs.net/2012Coimbra

Household Expenditure Pattern in Japan since 1970 The pattern in Grah4 for shows increasing expenditures of all goods categories up until the end of the
In order to explore possible factors affecting wine consumption in Japan, it would be worthwhile to Bubble, thereafter the expenditure pattern either stabilises for most categories apart from housing&utilities
which continues to go up even at a similar pace, and clothing&footware which reverts to a decreasing trend.
examine other goods consumption pattern during the same period. Japan has experienced a rapid economic There seems to be no particular trend differences between the basic needs goods and other goods in terms
growth especially for the period of 1954-1973, with the following period still exhibiting a high growth up until of the growth rate (log graph in second panel of Graph4), apart from the clothing&footware. Looking at the
the collapse of the Bubble in early 1993. ratio of expenditure to total household expenditure in Graph5, we see that the ratio of food&non-alcohol has
a decreasing trend while house&utilities has an increasing trend. Excluding these two which makes easier to
First, we shall look at the household expenditures patterns of general goods, then on the food-related see the trends of other more minor expenditure categories, a sharp decrease in clothing&footware
expenditure/consumption pattern in particular. The household expenditures are largely categorised in the expenditure is pronounced especially after the Bubble, and a slight decrease in alcohol&tobacco
following categories: (1) food and non-alcoholic beverages; (2) alcoholic beverages and tobacco; (3) clothing expenditure. Unfortunately, we cannot discern the composition of alcohol&tobacco with this annual data.
and footware; (4) housing and utilities; (5) furniture and household equipment; (6) medical expenses; (7) Thus we will now look at data on average monthly consumption expenditures of working households, which
transport; (8) communication; (9) leisure; (10) education; (11) eating out & trips; (12) other.1 Graph4 shows have more detailed categories.
the expenditure pattern for categories (1), (2), (3), (9) and (11). It can be said that a large part of (1)
composes a necessary expenditure, where the price elasticity of demand is generally low, whilst (2), (3) (9) Amongst various categories of household expenditures, we focus on the food-related categories that
and (11) have more features of non-necessary expenditure, and (4) being mixed. may be considered to correlate with alcohol beverages, either positively or negatively. These food items are:
(1) staple foods - (1a) rice, (1b) bread; (2) main dish - (2a) meat, (2b) fish & shellfish; (3) drinks – (3a) non-
Graph4: Household Expenditure Pattern in Japan: 1970-2009 alcohol beverages, (3b) alcohol beverages; (4) food consumption outside the household - (4a) eating-out,
(4b) social-expenses on food. Graph6 exhibits two panels, one in real expenditure and another in log
Selected Goods Categories (current price) expenditure. A notable trend is an increasing trend for eating-out with an average annual growth of 3.8%.
Whilst social-expenses on food show a decreasing trend after the Bubble, household eating-out trend
expenditure in billion yen Household Expenditure Pattern Household Expenditure(log) Pattern remains stable. We see that rice and fish consumption are on its decreasing trend from the mid-1980s to
20000 40000 60000 80000 early-1990s, while bread consumption has increased significantly during this period. This may suggest
log expenditure 9 10 11 12 increasing westernisation of food at the household level, which we shall consider more in detail in the
following section.
8
©www.vdqs.net/2012Coimbra
0 7

1970 1980 1990 2000 2010 1970 1980 1990 2000 2010
year year

bubble_econ food&non alcohol alchol_bev&tobacco cloth&footware
house&utilities leisure eatingout&trip

Data: Composition of Final Consumption Expenditure of Households Classified by Purpose
Data Source: Japan Statistics Bureau

1 There is no coherent data available for the whole period in concern, thus the data for 1970-1979 and 1980-2009 differ in
terms of tis categorisation and definition for some. In particular, categories (1) and (2), (7) and (8), (9) and (10), (11) and
(12) were put together in the earlier years’ data. We extrapolate the categorical figures by applying the categorical
distribution figure of 1980, as an approximation. As for (6), medical expenses, there is a definitional change so the
figures do not match in these data sets, although this category serves us a little interest in any case.

©www.vdqs.net/2012Coimbra

Graph5: Household Expenditure Pattern (%) in Japan: 1970-2009 Correlation Between Wine Consumption and Other Food-Related Expenditures
Selected Goods Categories (% of total hh exenditure) Obviously, we cannot determine whether the westernsation of food leads to an increasing wine

Graph5a: Household Expenditure Pattern (%) Graph5b: Household Expenditure Pattern (%) consumption, as wine can also be drunken with non-western food as well. Also, same food items can be used
in various cuisines, so we cannot conclude westernisation of food just from our data. Nonetheless, it would
expenditure ratio expenditure ratio be worthwhile to examine possible correlations between wine consumption and other food items as well as
.05 .1 .15 .2 .25 .05 .1 .15 other goods.

0 Given the fact that consumptions patterns exhibit autocorrelations, we apply several models in order
to explore the possible correlations between wine consumption and other commodities.2 Namely, we
1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 estimate the following two models: (1) a first differenced model - regressing the first difference (FD) of wine
year year consumption on the first difference of other food-related goods; (2) an autoregressive (AR) model -
regressing wine consumption onto its previous year’s consumption and onto other food-related item
bubble_econ food&nonalcohol alcohol_bev cloth&footware consumption. Both models apply a dummy variable for the boom of 1998.3 Given the fact that wine is often
house&utilities leisure eatingout taken with food, we may perhaps see some correlation between certain kinds of food consumption trends
and that of wine.
Data Source: Japan Statistics Bureau
(1) FDWineConsum(t) = γFDOtherConsm(t) + y1998 + ϵ (FD Model),
Graph6: Household Expenditure Pattern in Japan: 1970-2009 (2) where FDWineConsum(t)= FDWineConsum(t)- WineConsum(t-1) and,
(3) FDOtherConsm(t)=OtherConsm(t) - OtherConsm(t)
Selected Goods Categories (4) WineConsum(t) = α + βWineConsm(t-1) + γ OtherConsm(t) + y1998 + ϵ (AR Model).

Graph6a: Household Expenditure Pattern Graph6b: Household Expenditure(log) Pattern With the FD model examining various other goods expenditure, household utilities and
15000 eatingout&trips were found to both have positive significant impact on the overall wine consumption at the
log expenditure 5% significance level. Analysing another data on the working household expenditure which has more detailed
expenditure in yen (current) 10000 7 8 9 10 categories, we again have a significant positive coefficient for eatingout, as well as social-expenses-on-food,
whilst no particular food item seems to exert any significant correlations (Table1a shows only selected food
5000 items). Looking separately at domestic and imported wine, the coefficients on these two variables, eatingout
and social-expenses-on-food, are identical, although the coefficients for y1998 dummy is much larger for
0 6 imported wine, with no significance found for domestic wine (Table1b).

1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 The AR model estimation (Table2a) gives similar results to the FD estimation for eatingout and social-
year year expenses-on-food, whose coefficients are both positive and significant at 5% significance level, although
the magnitudes of coefficients are both minimal. For total wine consumption, the estimated coefficient for
bubble_econ rice bread fish meat the lagged-dependent variable is positive and highly significant at 1% level, with its magnitude being large
beverage alcohol_bev eatingout socialexp_food ― a one unit increase in previous year’s consumpƟon will give a nearly one unit rise in this year’s
consumption. The impact of 1998year dummy is pronounced, with its significance level at 1%. The results
Note: Average Monthly Expenditure Data for Working Households.
Data Source: Japan Statistics Bureau 2 As we are using annual data, seasonality is not an issue, however trends are clearly not stationary as shown in Graph6.
Both Dickey-Fuller test and Phillips-Perron test for wine consumption and other goods suggest a presence of unit root,
©www.vdqs.net/2012Coimbra i.e., stochastic trends, for domestic, imported, and total wine consumption, and majority of other goods. No
cointegration is suggested between wine consumption and other goods consumption.
3 Note that we cannot run autoregressive distributed-lag (ADL) model due to high multicollinearily.

©www.vdqs.net/2012Coimbra

not only confirm a high impact of the 1998 polyphenol boom and autoregressive nature of wine (1.90)*
consumption, but also suggest a possible increase in wine consumption due to consumption outside the
household. This matches the general observation made by Amine et Lacoeuilhe (2007) that wine is being Eatingout 0.0002
drunken at socialising occasions. Social exp_food (2.60)**

y1998 73.5696 73.51 72.1241 72.7854 71.6449 72.8168 0.0004
(5.50)*** (2.14)**
Table1a. FD Estimation of Wine Consumption (Total) and Other Food Related Expenditures: 1970-2009 (5.08)*** (5.40)*** (5.03)*** (5.11)*** (5.19)*** -7.5196 70.6154
(-1.02) (5.17)***
(Model1) First Difference of Wine Consumption Constant 12.6227 -7.0269 -0.6413 -3.7882 -9.8462 0.9822 -8.701
0.9807 (-0.95)
Fish&shell Alcohol Social -0.91 (-0.83) (-0.06) (-0.30) (-0.91) 0.9813
fish beverages expe_food 39 0.9797
Bread 0.007 Meat Eating out R-squared 0.9788 0.9813 0.9794 0.9795 0.9808
(1.21) 0.02 0.024 39
First Difference (xs) 0.033 0.011 (1.46) 0.015 (2.97)*** Adj-R-squared 0.977 0.9796 0.9776 0.9777 0.9791
(1.4) (1.55) (2.91)***
N 39 39 39 39 39

y1998 73.24 74.544 73.593 72.72 72.547 79.329 Note: t statistics in parentheses; * p<0.1, ** p<0.05, *** p<0.01

(5.05)** Data: Average Monthly Expenditure for Working Households, Japan Statistics Bureau
*
(5.09)*** (5.11)*** (5.03)*** (5.41)*** (5.86)***

R-squared 0.425 0.418 0.432 0.428 0.508 0.511 Analysis of Wine Consumption and Food Service Industry (FSI)
Given the possible correlation between wine consumption and eating-out and social-expenses on
Adj-R-squa 0.394 0.387 0.402 0.397 0.481 0.485
N 39 39 39 39 39 39 food, we first examine the graph of wine consumption and different types of food-service industry (FSI). For
food industry, we utilise two types of data: (1) data from Foodservice Industry Research Institute; (2) data
Note: t statistics in parentheses; * p<0.1, ** p<0.05, *** p<0.01 from Japan Food Service Association. The former provides data for the period of 1975-2010 on actual sales
(in current yen) for various categories within the FSI. We conduct analyses on selected categories that might
Data: Average Monthly Expenditure for Working Households, Japan Statistics Bureau have either positive or negative correlations with the wine consumption, namely: diner & restaurant, sushi,
domestic airplane, hotel, cafe, pub, bar & nightclub, ryotei (traditional Japanese-style restaurant), and take
Table1b. FD Estimation of Wine Consumption and Eating-out/Social-Expenditure on Food: 1970-2009 out. The latter provides data in the form of “percentage-change (Δ%) from the previous year” for sales,
customers, shop numbers and average spending per customer, for the following categories, fast-food, family-
(Model1) restaurant, pub, restaurant and café.

Wine total Domestic wine Imported wine Graph7 provides two graphs on FSI sales in current price and log, with an added line for the total wine
consumption in quantity due to the unavailability of sales data. We see a fairly steady increase of
D.eatingout 0.001 0.001 0.001 diner&restaurant sales up until 1997 followed by decrease and stagnation. This interestingly seems to fairly
D.social exp_food coincides the trend of wine consumption. On the other hand, we see a pronounced decrease in sales trend
y1998 (0.00)*** (0.00)** (0.00)* for hotel and nightclub after the burst of Bubble Economy, suggesting that these types of spending became
increasingly unaffordable. Another notable trend is a steady increase of takeout sales. Graph 8 shows the
0.001 0.001 0.001 yearly trends of Δ% in FSI, with or without Δ% in wine consumption in each panel. The left panel, Graph8a
(0.00)** (0.00)*** shows particularly high peaks for 1997 and 1998’s Δ% wine consumption. Excluding wine consumption in the
(0.00)** right panel, we see more clearly the trends in FSI. Generally speaking Δ% from a previous year seems to
72.08 79.329 fluctuate in the range of (-8%, +13%). A relatively sharp consecutive decrease in 2008 and 2009 for all FSI
72.547 4.022 7.249 68.524 (8.71)*** (13.53)*** categories perhaps reflects the Lehman Shock which was triggered by the company’s collapse in September
(13.41)*** (7.95) 2008.
(8.24) (8.82)***

R-squared 0.508 0.171 0.129 0.637 0.654 0.511
Adj-R-squa~d 0.481 0.126 0.082 0.617 0.635 0.485
N
39 39 39 39 39 39

Note: t statistics in parentheses; * p<0.1, ** p<0.05, *** p<0.01
Data: Average Monthly Expenditure for Working Households, Japan Statistics Bureau

Table2a. AR Estimation of Wine Consumption on Previous Wine Consumption and Other Food Related

Expenditure: 1970-2009

(Model2) Wine Consumption (total)

Lag-consumption 0.951 0.8986 0.9602 0.9532 0.9335 0.8762 0.928

(27.58)*** (25.00)*** (38.54)*** (38.47)*** (34.91)*** (22.79)*** (34.30)***

Rice 0

(-0.24)

Bread 0.001

(2.14)**

Fish & shellfish 0.0001

-0.96

Meat 0.0002

-1.08

Alcohol 0.0006

©www.vdqs.net/2012Coimbra ©www.vdqs.net/2012Coimbra

Graph7: Wine Consumption and Food-Service Industry Sales: 1975-2009 Next, we analyse the correlations between wine consumption and FSI with the following estimation
Selected Food-Service Categories equations with multiple FSI-sales covariates or a single FSI-growth-rate covariate, with 1998-year dummy:

sales (in 100million yen) & quantity (HKL) Graph7a: Wine Cons. & FSI Sales Graph7b: Wine Cons. & FSI Sales (log) (5) FDWineConsum(t) = α + γFDFSI (t) + y1998 + ϵ (FD Model),
0 20000 40000 60000 80000100000 (6) WineConsum(t) = α + βWineConsm(t-1) + γFSI (t) + y1998 + ϵ (AR Model),
log sales & log quantity10 12 14 (7) GRWineConsum(t) = α + γGRFSI (t) + y1998 + ϵ (GR/FD-log Model),

8 where FD and GR signifies first-difference and growth-rate (log difference), respectively, and FDFSI is a
vector of all first-differenced FSI sales listed above except for bar & nightclub which is omitted due to
6 multicollinearity problem. Looking at Table3, which shows the results of multivariate estimation of different
types of wine on various FSI sales, for both FD and AR estimations, we find the significant positive coefficient
1975 1980 1985 1990 1995 2000 2005 2010 1975 1980 1985 1990 1995 2000 2005 2010 for diner&restaurant for all estimations, suggesting the result robustness. The magnitude of coefficient is
year year much higher for a lagged dependent variable in all three types of wine consumption, i.e., total, domestic, and
imported. We also find, rather strangely, the coefficients of sushi being negative and of café being positive,
bubble_econ sales_dinerresto sales_sushi sales_other both significant at 5% level. Indeed, it should be noted that there are signs of multicollinearity for the AR
sales_domeplane sales_hotel sales_cafe sales_pub estimation, judging from the variance inflation factor analysis. Indeed, looking at univariate estimation of
sales_barnightclub sales_takeout Consumption Sum growth-rate of wine on growth-rate of FSI (Table4), diner&restaurant, sushi, pub and bar are found to be
positively correlated with wine consumption at 5% significance level which is a more understandable result.
Note: Household Expenditure Data; Food-Service Industry Market Size Trend Although we do not have the data in hands, most sushi restaurants now stock wine, particularly white wine,
Data Source: Japan Statistics Bureau; Foodservice Industry Research Institute indicating the existence of demand. The sign of coefficient is opposite for multivariate AR and univariate GR
estimation for sushi. Also café is no longer found significant in GR model. Although growth-rate of sales is not
Graph8: Wine Consumption and Food-Service Industry Sales: 1993-2009 same as autoregressive sales, it does suggest possible specification problem of multivariate AR model. Whilst
Percentage Chagnge from Previous Year 1998year dummy is found significant and positive for total and imported wine, it is not found so in domestic
wine for both FD and AR models.
Graph8a: Wine Cons. & FSI Sales Graph8b: FSI Sales
% change
% change .9 .95 1 1.05 1.1 1.15 Table3. Multivariate Estimation of Wine Consumption on FSI (FD & AR models): 1975-2009
1.1 1.2 1.3 1.4
(Model3) FD estimation (Model4) AR estimation
Wine total Domestic Imported Wine total Domestic Imported

Lag-dependent 0.753 0.458 0.597

(5.49)*** (2.50)** (4.07)***

1 diner&resto 0.006 0.003 0.003 0.006 0.003 0.002
sushi (2.67)** (1.99)* (2.36)** (4.32)*** (3.56)*** (2.86)***
-0.006 -0.009
.9 (-0.42) (-1.08) 0.003 -0.016 -0.019 -0.007
-0.35 (-1.60) (-2.70)** (-1.08)

1995 9798 2000 2005 2010 1995 2000 2005 2010 dome air -0.038 -0.016 -0.022 -0.04 -0.057 -0.02
hotel (-0.58) (-0.42) (-0.52) (-0.84) (-1.69) (-0.69)
year year café -0.002 -0.001 -0.001 -0.001 0.001 -0.001
(-1.27) (-1.37) (-0.73) (-0.66) -0.51 (-0.43)
pchng_fastfood family rest pchng_pub pchng_restaurant 0.002 0.003 -0.002 0.002 0.005 0.001
pchng_cafe pchng_other wine consm

Note: Food-Service Industry Market Trend Survey (0.36) (1.20) (-0.55) (0.59) (2.30)** (0.36)
Data Source: Japan Food-Service Association
pub -0.008 -0.001 -0.007 -0.004 0 -0.003
ryotei (-0.70) (-0.18) (-0.93) (-0.46) (-0.07) (-0.51)
0.01 0.013 -0.003 0.003 0.015 0.002

©www.vdqs.net/2012Coimbra ©www.vdqs.net/2012Coimbra

takeout (0.47) (1.04) (-0.23) (0.18) (1.30) (0.22) sufficient care, and yet the price starts from as low as 100 yen a glass.4 They also claim that they are the
0.001 0.003 -0.002 -0.002 0.000 0.000 largest Italian wine importer amongst the Italian food restaurants. Given the fact that this family restaurant
y1998 (0.45) (2.02)* (-1.17) (-1.26) (0.20) (-0.30) chain offers reasonably-priced menu and is popular especially amongst the younger generations, it can be
72.215 -13.805 86.02 63.808 2.346 61.22 inferred that younger generations and casual eating-out occasions are the key to the future wine
Constant (3.10)*** (-1.01) (5.82)*** (5.24)*** -0.28 (7.68)*** consumption in Japan, contrary to the trends in France where younger generations are increasingly unlikely
-5.496 -5.602 0.106 1.337 -18.061 -6.465 to drink wine, and where wine is becoming more a drink of the wealthier class, taken less regularly (Amine
R-squared (-0.98) (-1.70) (0.03) (0.03) (-0.66) (-0.25) and Lacoeuilhe, 2007; Aurier, 2007; Ebihara and Omura, 2009). This market trend in Japan may warrant
Adj-R-squared 0.627 0.41 0.747 0.991 0.953 0.993 further investigation.
N 0.465 0.153 0.637 0.986 0.931 0.989

34 34 34 35 35 35

Table4. Univariate Estimation of Wine Consumption on FSI (FD log): 1975-2009 Table5. Wine Consumption on FSI (PC from Previous Year): 1993-2009

(Model5) Growth Rate Wine Consumption (total) (y) (Model6) (Model7) AR

Diner&r Sushi Othe Dome Hotel Café Pub Bar Ryotei takeo Wine Consumption Percentage-Change (PC) from the Previous Year
esto 0.676 r _air 0.436 0.576 0.599 ut
Growth 0.906 (2.08)** 0.632 0.388 (1.75)* (2.09)** (2.13)** 0.737 Wine total Domestic Imported Wine total Domestic Imported
Rate_FIS (x) 0.29 (1.48 0.293
(2.91)** (0.79 (1.99)* (1.77)* Lag-wine_type 0.057 0.057 0.057
* ) (1.58)
) 0.236 (0.26) (0.26) (0.26)
0.195
y1998 0.283 0.257 0.227 0.243 0.25 0.24 0.257 0.257 (2.40)** Fastfood 0.479 -0.727 1.43 0.541 0.541 0.541
(1.93)*
(2.23)* (2.48)* 0.049 (0.40) (-0.28) (1.09) (0.41) (0.41) (0.41)
* * (2.82)** 0.03
(3.03)*** (2.63)** (2.49)** (2.44)** (2.63)** (2.64)** family 3.33 2.371 4.568 3.253 3.253 3.253
* (1.21) restaurant
Constant 0.018 0.037 0.045 0.032 0.05 0.053 0.041 0.041 0.212
0.162 0.198 (2.69)** (0.89) (3.38)*** (2.41)** (2.41)** (2.41)**
(2.79)* (3.10)** 0.146
(0.88) (1.92)* (1.93)* -1.51 ** * (2.22)** (2.27)** 34 dinner 0.72 0.284 0.953 0.827 0.827 0.827
34 restaurant
R-squared 0.319 0.239 0.15 0.231 0.19 0.211 0.24 0.243
(0.68) (0.12) (0.82) (0.69) (0.69) (0.69)
Adj-R-squar 0.276 0.19 0.095 0.182 0.138 0.16 0.191 0.195
Café -1.535 -1.319 -1.905 -1.511 -1.511 -1.511
N 34 34 34 34 34 34 34 34
(-1.66) (-0.67) (-1.90)* (-1.53) (-1.53) (-1.53)
Note: t statistics in parentheses; * p<0.1, ** p<0.05, *** p<0.01
Pub -1.408 -0.396 -2.155 -1.498 -1.498 -1.498
Data: Japan Statistics Bureau; Foodservice Industry Research Institute
(-1.52) (-0.20) (-2.13)* (-1.43) (-1.43) (-1.43)

y1997 0.422 0.393 0.431 0.425 0.425 0.425

For the second set of data on FSI, we utilize the following estimation equation for Δ% in FSI sales: (4.90)*** (2.13)* (4.60)*** (4.61)*** (4.61)*** (4.61)***
(8) Δ%WineConsum(t) = α + γΔ%FSI (t) + y1997 + y1998 + ϵ
(9) Δ%WineConsum(t) = α + βΔ%WineConsm(t-1) + γΔ%FSI (t) + y1997 + y1988 + ϵ y1998 0.354 0.046 0.633 0.338 0.338 0.338

The equation (6) simply regresses Δ% wine consumption onto Δ%FSI, whilst the equation (7) has an (4.20)*** -0.26 (6.90)*** (3.06)** (3.06)** (3.06)**
additional lag-dependent variable. Both estimation models have 1997 and 1998 year dummies, which has
extreme peaks of Δ% wine consumption. For this data set, we have a more detailed groups for Constant -0.505 0.859 -1.795 -0.589 -0.589 -0.589
diner&restaurant category than the first FIS data set used above, namely, fast-food, family restaurant, and
dinner restaurant. According to the results shown in Table5, apart from y1997 and y1998 dummies, the only (-0.37) (0.30) (-1.22) (-0.40) (-0.40) (-0.40)
variable that comes up significant at 5% level is family restaurant. This is rather unexpected since family
restaurant is regarded more as a place to eat rather than drink alcohol. The results are robust in terms of the R-squared 0.869 0.498 0.913 0.87 0.87 0.87
similarity of results amongst different estimations. Because the data is only in Δ% form, this may be a simple
coincidence. Nonetheless, it is also true that there are family restaurants that carry wine in their menu. A Adj-R-squared 0.754 0.058 0.837 0.721 0.721 0.721
well-known family restaurant that carries wine is called Saizeria which offers Italian food and wine. According
to the information published on their website, they offer ranges of Italian wines selected and imported with N 16 16 16 16 16 16

Conclusion
Although Japanese consumers do not drink as much wine as their Westerns counterparts, we see a

steady growing trend of wine consumption. As this may suggest a possibility of wine steadily gaining its place

©www.vdqs.net/2012Coimbra 4 See http://www.saizeriya.co.jp/corporate/effort/taste/ for their efforts and care on their offered wine, and see
http://www.saizeriya.co.jp/menu/wine.html and http://www.adda-tours.co.jp/ for their wine internet shopping (all web
sites are in Japanese).

©www.vdqs.net/2012Coimbra

in Japanese life, we have analysed in this paper the possible inter-linkages between the Wine Sector and the Common Market Reform: An Analysis on Consumer Behaviour and Production
consumption/expenditure patterns of wine and other goods, in particular food-related items. We have Adjustment Policy),” Meiji Gakuin Law Journal, 87: 23-62.
utilised a time-series data on wine consumption, household expenditure patterns, and food service industry WANDS. 2012. “2011/2012 Wine Market Review,” WANDS, April 2012: 8-25.
sales for 1970-2009 for most analysis, and for a shorter time period in case of data restriction. With time-
series data, several estimation models are employed. DATA SOURCES
Foodservice Industry Research Institute. 2012. Food-Service Industry Market Size Trend.
Whilst there are some differences in estimation results between the models, the analyses do not give Japan Food-Service Association. 2012. Food-Service Industry Market Trend Survey.
strong evidence that wine consumption at a household level is likely to be correlated with the consumption Mercian. 2011. “Wine Consumption Trend,” Wine Reference Material. (data based on the Japan National Tax
of other food items. Thus we cannot establish a hypothesis of linking food westernisation and wine Agency Reports; original reports were unavailable for verification).
consumption at the level of household food consumption. Nonetheless, recent studies investigating wine Statistical Survey Department, Statistics Bureau, Ministry of Internal Affairs and Communications. 1998,
sales by different types of shops suggest that home wine consumption is on its increasing trend since the 2009, 2010. Gross Domestic Expenditure.
Lehman Shock in late 2008, and that this trend is strengthened by the North-Eastern Japan Earthquake in _______. 2012. Final Consumption and Household Expenditure.
March 2011 (WANDS, 2012). On the other hands, through the long-term trend analysis we conducted, wine _______. 2012. Annual Average of Monthly Receipts and Disbursements per Household (Workers'
consumption is found to be positively correlated with expenditures on eating-out and social-expenses, Households).
suggesting that wine is consumed mostly outside household and amongst family, friends, and/or colleagues.
Therefore, we are not certain at this point whether a recent trend of consuming wine at home will be a ©www.vdqs.net/2012Coimbra
stable trend although it may well be so.

It is also estimated that diner and restaurant sales are the most important factor correlated with wine
consumption, rather than pub or other types of food-service industry. Looking closely at diner and restaurant
category, family restaurant is found to be significantly correlated with wine consumption. Such findings may
suggest younger generations, reasonably-priced wines and casual eating-out occasions being the key to the
future wine consumption in Japan, contrary to the trends recently seen in France. Indeed the study cited
above by WANDS (2012) finds that low price-range bottles (\500~\1,000) are on a rapid increase, now
extended to 49% share of all imported wines. While we give preliminary analyses on wine consumption and
other food-related expenditures in Japan, with a possible hypothesis regarding a future trend, we have not
really distinguished different types of wine, in terms of its type, origin and price range. These factors remain
to be investigated in our future research.

REFERENCES
Amine, Abdelmajid and Jérôme Lacoeuilhe. 2007. “Les Pratiques de Consommation du Vin: Rôle des
Représentations et des sitautions de cosommation,” Actes du XXIIIème Congrès International de l’AFM – 31
mai & 1er juin 2007, Aix-les-Bains.
Aurier, Philippe. 2007. Vins, Boissons et Contextes de Consommation : Une Analyse du Statut du Vin en
France, in Jean-Pierre Couderc, Hervé Hannin, François d' Hauteville et Etienne Montaigne (dir.), Bacchus
2008 : Enjeux, stratégies et pratiques dans la filière vitivinicole, Dunod.
Ebihara, Kensuke and Makiko Omura. 2009. “Oshu-Kyodotai ni okeru Wain Sangyo no Jizoku-kanosei to
Kyotsu-shijo Seido-kaikaku: Shohi-doko oyobi Seisanchosei-seido ni kansuru Bunseki (The sustainability of

©www.vdqs.net/2012Coimbra

APPENDIX I: Basic Statistics Table

Observation Mean Std. Dev. Min Max

Annual Data 40 125.3 93.4 5.717 297.883
total wine consumption(a)
domestic wine consumption 40 56.9 31.6 4.934 122.798
imported wine consumption
GDP expenditure(b) 40 68.4 63.9 0.783 184.985

GDP per capita 40 363408.6 154082 73345 523198

40 2.940891 1.153575 0.707144 4.146902

Monthly Household Expenditure (c) 40 436166.3 146360.2 112,949 595,214
Income 40 277387.3 82947.0 82,582 357,636
Consumption 40 1287.1 2,243 6,107
Rice 40 4234.1 2,583
Bread 40 1991.8 648.9 488 1,694
Noodle 40 1322.3 336.1 432 10,533
Fish 40 7985.1 1962.4 3,386 8,082
Meat 40 6699.0 1371.6 2,658 4,083
dairy&egg 40 3520.3 509.7 2,090 10,513
Vegetables 40 8067.0 1772.2 3,483 3,725
Fruits 40 2935.6 553.7 1,760 3,895
Beverages 40 2914.0 773.3 1,135 4,110
Alcohol 40 3222.5 747.7 1,327 13,192
eating out 40 9731.7 3596.7 2,112 6,745
social expenditure 5051.4 1371.2 1,610

Food Service Industry Annual Sales Data(d)

diner & restaurant sales 35 69886.6 23871.5 21,838 97,332

sushi sales 35 12276.4 2914.5 5,074 15,485

other sales 35 8643.3 3076.0 3,266 13,447

domestic airplane sales 35 1989.1 640.7 646 2,581

hotel sales 35 33433.3 10350.9 15,174 49,546

café sales 35 13379.4 2626.6 7,375 17,396

pub sales 35 10786.8 2943.9 4,035 14,629

bar & night club sales 35 26362.9 7175.6 10,267 35,752

takeout sales 35 28544.9 19823.2 2,016 56,581

(a): units in 1,000KL; source Mercian(2011), based on National Tax Agency reports (original source could not

be traced).

(b): Current GDP in billions JPY. Data for 1970-1979, 1980-1994, 1995-2009, take from 2000, 2009 and 2010

statistics respectively.

(c): Annual Average of Monthly Receipts and Disbursements in JPY per Household (Workers' Households); All

Japan, Cities with Population of 50,000 or more (1963-2010); source for (b) & (c): Ministry of Internal Affairs

and Communications.

(d): annual sales in JPY 100million; source: Foodservice Industry Research Institute (2012).

©www.vdqs.net/2012Coimbra