Understanding International Food Consumption

Understanding International Food Consumption Patterns
Yanrui Wu
Economics Program
School of Economics and Commerce
University of Western Australia
35 Stirling Highway, Crawley WA 6009
Tel: 08 9380 3964
Fax: 08 9380 1016
[email protected]
This paper is part of the outcome of the project on the comparison of international food
consumption patterns. The project is funded by an Australian Centre for International
Agricultural Research (ACIAR) grant. I would like to thank Ray Trewin for his support,
Kenneth Clements for his encouragement, helpful discussions and comments, and Jing
Zhang and Nicole Guok for their excellent research assistance.
Summary
International food consumption has attracted a lot of attention in the literature. In
particular, a series of studies have been conducted using the popular International
Comparison Project (ICP) databases (released in the 1970s and 1980s). This paper adds
to the growing literature by presenting an analysis of the 1995 ICP data. It aims to
examine cross-country consumption patterns of individual food items such as cereals,
meat, dairy products and other foods. Especially, it attempts to identify international food
consumption norms and outliers, and gain insight into the impact of country-specific
factors (eg. income, geography, culture etc) on food consumption.
According to this study, as income rises, an average country in the world tends to spend
proportionally less on food and its demand for food becomes less elastic too. This trend is
however not very clear if the focus is the regions instead of income groups. At the
disaggregate level, most food items (eg. cereals, meats, fruits and vegetables, dairy
products and oils and fats) are found to be necessities. Aquatic products, alcoholic and
non-alcoholic beverages however appear to be luxuries for most countries with the
exception of the high-income ones.
Income elasticities of demand in South Asia are found to be high for all food items except
cereals and dairy products. In contrast to South Asia, dairy goods account for a small
budget share in East Asia. It is also found that the Chinese diet mainly consists of cereals,
meats, aquatic products, and fruits and vegetables. There are however considerable
2
variations among the regions in large countries such as India and China. Regional issues
can only be addressed by conducting detailed studies using household or regional data.
3
1
Introduction
International food consumption continuously attracts the attention of researchers and
policy makers not only because it is concerned with a scarce commodity in the world but
also because it is so dynamic and diverse. The pioneers in this field include Working
(1943), Leser (1963), Barten (1977), Clements et al. (1979), and Seale and Theil (1986),
to cite a few. In particular, the analytical framework proposed by Working has been
widely applied in empirical studies due to its simplicity and expanded over time because
of the availability of better statistics. This paper falls into the category of applying the
Working’s model to the latest International Comparison Project (ICP) database. The
objective of this paper is to examine food consumption patterns across the countries and
investigate how country-specific factors such as the geographical location and income
level affect international food consumption patterns. The rest of the paper is organised as
follows. The data issues are described in Section 2. The preliminary analysis of the
database is presented in Section 3. The analytical issues are then discussed in Section 4.
This is followed by the presentation of the estimation results and their interpretation in
Section 5. A more detailed analysis of the Asian economies is reported in Section 6. The
final section presents some concluding remarks.
2
Data Issues
The ICP datasets have been compiled for the years 1970, 1973, 1975, 1980, 1985, 1990
and 1995. Many reports based on pre-1995 databases have been released (eg. Kravis et al.
4
1982, Theil and Clements 1987, Rimmer and Powell 1992). There are however few
studies based on the 1995 dataset. Regmi et al. (2001) and Seale and Regmi (2002) are
two exceptions. Regmi et al. employed a two-stage budgeting process to estimate income
and price elasticities of demand for individual food items. Seale and Regmi applied a
modified Working’s model to examine income and price effects on the consumption of
nine broad categories of food and non-food commodities. The present study employs part
of the 1995 ICP database, ie. the expenditure data which were privately obtained. Due to
the nature of the data available, this study focuses on examining the impact of income on
food consumption patterns in the world. It particularly focuses on the impact of countryspecific factors such as the geographical location and income level. The 1995 ICP
database covers 114 countries. Preliminary checks of the data shows that there are serious
problems with data from thirty countries. As a result, these countries are omitted from the
empirical analysis.1 It is unfortunate that the 1995 ICP dataset does not include the
world’s two most populous countries, ie. India and China. To overcome this problem,
other secondary sources are used to compile data so that India and China are included in
the analysis.2 In order to examine food consumption patterns in the Chinese communities
(eg. Mainland China, Hong Kong and Singapore), Taiwan is also added to the database
by using data from secondary sources.3 As a result, the final sample contains 87
countries. They are divided into seven groups according to their geographical locations
1
The omitted countries are Albania, Belarus, Bulgaria, Congo, Georgia, Grenada, Jamaica, Lebanon,
Macedonia, Madagascar, Malawi, Mexico, Moldova, Nigeria, Paraguay, Poland, Romania, Russia, Sierra
Leone, St. Vincent & Grenadines, Syria, Tajikistan, Tanzania, Turkey, Turkmenistan, Uruguay, Venezuela,
Vietnam, Yemen and Zambia.
2
Indian data are drawn from Statistical Abstract, India, Delhi: Manager of Publications, 1998. Chinese data
are weighted average of rural and urban consumption figures derived from China’s Statistical Yearbook
1996 (National Statistical Bureau 1996).
3
Data for Taiwan are derived from the official statistical yearbook (Statistical Yearbook of the Republic of
China 1996) and Huang and Bouis (2001).
5
and four categories according to the level of income.4 In particular, 57 developing
countries are included in the sample (see Table 1).
One of the main problems with international comparative studies is the choice of
exchange rates so that national incomes expressed in local currencies can be converted
and become compatible. For this purpose, several approaches have been developed.5 This
paper adopts the concept of the purchasing power parity (PPP) and hence employs the
conversion ratios of local currency over international dollar estimated by the World Bank
(Easterly and Yu 2000). The same ratios are also used to convert household expenditure
data. Some summary statistics about the sample are presented in Table 1. Several
observations can be made from this table. First, as expected, South Asia and Sub Sahara
Africa are the poorest regions in the world. Second, total expenditure on an average
accounts for more than 70 per cent of GDP among the regions with the exception of East
Asia and the Pacific where an average country tends to save more than those in the rest of
the world. Finally, food is the largest consumption item followed by housing and
transport. On an average, food expenditure amounts to about one third of total
expenditure in the world. However, there are considerable variations among the regions.
Countries in South Asia spend the most on food (59%) and those in West Europe and
North America the least (16%). These observations will be further investigated later by
analyzing cross-country statistics.
4
5
The lists of the countries are reported in Appendices 1 and 2.
For a brief review, see Clements et al. (2003).
6
Table 1 Summary Information of the Sample
____________________________________________________________________________________________________________
Geographical regions
No. of GDPpc Exp
Food
Clothing Housing Durable Medical Transport
Recreation Other
and income groups
countries (ppp$) share
____________________________________________________________________________________________________________
East Asia and Pacific
East Europe and Central Asia
Middle East and North Africa
South Asia
West Europe and North America
Sub Sahara Africa
Latin America and Caribbean
13 12,016 0.66
13 4,675 0.72
11 9,543 0.73
5 1,469 0.80
18 21,313 0.72
12 3,248 0.72
15 7,525 0.72
0.33
0.46
0.34
0.59
0.16
0.46
0.32
0.07
0.07
0.09
0.07
0.06
0.10
0.07
0.16
0.12
0.18
0.13
0.19
0.12
0.18
0.07
0.04
0.08
0.04
0.06
0.09
0.08
0.07
0.08
0.07
0.03
0.12
0.05
0.08
0.11
0.10
0.12
0.07
0.14
0.11
0.12
0.08
0.05
0.04
0.02
0.08
0.02
0.05
0.11
0.09
0.08
0.04
0.18
0.06
0.11
Total
87
0.34
0.07
0.16
0.07
0.08
0.12
0.05
0.11
Lower income
Lower middle income
Upper middle income
High income
18 1,681
21 3,917
18 8,814
30 19,788
0.55
0.42
0.32
0.18
0.10
0.08
0.06
0.06
0.11
0.14
0.19
0.18
0.05
0.07
0.07
0.07
0.03
0.06
0.09
0.11
0.08
0.11
0.12
0.14
0.02
0.04
0.06
0.08
0.05
0.08
0.09
0.17
9,940 0.72
0.77
0.73
0.66
0.72
Developing countries
57 4,757 0.72
0.43
0.08
0.15
0.07
0.06
0.10
0.04
0.07
____________________________________________________________________________________________________________
Notes “GDPpc” stands for the mean GDP per capita in international dollars (ppp$) and “exp share” the mean ratio of expenditure over GDP. The figures in the
commodity columns are the mean budget shares of individual commodity expenditure over total expenditure.
Sources The expenditure share figures are calculated from the ICP database. The GDP figures are drawn from Easterly and Yu (2000).
7
3
Preliminary Analysis
Figure 1 demonstrates the relationship between food budget shares and per capita
income. In general, as income rises, countries spend proportionally less on food. The
poorest countries spend proportionally the most on food consumption. There are however
some outliers. For example, four countries (Azerbaijan, Amenia, Sri Lanka and Gabon)
spend proportionally far more on food than other countries even when income is
controlled for. Mali, Zimbabwe and Barbados are at the other end of the extremes, ie.
these countries spend proportionally less on food than others. It is however worthy of
notice that at the aggregate level both China and India appear to follow the international
norm in terms of food consumption according to Figure 1.
0.80
0.70
China
Azerbaijan
Amenia
Sri Lanka
Food budget shares
0.60
Gabon
Mali
0.50
0.40
India
0.30
Zimbabwe
0.20
0.10
Barbados
0.00
6
7
8
9
Natural log of GDP per capita
Figure 1 Food Budget Shares vs GDP per capita.
8
10
11
Among individual food items, rich countries tend to consume proportionally less on
cereals but more on meat products (see Figures 2 and 3). In particular, Nepal and
Bangladesh, two of the poorest countries, spend relatively the most on cereals. In the
meantime, Mali, Mongolia and Jordan spend proportionally the most on meat
consumption. Sri Lanka and Japan are also outliers. The former spends proportionally the
least on meat products. The latter, Japan, is an outlier among the developed countries.
Japanese consumers spend proportionally more on cereals and less on meat. This
consumption pattern may be due to two reasons. First, it is well known that the Japanese
agricultural sector has been heavily protected and, as a result, the prices of Japanese
cereal products, eg. rice, are much higher. Second, it appears that fish consumption
accounts for a significant part of the Japanese diet. This is evident in Figure 4 which
presents the combined budget shares of meat and fish products against the natural log of
GDP per capita. Japan is not an outlier anymore. Mongolia is still an outlier due to its
large consumption of meat. Hong Kong however leads the developed economies and its
citizens spend relatively the most on meat and fish products combined. Among the
developing economies, Nepal and Kenya are two outliers, their citizens spending
proportionally least on meat and fish products. In terms of dairy products, Pakistan, India,
Mongolia and Kenya are clearly the outliers according to Figure 5. As for the
consumption of fruits and vegetables, Gabon appears to be an outlier. Consumers there
spend proportionally much more on fruits and vegetables (see Figure 6). As expected,
Mongolia is at the other end of the extremes. Consumers there spend relatively little on
fruits and vegetables.
9
0.7
Cereals Budget shares
0.6
Nepal
0.5
Bangladesh
0.4
0.3
Japan
0.2
0.1
0.0
6
7
8
9
10
11
Natural log of GDP per capita
Figure 2 Budget Shares of Cereal Products vs GDP per capita
0.35
Mongolia
Meat budget shares
0.30
Jordan
0.25
0.20
0.15
Mali
0.10
Japan
0.05
Sri Lanka
0.00
6
7
8
9
Natural log of GDP per capita
Figure 3 Budget Shares of Meat Products vs GDP per capita
10
10
11
0.45
Meat&fish budget share
0.40
Hong Kong
Mongolia
0.35
0.30
Japan
0.25
0.20
0.15
0.10
Kenya
0.05
Belize
Nepal
0.00
6
7
8
9
10
11
Natural log of GDP per capita
Figure 4 Budget Shares of Meat and Fish Products vs GDP per capita
0.30
Pakistan
Dairy budget shares
0.25
Mongolia
0.20
India
0.15
Kenya
0.10
0.05
0.00
6
7
8
9
Natural log of GDP per capita
Note The dot on the x-axis represents Taiwan due to missing data.
Figure 5 Budget Shares of Dairy Products vs GDP per capita
11
10
11
0.50
Fruit&veg budget shares
0.45
Gabon
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
Mongolia
0.00
6
7
8
9
10
11
Natural log of GDP per capita
Figure 6 Budget Shares of Fruit and Vegetable Products vs GDP per capita
In terms of consumption of alcoholic and non-alcoholic beverages, the United Kingdom
and Botswana are two outliers. Consumers in both countries spend a relatively high
proportion on alcoholic drinks (see Figure 7). There are however countries where there is
little or no consumption of alcohol largely due to religion. These countries include India,
Pakistan, Kenya, Qatar, Oman, Mali and Jordan. In the meantime, Barbados, Bolivia,
Thailand and Zimbabwe appear to be outliers, spending proportionally more on nonalcoholic beverages (see Figure 8). However, Luxembourg and Portugal are two highincome economies spending relatively less on non-alcoholic beverages. As for the
consumption of oil and fat products, Senegal is clearly an outlier spending proportionally
far more than other countries according to Figure 9. Korea appears to be at the other end
of the extreme, spending relatively less on oils and fats.
12
Budget shares of beverage
0.14
0.12
Barbados
Thailand
0.10
0.08
Zimbabwe
Bolivia
0.06
0.04
0.02
Portugal
Luxembourg
0.00
6
7
8
9
Natural log ofGDP per capita
10
11
Note The dots on the x-axis represent countries with zero consumption (ie. Egypt, India, Jordan, Kenya,
Mali, Oman, Pakistan and Qatar) or missing data (ie. Taiwan).
Figure 7 Budget Shares of Alcoholic Beverages vs GDP per capita
Budget shares of beverage
0.14
0.12
Barbados
Thailand
0.10
0.08
Zimbabwe
0.06
Bolivia
0.04
0.02
0.00
6
7
8
9
Natural log of GDP per capita
10
11
Note The dots on the x-axis represent countries with zero consumption (ie. Nepal and Sri Lanka) or
missing data (ie. Taiwan).
Figure 8 Budget Shares of Non-alcoholic Beverages vs GDP per capita
13
Budget shares of oil and fa
0.16
Senegal
0.14
0.12
Azerbaijan
Pakistan
0.10
0.08
Mali
0.06
0.04
0.02
Korea
0.00
6
7
8
9
Japan
10
11
Natural log of GDP per capita
Note The dot on the x-axis represents Taiwan due to missing data.
Figure 9 Budget Shares of Oil and Fat Products vs GDP per capita
In summary, with the exception of a few outliers, countries in the world tend to follow
some norm in terms of consumption of food and individual food items. First, food
consumption budget falls as income rises. Second, the budget shares of cereal, oil and fat
products tend to decline as income increases. Third, consumption of beverages, meat and
fish appears to rise as income increases. Finally, there seems to be more diversity in the
consumption of dairy products, fruits and vegetables. These preliminary observations will
be further investigated in the following sections.
14
4
Analytical Framework
To examine the effect of income and country-specific factors on food consumption across
countries, a set of Engel functions are estimated. The choice of this approach is largely
due to the constraint of the database which we have access to and which only contains
information about expenditures (see Appendix 3). An Engel function relates the budget
share of a commodity to total household expenditure. This concept can also be applied to
a group of countries. Following Working (1943) and Leser (1963), an Engel function can
be expressed as
ω ij = α i + β i log(M j )
(1)
where ω ij is the budget share of the ith commodity consumed by the jth country, M j total
expenditure per capita in country j and α i and β i parameters to be estimated for the ith
commodity. Given equation (1), the income elasticity of demand for the ith commodity,
η i , can be computed as follows
ηi = 1 + β i / ω i
(2)
where ω i is the mean budget share of the ith commodity consumed.
Clements et al. (1995) argued that Model (1) could be treated as a simplified version of
more complicated systems under the assumption of constant prices. This is obviously a
strong assumption. Model (1) can be extended to incorporate country-specific factors
such as geographical grouping and income clustering. Symbolically, the extended model
can be expressed as
15
ω ij = α i + Σα ik Z k + β i log(M j ) + Σβ ik Z k log(M j )
(3)
where Z k is the dummy variable which represents the kth country-specific factor. The
above specification implies that the country-specific factors affect both the intercept and
the slope of the Engel function. Given model (3), the income elasticity of demand, η ik ,
for the kth factor and ith commodity can be computed as
η ik = 1 + ( β i + β ik ) / ω i
(4)
The empirical analysis follows the conventional approach by adopting a two-stage
budgeting procedure (see Deaton and Muellbauer 1984). In the first stage, consumers
allocate their budget to broad consumption groups (eg. food, clothing etc). In the second
stage, consumers make budget decisions concerning individual items within the food
group. Following this procedure, the income elasticity for the ith individual food item and
kth country-specific factor can be calculated as
θ ik = η kη ik
(5)
where θ ik represents the income elasticity for the ith food item, η k the income elasticity
of food and η ik the expenditure elasticity for the ith item within the food group.
5
Empirical Findings
The analytical framework described in the preceding section is applied to the sample
discussed in Section 2. The focus of the analysis is on food consumption patterns of
countries in several geographical regions and at different income levels. For this purpose,
16
the countries are divided into four geographical regions and four income groups,
respectively (see Appendices 1 and 2 for details). The criteria applied to divide the
sample include the number of observations in each group or region, geographical location
and cultural background. In the empirical estimation, the outliers identified in Section 3
together with missing observations are excluded so that the estimated results better reflect
the true pattern of consumption in each group or region. The estimated income elasticities
are presented in Table 2.6
According to Table 2, food at the aggregate level is a normal good for all income groups
and regions. Demand for food however becomes less elastic as income rises, a conclusion
also derived by Regmi (2001). Among the regions, there is not much difference in terms
of food consumption – the income elasticity of demand for food ranges from 0.575 in
East Asia, South Asia and the Pacific to 0.715 in Sub Sahara Africa, Latin America and
the Caribbean. At the disaggregate level, the impact of income level on consumption is
clearly evident in Table 2. According to this table, the estimated income elasticity for all
food items is less than 0.40 for the high income group while those for other three groups
are greater than 0.40 with the exception of income elasticity of demand for beverages in
the lower middle income group. These results are consistent with those in Regmi (2001)
who applied a slightly different grouping. As expected, the estimated income elasticities
of demand for individual food items have the largest values for the low-income group.
However, there is no sharp difference in the food consumption patterns among the four
regions. The possible explanation is that there are poor and rich countries in each region.
6
Regression results are reported in Appendix 4.
17
Table 2 Estimated Income Elasticity of Demand
__________________________________________________________________________________________________________
Geographical regions
Income groups
Low
LM
UM
High
Goods
I
II
III
IV
__________________________________________________________________________________________________________
Cereals
Meats
Aquatic products
Fruits and vegetables
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
0.224
0.680
0.879
0.655
1.062
1.147
0.263
0.015
0.602
0.409
0.696
1.242
0.465
0.703
1.280
0.661
0.333
0.646
0.462
0.738
1.055
0.752
0.864
0.804
0.896
0.600
0.698
0.358
0.647
0.761
0.460
0.848
0.977
0.586
0.295
0.811
0.681
0.844
1.155
0.961
1.224
1.908
0.437
0.906
0.723
0.702
0.489
1.229
0.676
0.940
0.270
0.506
0.690
0.666
0.584
0.406
1.184
0.918
0.480
0.761
0.517
0.400
0.413
0.215
0.230
0.385
0.159
0.265
0.291
0.294
0.221
0.196
Divisia variance
0.277
0.192
0.201
0.072
0.082
0.141
0.196
0.522
Food
0.575
0.621
0.715
0.667
0.735
0.653
0.559
0.200
__________________________________________________________________________________________________________
Notes The classification of the regions follows the World Bank system. The sample is divided into four geographical regions, that is,
I
East Asia and Pacific and South Asia;
II
East Europe and Central Asia and Middle East and North Africa;
III
Sub Sahara Africa and Latin America and Caribbean;
IV
West Europe and North America.
The sample can also be divided into four groups according to the level of income, that is,
Low
low income;
LM
lower middle income;
UM
upper middle income;
High
high income.
18
The average country in each region is thus not a good representative of individual
countries within the region. Income elasticity based on the regional mean budget shares is
therefore biased. This bias may cause serious problems in demand projections if regional
means are used as the estimates of demand parameters for individual countries. Table 2
also shows that, in general, demand for aquatic, beverage and alcoholic goods is more
elastic than that for other food products. In fact, these three goods seem to be luxuries for
some regions and income groups. In addition, the dispersion of the income elasticities can
be measured by their Divisia (budget share weighted) variance
DV = ∑ ω l (η l − 1) 2
(6)
As shown in Table 2, there is little dispersion in the income elasticities for West Europe
and North America and more dispersion for Asia and the Pacific regions. In terms of
income groups, there is however less dispersion in the income elasticities for the low
income group as income elasticities for this group are close to one. Furthermore, the
Divisia variance for the high-income group is high, reflecting the fact that income
elasticities for this group are generally far smaller than one as shown in Table 2.
6
Food Consumption Patterns in Asia
Food consumption patterns in Asia are of particular interest for several reasons. First, the
world’s two most populous nations, ie. India and China, are located in Asia. Both
countries have recently experienced rapid economic growth and hence rising income. The
rising income and resultant changes in these two countries have important implications
19
for food consumption in the world. Second, Asian countries are highly diverse in terms of
affluence, culture and diets. This section focuses on three particular groups ie. South
Asia, East Asia and the Chinese communities.
South Asia
Among the sample, five countries are located in South Asia. Table 3 reports the budget
shares and income elasticities in these countries. It shows that food consumption
expenditure in South Asian countries is dominated by cereals and fruits and vegetables,
and dairy products in the case of India and Pakistan. In addition, food spending accounts
for more than a half of total expenditure. It seems that the income elasticities of all foods
except cereals and dairy products are close to or greater than one – implying that most
food items appear to be luxury goods in the five South Asian countries. This is confirmed
by the small values of the Divisia variance of income elasticities for each country. The
findings showed in Table 3 are generally consistent with the fact that South Asia is one of
the poorest regions in the world.
East Asia
There are nine East Asian economies in the sample. They have gone through different
stages of development, including low-income economies such as China, Indonesia and
the Philippines, middle-income ones of Thailand, Taiwan and Korea and high-income
economies such as Japan, Hong Kong and Singapore. Table 4 demonstrates the budget
shares and income elasticities in these economies. It shows that food budget accounts for
20
Table 3 Budget Shares and Income Elasticities of South Asian Countries
Bangladesh
India
Nepal
969
1,490
1,039
1,561
2,285
Budget shares
Cereals
Meat
Aquatic products
Fruits and Vegetables
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
0.502
0.044
0.092
0.096
0.004
0.008
0.032
0.039
0.155
0.308
0.090
0.576
0.033
0.006
0.146
0.015
0.212
0.077
0.007
0.173
0.218
0.018
0.124
0.264
0.048
0.033
0.178
0.084
0.115
Food
0.674
0.754
1.126
0.965
1.280
7.142
2.867
-0.010
1.075
0.779
1.210
0.595
0.850
0.717
Divisia variance
0.270
Food
0.788
GDP pc (ppp$)
Income elasticity
Cereals
Meat
Aquatic products
Fruits and Vegetables
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
0.140
Pakistan Sri Lanka
0.054
0.043
0.044
0.006
0.268
0.101
0.118
0.067
0.014
0.145
0.525
0.618
0.491
0.653
0.677
0.880
0.740
1.207
3.307
1.085
2.528
0.637
0.882
2.788
0.954
0.833
1.040
1.027
0.799
1.240
0.302
1.020
0.739
3.335
0.623
0.808
0.698
0.519
0.926
0.794
0.075
0.139
0.136
0.041
0.728
0.769
0.709
0.781
Notes Income elasticities are based on regression results in Table A2, Appendix 3. Blank cells are due to
missing numbers.
21
Table 4 Budget Shares and Income Elasticities of East Asian Countries
Mainland
China Taiwan Hong Kong Singapore
GDP pc (ppp$)
Indonesia Philippines Thailand
Korea
Japan
2,661
14,222
23,153
25,502
3,205
3,290
6,473
12,275
22,770
Food
0.541
0.255
0.110
0.139
0.513
0.489
0.302
0.353
0.159
Cereals
Meats
Aquatic products
Fruits and vegetables
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
0.324
0.196
0.048
0.176
0.035
0.005
0.040
0.040
0.087
0.150
0.187
0.159
0.108
0.000
0.000
0.000
0.000
0.397
0.090
0.227
0.197
0.118
0.045
0.069
0.034
0.033
0.122
0.103
0.133
0.150
0.181
0.086
0.066
0.050
0.018
0.113
0.335
0.051
0.087
0.237
0.004
0.007
0.057
0.047
0.072
0.297
0.145
0.145
0.111
0.044
0.024
0.067
0.018
0.098
0.161
0.186
0.033
0.164
0.124
0.100
0.052
0.028
0.090
0.207
0.127
0.117
0.212
0.046
0.064
0.050
0.009
0.100
0.223
0.078
0.170
0.128
0.083
0.067
0.048
0.007
0.115
Budget shares
22
Income elasticities
Food
0.735
0.439
0.200
0.200
0.721
0.708
0.527
0.595
0.103
Cereals
Meats
Aquatic products
Fruits and vegetables
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
0.687
0.806
1.052
0.986
1.464
3.710
0.142
0.993
0.721
0.460
0.330
0.642
0.913
0.400
0.220
0.224
0.236
0.150
0.355
0.262
0.478
0.217
0.196
0.217
0.241
0.248
0.167
0.281
0.265
0.392
0.231
0.196
0.675
0.985
0.893
0.903
7.740
2.885
0.310
0.937
0.704
0.742
0.489
0.899
0.747
1.169
0.295
0.472
0.815
0.725
0.575
0.401
1.152
0.547
0.648
0.453
0.302
0.578
0.541
0.616
0.377
0.969
0.921
0.450
0.721
0.509
-0.092
0.387
0.107
0.139
0.124
0.079
0.146
0.136
0.205
0.147
0.101
Divisia variance
0.123
0.292
0.556
0.533
0.260
0.106
0.225
0.160
0.714
Notes Income elasticities are based on regression results in Table A2, Appendix 3. Blank cells are due to missing numbers.
23
about a half of total expenditure in low-income countries but less than 20 per cent of total
expenditure in high-income economies. Table 4 also shows that food budget in East Asia
is dominated by cereals and meat or aquatic products. The next major items are fruits and
vegetables.
As in other regions, demand for food is highly elastic in low-income East Asia such as
China and Indonesia, but less elastic in high-income economies such as Japan, Singapore
and Hong Kong. The value of Divisia variance shows that there is little dispersion in the
income elasticities of demand in the low-income economies eg. China, Indonesia and the
Philippines, implying that income elasticities in these economies are close to or greater
than one. However, there is more dispersion in the income elasticities of demand in the
high-income economies eg. Japan, Hong Kong and Singapore, an indication of low
income elasticities in these economies.
Chinese Societies
It is interesting to compare food consumption patterns among the four Chinese societies,
ie. Mainland China, Hong Kong, Taiwan and Singapore, as they have gone through
different stages of development and they share a similar cultural background and dietary
habits. In particular, one can speculate what is going to happen if China follows the
patterns of other three Chinese-dominated societies (ie. Taiwan, Singapore and Hong
Kong). The findings could have important implications given China’s sheer size of
population. According to Table 4, it can be anticipated that China’s demand for aquatic
products will increase substantially as income rises. Other potential areas of growth in
24
demand include alcoholic and non-alcoholic beverages. Table 4 also shows that the
budget shares of dairy products in the four Chinese societies are very close. One may
conclude that, in terms of budget share, dairy consumption in China has already reached
its potential level. Future growth will mainly be in the form of an increase in the absolute
quantity. Of course, it has to be noted that there are considerable regional variations
inside China in terms of affluence and dietary traditions. For some regions, there will be
growth in both relative and absolute terms. The same argument can also be applied to
India, the world’s second largest country in terms of population. To gain further insight
into the regional issues, country-specific projects have to be conducted.
7
Conclusions and Comments
This study estimates a series of Engel functions using the 1995 ICP expenditure data. It
particularly incorporates regional grouping and income clustering variables into the Engel
functions. The results are used to derive estimates of income elasticities of demand for
various groups and countries. As stated in the main body of the paper, this study is
constrained by the availability of data. Thus all conclusions in this paper are subjected to
qualifications. However, the findings in this study still confirm some results in the
existing literature and gain fresh insights into cross-country food consumption patterns.
This study particularly points out the importance of country-specific studies in terms of
understanding food consumption patterns. The major points are summarised as follows:
25
•
Countries in the world are found to follow some common norm in terms of food
consumption. As income rises, an average country in the world tends to spend
proportionally less on food and its demand for food becomes less elastic too. This
trend is however not very clear if the focus is the regions instead of income
groups. Thus, regions are poor representatives of individual countries.
•
At the disaggregate level, most food items (eg. cereals, meats, fruits and
vegetables, dairy products and oils and fats) are found to be necessities for all
groups (by either region or income). Aquatic products, alcoholic and nonalcoholic beverages however appear to be luxuries for most groups except the
high-income one.
•
Food consumption patterns in Asia are of special interest partly because China
and India are in the region and both countries are experiencing rapid
transformation. With the exception of alcoholic and non-alcoholic beverages, all
food items are shown to be necessities in Asia. The value of income elasticity of
demand for cereals in Asia is the smallest among the regions, a reflection of the
importance of cereals in Asian diet.
•
There is however considerable diversity in Asia. South Asia has the lowest
income in Asia. Income elasticities of demand in South Asia are found to be high
for all food items except cereals and dairy products. It is however difficult to
speculate how food consumption patterns will evolve in South Asia as the five
26
countries considered in this paper are themselves very diverse. India and Pakistan
seem to follow a similar pattern, with cereals, fruits and vegetables and dairy
products being the dominant food items in their diets. The other three countries ie.
Bangladesh, Nepal and Sri Lanka appear to have different dietary habits,
consuming far less dairy goods than India and Pakistan. Unless country-specific
studies are carried out, it is difficult to draw any conclusion about future
consumption trend.
•
Food consumption pattern in East Asia is dominated by the spending on cereals,
meat or aquatic products and fruits and vegetables. In contrast to South Asia,
dairy goods account for a small budget share in East Asia even in the most
developed economy ie. Japan. It is anticipated that growth in the consumption of
dairy products will only be in absolute term in East Asia. In addition, developed
East Asian members consume relatively more aquatic products and beverages
(both alcoholic and non-alcoholic). Demand for these products will certainly
increase in the future.
•
The four Chinese-dominated economies offer an interesting case study though it
is subjected to qualifications due to missing observations in Taiwan’s data. It is
found that the Chinese diet mainly consists of cereals, meats, aquatic products,
and fruits and vegetables. If mainland China is to follow the same patterns of the
other three societies, there will be a huge growth in the demand for aquatic goods.
In the meantime, demand for cereals in China will decline relatively. Another area
27
of growth will be demand for beverages (both alcoholic and non-alcoholic) in
China. There are of course considerable variations among the Chinese regions.
Those regional issues can only be addressed by conducting detailed studies using
household or regional data.
28
Appendix 1 List of countries by geographical distribution
East Asia and Pacific (13)
Middle East and North Africa (11)
Australia
China
Fiji
Hong Kong
Indonesia
Japan
Korea
Mongolia
New Zealand
Philippines
Singapore
Taiwan
Thailand
Bahrain
Egypt
Greece
Iran
Israel
Jordan
Morocco
Oman
Portugal
Qatar
Tunisia
East Europe and Central Asia (13)
South Asia (5)
Armenia
Azerbaijan
Czech Republic
Estonia
Hungary
Kazakhstan
Kyrgyz Republic
Latvia
Lithuania
Slovakia
Slovenia
Ukraine
Uzbekistan
Bangladesh
India
Nepal
Pakistan
Sri Lanka
29
West Europe and North America (18)
Latin America and Caribbean (15)
Austria
Belgium
Denmark
Finland
France
Germany
Iceland
Ireland
Italy
Luxembourg
Netherlands
Norway
Spain
Sweden
Switzerland
UK
Canada
USA
Antigua & Barbuda
Argentina
Bahamas
Barbados
Belize
Bermuda
Bolivia
Brazil
Colombia
Dominica
Ecuador
Peru
St. Kitts & Nevis
St. Lucia
Trinidad & Tobago
Sub Sahara Africa (12)
Benin
Botswana
Cameroon
Cote d'Ivoire
Gabon
Guinea
Kenya
Mali
Mauritius
Senegal
Swaziland
Zimbabwe
30
Appendix 2 List of countries by income group
Low income (18)
Lower middle income (21)
Benin
Cameroon
Cote d'Ivoire
Guinea
Kenya
Mali
Senegal
Zimbabwe
Bangladesh
India
Nepal
Pakistan
Armenia
Azerbaijan
Kyrgyz Republic
China
Indonesia
Mongolia
Belize
Bolivia
Colombia
Dominica
Ecuador
Peru
Swaziland
Sri Lanka
Egypt
Iran
Jordan
Morocco
Tunisia
Kazakhstan
Latvia
Lithuania
Ukraine
Uzbekistan
Fiji
Philippines
Thailand
31
Upper middle income (18)
High income (30)
Antigua & Barbuda
Argentina
Barbados
Brazil
St. Kitts & Nevis
St. Lucia
Trinidad & Tobago
Botswana
Gabon
Mauritius
Bahrain
Oman
Czech Republic
Estonia
Hungary
Slovakia
Korea
Taiwan
Bahamas
Bermuda
Canada
USA
Austria
Belgium
Denmark
Finland
France
Germany
Iceland
Ireland
Italy
Luxembourg
Netherlands
Norway
Spain
Sweden
Switzerland
UK
Greece
Israel
Portugal
Qatar
Slovenia
Australia
Hong Kong
Japan
New Zealand
Singapore
32
Appendix 3: List of the data
Countries
GDP
M
Mali
Bangladesh
Nepal
Kenya
Benin
Mongolia
India
Pakistan
Azerbaijan
Senegal
Cote d'Ivoire
Guinea
Cameroon
Kyrgyz Republic
Armenia
Zimbabwe
Sri Lanka
Ukraine
China
Bolivia
Egypt
Uzbekistan
705
969
1039
1164
1200
1477
1490
1561
1594
1634
1710
1725
1811
1950
2166
2198
2285
2471
2661
2771
2839
2928
602
862
968
851
1114
1377
812
1248
1506
1300
1192
1319
1534
1134
1273
1844
1948
1159
1072
2538
2590
1657
Food Cereals Meats Aquatic Dairy Oils & fats Fruits & vegs Other food Beverages Alcohol
54.5
67.4
61.8
51.1
57.5
59.3
52.5
49.1
75.9
56.6
52.5
46.7
46.7
49.9
71.3
30.8
65.3
50.0
54.1
43.8
50.2
54.3
34.4
50.2
57.6
32.5
23.6
30.4
30.8
21.2
39.0
26.5
19.6
16.1
16.1
21.1
18.9
23.7
21.8
17.8
32.4
21.9
24.6
27.3
14.1
4.4
3.3
5.1
14.3
31.2
9.0
7.7
14.4
13.9
14.4
16.2
16.2
9.6
8.2
22.0
1.8
21.6
19.6
23.9
23.6
10.7
3.0
9.2
0.6
0.4
7.6
0.0
0.7
1.1
13.1
2.2
4.7
4.7
0.3
1.6
2.6
12.4
2.5
4.8
0.9
4.6
0.2
33
3.8
3.2
5.4
15.1
4.1
18.1
17.8
26.8
5.6
4.4
4.4
1.2
1.2
8.1
6.2
9.0
6.7
14.0
4.0
5.9
10.1
12.0
8.1
3.9
4.3
2.6
4.5
3.5
8.4
10.1
10.2
13.9
1.5
3.8
3.8
6.1
9.6
6.7
1.4
4.2
4.0
3.2
8.4
5.1
9.9
9.6
14.6
17.6
33.2
3.8
14.0
17.3
13.0
13.1
23.3
31.2
31.2
33.9
34.4
10.0
26.4
19.9
17.6
22.2
12.5
19.3
19.9
15.5
4.4
11.2
3.3
6.9
11.5
11.8
13.7
8.5
15.2
7.6
7.6
10.0
16.0
12.0
14.5
10.6
8.7
8.6
6.9
20.7
1.2
0.8
0.0
2.2
2.5
0.6
3.3
0.6
0.2
1.9
1.6
1.8
1.8
1.6
0.3
7.5
0.0
1.4
0.5
7.8
0.8
0.8
0.0
0.4
1.5
8.4
5.7
3.6
0.0
1.0
2.2
7.9
14.5
14.5
7.5
2.0
5.3
4.8
5.8
3.5
4.6
0.0
3.4
Appendix 3: List of the data (continued)
Countries
GDP
M
Morocco
Indonesia
Philippines
Swaziland
Kazakhstan
Latvia
Jordan
Lithuania
Fiji
Dominica
Belize
Estonia
Peru
Ecuador
Tunisia
St. Lucia
Iran
Brazil
Trinidad & Tobago
Thailand
Colombia
Hungary
3110
3205
3290
3322
3403
3480
3562
3783
4032
4126
4346
4422
4429
4773
4826
4978
5462
6175
6444
6473
6536
6701
2374
2332
2728
2196
2950
2269
2050
3317
3140
2913
3222
3440
3556
4272
3288
3880
4656
5038
4731
3833
1689
5537
Food Cereals Meats Aquatic Dairy Oils & fats Fruits & vegs Other food Beverages Alcohol
50.1
51.3
48.9
31.4
56.1
45.7
40.9
43.5
44.8
42.3
33.7
36.4
32.5
30.0
39.0
48.8
34.8
23.2
23.7
30.2
23.4
24.2
20.2
33.5
29.7
25.3
30.8
12.9
8.4
12.9
15.7
16.9
10.9
16.1
21.3
14.8
13.8
14.4
24.8
16.8
14.2
16.1
21.5
10.9
19.9
5.1
14.5
22.9
17.8
18.9
27.6
20.7
12.8
11.5
6.5
20.3
22.2
19.5
13.6
21.2
23.9
24.5
16.0
18.6
21.8
20.5
1.9
8.7
14.5
2.3
1.6
3.0
1.7
3.5
11.8
9.6
0.9
3.0
4.7
5.5
5.0
7.4
1.7
2.3
5.7
3.3
2.1
0.8
34
6.6
5.7
6.7
9.4
9.2
14.9
11.3
14.1
7.0
8.7
10.3
13.2
9.6
12.9
10.6
11.5
11.2
14.0
9.4
5.2
11.2
12.8
8.6
4.7
1.8
4.4
7.1
4.3
9.0
4.8
4.0
2.1
4.4
4.7
3.7
5.9
4.3
2.8
7.0
3.6
5.1
2.8
4.6
4.7
18.4
23.7
11.1
11.3
9.3
17.8
15.5
12.0
21.2
29.7
7.4
10.2
21.4
21.1
28.2
30.3
18.6
14.8
14.6
16.4
17.3
12.7
12.6
7.2
9.8
12.5
13.0
9.3
16.4
12.2
9.4
15.5
44.7
11.2
7.9
10.5
10.8
4.7
8.1
11.5
17.9
9.0
8.1
14.1
2.5
0.7
2.4
4.6
0.3
1.1
2.0
1.2
5.9
1.6
4.4
1.0
3.3
3.4
2.3
2.7
0.4
2.8
4.6
10.0
4.5
3.9
1.6
0.4
4.4
5.2
8.9
15.5
0.0
15.8
7.4
3.4
9.4
13.2
5.0
5.0
1.1
2.9
0.0
1.8
8.7
12.4
3.1
13.3
Appendix 3: List of the data (continued)
Countries
Slovakia
Gabon
St. Kitts & Nevis
Botswana
Mauritius
Antigua & Barbuda
Argentina
Czech Republic
Oman
Bahamas
Barbados
Slovenia
Greece
Korea
Portugal
Taiwan
Spain
Qatar
Bahrain
New Zealand
Ireland
Israel
Finland
GDP
M
6994
7392
7552
7560
8550
9182
9259
9945
9980
10779
10924
10942
12061
12275
13218
14222
15041
16077
16091
17258
17490
17752
18515
4854
3617
5007
2610
6295
4946
6848
7442
6876
8371
8077
9097
10900
8165
11042
6990
11550
8854
6159
12746
12044
15785
13316
Food Cereals Meats Aquatic Dairy Oils & fats Fruits & vegs Other food Beverages Alcohol
35.0
51.6
39.4
39.9
29.6
38.9
33.3
27.0
26.0
40.0
12.2
23.3
22.2
35.3
25.5
25.5
18.8
28.1
32.6
16.3
18.1
20.0
16.1
10.0
10.7
25.8
24.2
10.1
25.8
14.6
10.2
16.8
14.1
13.1
10.1
7.3
20.7
13.1
15.0
12.5
10.6
13.1
12.6
9.5
14.4
11.4
20.6
9.2
7.6
11.9
15.6
7.6
26.1
21.3
16.4
23.3
22.1
22.1
16.0
12.7
22.4
18.7
24.0
23.2
13.5
13.9
16.4
14.1
15.2
1.7
14.5
10.4
0.7
8.4
10.4
1.4
1.8
7.7
6.2
4.9
1.9
4.5
11.7
12.2
15.9
10.3
5.6
9.5
1.7
2.0
2.5
2.8
35
13.9
3.9
11.0
4.7
10.5
11.0
12.7
11.6
11.2
11.1
9.0
11.4
13.6
5.0
8.5
4.6
2.5
4.7
2.3
5.2
4.7
3.5
4.0
4.6
5.5
3.3
3.0
5.4
0.9
3.7
11.6
10.4
10.4
9.2
10.1
13.0
12.6
4.8
2.7
3.1
2.3
2.7
1.9
2.0
13.4
44.8
16.6
6.2
17.9
16.6
17.2
12.4
22.3
11.3
18.0
17.2
17.3
21.2
14.5
10.8
13.8
21.0
25.5
16.9
13.4
19.4
13.5
10.4
4.9
15.1
13.6
7.8
15.1
9.5
10.6
16.3
6.7
11.5
10.2
11.4
10.0
4.2
39.7
5.3
19.2
17.0
10.5
8.6
16.1
11.1
2.4
2.6
2.5
5.4
4.8
3.1
5.3
3.4
3.7
2.5
12.1
3.2
2.5
6.4
1.1
15.1
5.1
5.2
26.0
13.4
4.6
4.6
15.2
0.0
15.4
3.9
10.4
4.9
4.6
10.5
2.4
5.1
5.1
4.2
4.5
8.9
2.6
6.6
0.0
0.0
16.6
12.6
3.9
19.8
Appendix 3: List of the data (continued)
Countries
Sweden
UK
Netherlands
Italy
Australia
Bermuda
Germany
Iceland
France
Austria
Canada
Belgium
Denmark
Norway
Japan
Hong Kong
Switzerland
Singapore
USA
Luxembourg
GDP
M
19269
19498
19869
19887
19894
20605
20646
20823
21242
21476
21726
21870
22151
22559
22770
23153
24838
25502
27333
29396
13848
15626
13374
14876
14633
14757
14865
16307
15776
15582
15651
15375
16450
14946
15281
15511
17031
9188
21437
19046
Food Cereals Meats Aquatic Dairy Oils & fats Fruits & vegs Other food Beverages Alcohol
14.3
17.6
14.4
17.8
16.2
14.8
13.9
20.5
16.6
14.8
12.9
15.8
15.4
17.4
15.9
11.0
15.8
13.9
10.6
18.7
11.4
8.3
12.4
11.3
13.5
10.1
14.9
11.9
10.9
13.4
11.4
10.8
8.9
7.7
22.3
9.0
10.7
10.3
11.4
8.9
15.2
12.6
18.7
23.6
16.9
11.5
20.3
16.4
24.9
21.0
16.5
24.7
20.4
16.3
7.8
22.7
16.5
13.3
19.6
18.3
4.4
2.3
2.2
5.4
3.1
3.0
1.9
5.0
4.8
1.6
2.7
6.1
2.0
4.9
17.0
19.7
1.8
15.0
1.2
2.3
11.7
6.9
12.6
13.9
9.7
7.8
7.1
11.6
11.8
11.3
11.2
11.0
11.1
12.8
4.8
3.4
15.2
5.0
8.6
7.8
2.3
1.3
2.2
3.9
1.7
4.3
2.3
1.6
2.9
3.8
2.1
3.9
2.2
1.5
0.7
3.3
2.0
1.8
1.8
1.9
14.4
12.0
15.7
19.1
18.3
13.5
8.3
10.8
12.4
14.1
18.1
12.4
11.9
11.1
12.8
11.8
17.0
18.1
14.7
11.6
13.1
9.2
12.3
6.6
11.6
29.4
17.1
15.2
11.0
11.0
8.6
10.2
14.7
15.7
11.5
12.2
10.6
11.3
14.1
6.1
3.2
4.2
3.8
2.3
6.6
6.8
4.2
9.0
2.6
2.9
3.7
3.3
4.3
7.2
6.7
6.9
3.7
6.6
8.8
1.7
14.0
30.5
9.9
4.9
7.2
10.0
13.2
11.3
10.9
11.2
13.0
7.7
12.9
11.8
8.3
4.5
13.6
8.6
10.7
13.8
Notes GDP = gross domestic product per capita in international dollars. M = total expenditure per capita in international dollars. Food = food budget shares over
total expenditure. Others are budget shares over total food expenditure. The figures for tobacco are omitted. Blank cells are due to missing values. Zero values
are due to rounding.
36
Appendix 4: Regression results
In order to run the regressions, eight dummy variables are defined as follows:
D1 = 1 for East Asia & Pacific and South Asia
= 0, otherwise
D2 = 1 for East Europe & Central Asia and Middle East & North Africa
= 0, otherwise
D3 = 1 for West Europe and North America
= 0, otherwise
D4 = 1 for Sub-Sahara Africa and Latin America & Caribbean
= 0, otherwise
D11
= 1 for low income economies
= 0, otherwise
D12 = 1 for lower middle income economies
= 0, otherwise
D13 = 1 for upper middle income economies
= 0, otherwise
D14 = 1 for high income economies
= 0, otherwise
For the “food” regressions reported in Table A1 and A2, the dependent variables are food
budget shares over total expenditure per capita (M) and, for individual food items, the
dependent variables are budget shares over total food expenditure (M). “Log” stands for
the natural logarithm and “mean” for the mean of the dependent variables. “adj. R2” is the
adjusted R-squared and “N” the number of observations. The following outliers or
countries with zero budget shares (due to missing observations) are identified and
excluded from relevant regressions:
Aquatic products:
Alcoholic drinks:
Beverages:
Dairy products:
Oils and fats:
Other foods:
Food:
Hong Kong, Japan, India, Singapore and Taiwan.
Bahrain, Botswana, Egypt, India, Iran, Jordan, Mali, Oman,
Pakistan, Qatar, Taiwan and the UK.
Barbados, Bolivia, Luxembourg, Nepal, Portugal, Sri Lanka,
Taiwan, Thailand and Zimbabwe.
Taiwan.
Taiwan
Belize, Bermuda, Luxembourg, Mali, Taiwan and Uzbekistan.
Amenia, Azerbaijan, Barbados, Gabon, Mali, Sri Lanka and
Zimbabwe.
37
Table A1 Estimation results from regressions incorporating geographical dummy variables
Intercept
Cereals
Meats
Fruits and vegetables
Aquatic products
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
Food
logM logM*D1 logM*D2 logM*D3 Adj. R2
D1
D2
D3
0.665
0.160
0.128
0.136
0.123
0.158
-0.119
0.079
0.014
0.110
-0.026
0.040
-0.065
0.089
0.095
0.045
0.124
0.092
0.645
0.261
-0.164
0.222
-0.120
0.259
-0.037
0.148
-0.262
0.193
-0.173
0.067
0.455
0.146
0.167
0.074
-0.059
0.141
-0.077
0.279
-0.106
0.237
0.392
0.277
-0.089
0.138
-0.099
0.192
-0.162
0.071
0.123
0.155
0.144
0.078
-0.039
0.151
-0.159
0.780
0.105
0.662
0.340
0.773
0.115
0.386
0.216
0.500
-0.382
0.229
0.279
0.433
0.039
0.219
-0.198
0.480
-0.068
0.023
0.005
0.019
0.010
0.023
0.025
0.011
0.008
0.016
0.008
0.006
0.021
0.013
-0.007
0.006
-0.003
0.013
-0.082
0.037
0.018
0.031
0.012
0.037
0.007
0.021
0.035
0.027
0.025
0.010
-0.065
0.021
-0.025
0.011
0.009
0.020
0.010
0.039
0.017
0.033
-0.056
0.039
0.009
0.019
0.015
0.027
0.021
0.010
-0.014
0.022
-0.018
0.011
0.008
0.021
0.017
0.101
-0.011
0.085
-0.052
0.100
-0.020
0.050
-0.022
0.065
0.050
0.030
-0.035
0.056
-0.007
0.028
0.027
0.062
1.256
0.142
0.461
0.182
0.284
0.206
-0.580
0.971
-0.109
0.018
-0.054
0.022
-0.031
0.025
0.055
0.101
38
N Mean
0.444
87
0.180
0.070
87
0.167
0.056
87
0.171
0.152
82
0.045
0.176
75
0.081
0.369
78
0.033
0.130
86
0.097
0.343
86
0.041
-0.027
81
0.111
0.840
80
0.330
Table A2 Estimation results from regressions incorporating income group dummy variables
Intercept
Cereals
Meats
Fruits and vegetables
Aquatic products
Alcoholic drinks
Beverages
Dairy products
Oils and fats
Other foods
Food
D11
D12
0.046 0.386 0.048
0.467 0.572 0.558
-0.030 0.036 0.521
0.403 0.493 0.481
0.375 -0.567 -0.238
0.469 0.574 0.560
-0.241 0.145 0.011
0.320 0.368 0.357
-0.167
0.034
0.080
0.229
-0.123
0.244
0.030
0.086
-0.270 0.560 0.525
0.282 0.345 0.337
0.005 -0.036 0.024
0.144 0.176 0.172
0.131 -0.020 -0.040
0.263 0.332 0.308
1.546
0.062
D13
0.059
0.623
0.544
0.537
-1.055
0.626
-0.242
0.386
0.330
0.313
0.059
0.125
0.427
0.376
0.108
0.192
0.245
0.341
logM logM*D11 logM*D12 logM*D13
0.009
0.060
0.027
0.052
-0.030
0.060
0.036
0.041
0.035
0.011
0.022
0.004
0.048
0.036
0.003
0.019
-0.002
0.034
-0.143
0.007
39
-0.030
0.079
-0.009
0.068
0.089
0.079
-0.015
0.050
-0.080
0.048
0.011
0.024
0.001
0.046
0.006
0.074
-0.072
0.064
0.036
0.075
0.003
0.047
-0.006
0.033
-0.036
0.012
-0.070
0.045
0.000
0.023
0.005
0.041
-0.002
0.082
-0.074
0.071
0.146
0.082
0.038
0.050
-0.046
0.042
-0.008
0.017
-0.055
0.049
-0.013
0.025
-0.033
0.045
R2
N Mean
0.464
87
0.180
0.077
87
0.167
0.068
87
0.171
0.092
82
0.045
0.113
75
0.081
0.294
78
0.033
0.007
86
0.097
0.242
86
0.041
-0.040
81
0.111
0.831
80
0.330
References
Barten, A.P. (1977), “The Systems of Consumer Demand Functions Approach: A
Review”, Econometrica, 45, 23-51.
Clements, K.W., F.E. Suhm and H. Theil (1979), “A Cross-country tabulation of Income
Elasticities of Demand”, Economics Letters, 3, 199-202.
Clements, K.W., S. Selvanathan and E.A Selvanathan (1995), “The Economic theory of
the Consumer”, in E.A. Selvanathan and K.W. Clements (eds.) Recent Development
in Applied Demand Analysis: Alcohol, Advertising and Global Consumption,
Berlin/Heidelberg: Spring-Verlag, chapter 1, 1-72.
Clements, K.W., Y. Wu and J. Zhang (2003), “Comparing International Consumption
Patterns”, unpublished research report, School of Economics and Commerce,
University of Western Australia.
Deaton, A. and J. Muellbauer (1984), Economics and Consumer Behaviour, Cambridge:
Cambridge university Press.
Easterly, W. and H. Yu (2000), “Global Development Network Growth Database”,
unpublished, the World Bank.
Huang, J. and H. Bouis (2001), “Structural Changes in the Demand for Food in Asia:
Empirical Evidence from Taiwan”, Agricultural Economics, 26, 57-69.
Kravis, I.B., A.W. Heston and R. Summers (1982), World Product and Income:
International Comparisons of Real Gross Product, Baltimore: Johns Hopkins
University Press.
Leser, C.E.V. (1963), ‘Forms of Engel functions’, Econometrica, 8, 694-703.
40
National Statistical Bureau, 1996, China’s statistical yearbook 1996, Beijing: Statistical
Publishing House.
Regmi, A., M.S. Deepak, J. Seale and J. Bernstein (2001), “Cross-Country Analysis of
Food Comparison Patterns”, in A. Regmi (ed.) Changing Structure of Global Food
Consumption and Trade, Economic Research Service, US Department of
Agriculture.
Rimmer, M.T. and A.A. Powell (1992), “Demand Patterns Across the Development
Spectrum: Estimates of AIDADS”, IMPACT Working Paper, OP-75, Monash
University.
Seale, J. and A. Regmi (2002), “International Consumption Patterns: Evidence from the
1996 International Comparison Project”, unpublished manuscript, Department of
Food and Resource Economics, University of Florida.
Seale, J. and H. Theil (1986), “Working’s Model for Food in the Four Phases of the
International Comparison Project”, Economics Letters, 22, 103-104.
Statistical Yearbook of the Republic of China 1996, Directorate-General of Budget,
Accounting and Statistics, Executive Yuan, Republic of China.
Theil, H. and K.W. Clements (1987), Applied Demand Analysis: Results from SystemWide Approaches, Cambridge: Ballinger Publishing Company.
Working, H. (1943), “Statistical laws of family expenditure”, Journal of the American
Statistical Association, 38, 43-56.
41