Vous êtes sur la page 1sur 16

______________________________________________________________________________

Household Food Consumption Characteristics in


Vietnam: an Empirical Analysis
Charles F. Nicholson, Research Associate;
REI-Vietnam, 1855-C Austin Bluffs Parkway, Colorado Springs, Colorado 80918

______________________________________________________________________________

Abstract

This study provides updated estimates of price elasticities and expenditure elasticities
using recent household data. The estimates provide a general understanding of the potential shift
in household consumption patterns due to a wide variety of reasonable development scenarios,
including trade liberalization, domestic institutional reform, the adoption of cost-reducing
technology and the general upward trend of household income for low-income households. These
critical parameters are relevant to both public sector and private sector agents, who have a stake
in the markets for food items in Vietnam.

Introduction

Vietnam’s economic development during the last decade has received notable attention
for the significant improvements in national income and household welfare. Market reforms
deserve most of the credit for these improvements as the government has followed a course that
consistently moves away from central planning and towards a market-oriented economy. Striking
advances have been noted in a variety of areas, including poverty reduction, children’s education,
total production levels, and average household income. In addition, these accomplishments have
not been associated with common ailments of development, exhibiting only moderate rates of
inflation, only small changes in income disparity and urbanization that has not resulted in much
strain on infrastructure.

This paper focuses attention on the current features of food demand in Vietnam, taking
note of the development path that is being followed by Vietnam. The effects of continued
development on the markets for household food items deserves particular attention because of the
particular development stage Vietnam is currently in. That is, despite remarkable growth over the
past decade, Vietnam is still a low-income country. As such, household food consumption
represents a large share of total household expenditures. Thus, an understanding of the nature of
household food consumption patterns will provide further insight into major markets in Vietnam
and the ways in which those markets may change as household incomes increase.

Food policies may also change markets in significant ways. For instance, one of the most
notable accomplishments is the drastic improvement in rice production that has turned the
country from an importer of rice to one of the world’s largest exporters of rice. Much credit for
that accomplishment goes to the progressive liberalization of the rice market from collective

1
farming in the 1980’s to a contract farming system in the late 80’s to secure land rights in the
early 90’s and finally to relaxed export restrictions in the late 90’s. Minot and Goletti (2000)
provide an excellent review and analysis of the impact of rice market liberalization in Vietnam.
They note that export quotas have been relaxed to nearly non-binding levels and that little
incentive remains for further liberalization. Specifically, as a large exporter of rice Vietnam has
incentive to maintain some form of trade restriction, either an export quota or export tax, in order
to capitalize on it’s ability to influence the world price. Thus, the effects on household
consumption patterns due to rice market liberalization may have already come to completion.
However, many other relevant food policies continue to be discussed, which include policies that
dictate the trade of meat, vegetables, tea and coffee, any one of which can have price increases
that shift household consumption patterns.

This study provides updated estimates of the expected changes in food demand given
price or income changes. In particular, price elasticities and expenditure elasticities are estimated
using recent household data. These critical parameters are relevant to both public sector and
private sector interests in the markets for food items in Vietnam.

The rest of the paper is organized as follows. Section 1 briefly explains a model of
household food demand. Section 2 presents the empirical model and a description of the data.
Section 3 discusses the regression results and the elasticities, which are derived from the
regression results. Section 4 summarizes the main observations of the analysis of household food
consumption.

1. The model of household food demand

We start with a two-stage budgeting process for the household. In the first stage the
household allocates total available income across groups of consumption possibilities. In
particular, we will assume that one of those groups represents prominent food items. Prominent
food items do not necessarily have to be consumed and, therefore, should not be regarded as
essential in the sense that the household must consume the items to survive. However, prominent
food items will include essential food items. The model, therefore, implicitly permits households
to have zero expenditures on items in the prominent foods group.

Once the household has set a budget for the prominent food group, the second stage of
the budgeting process involves sub-allocation of income for the purchase of individual food items
in the prominent food group. The model assumes that the form of household utility is separable,
according to the broad groups of consumption possibilities.1

The almost ideal demand system of Deaton and Muellbauer (1980b) has been widely
applied for the estimation of demand elasticities. The model represents the technique of choice
because of the linearity of the model, the link with a rich theory of consumer demand and
straightforward means of calculating elasticities from the estimated parameters. As a result, the
following discussion draws heavily from Deaton and Muellbauer (1980b). In particular, I will
adopt their functional form for the minimum household expenditure function given a particular
utility level.

log e(p, u ) = a (p) + ub(p)


where
1
The reader should consult Deaton and Muellbauer (1980a) for a complete discussion of the
theoretical implications of the household decision model discussed here.

2
1
a (p) = α 0 + ∑ α k log( pk ) + ∑∑ γ klc log( pk ) log( pl )
k 2 k l
b(p) = β 0 ∏ pkβ k
k
The budget share that is based on the Hicksian demand for good k is given by

∂ log(e(p, u )) ∂a (p) ∂b(p)


wk = = +u
∂ log( pk ) ∂ log( pk ) ∂ log( pk )

After substitution of direct utility for indirect utility and employing the explicit forms of
the derivaties, the budget share equations, which are based on Marshallian demands, can now be
written as
1 1
wk = α k + ∑
2 l
γ klc log( pl ) + ∑ γ cjk log( pj ) + βk log( y / P )
2 j
wk = α k + ∑ γ kl log( pl ) + βk log( y / P )
l

where the parameters γ kl are defined as


1
γ kl = (γ klc + γ lkc ) = γ lk
2

The coefficients from the system of equations described above can be used to obtain
elasticities of demand for the food items in the demand system. The following expenditure and
price elasticities are close approximates to the exact formulae derived by Green and Alston
(1990).
βi γ γ β
ei = + 1; ηii = ii + βi − 1; ηij = ij − i wj
wi wi wi wi

The values of the elasticities in represent dimensionless demand sensitivity parameters to


changes in prices and income.

2. The empirical model and data

The data used in this analysis comes from the 1997-1998 Vietnam Living Standards
Survey (VNLSS) and represents some of the most recent data available on household
consumption patterns in Vietnam. The survey covers 6000 households in all provinces in
Vietnam and includes a large section on household expenditures, from which the data in this
analysis is derived. In particular, the household expenditure data on 22 food items was used to
calculate household budget shares for each of the 22 food items. The household expenditure data
indicated that these 22 food items represent over 80% of total household food expenditures, and,
therefore, is properly considered a group of prominent food items. The survey also included
commodity prices at the level of the commune for the sampled households. Thus, local market
prices were available for most of the consumption goods in the survey. In the few cases where
prices were not available (7.7 percent of items in the price data), regional prices were estimated
and then inflated to account for transportation costs incurred by the household in traveling to
other areas to purchase the good, which was not available in the commune. Of the 6000

3
households in the survey, only 6 did not have useable data so that the analysis of this paper uses
5994 households in the data set.

The Stone household price index, S h = ∑w


k
k log( phk ) , was used in this analysis as an

approximation to the price index, P, which is based on a(P), and as a means of deflating total
expenditures in equation . The Stone index is a convenient means of maintaining the linearity of
the model and maintaining reasonable approximations for the estimated parameters. Because the
system is linear, we can apply the technique of seemingly unrelated regressions (Zellner (1962))
to get efficient estimates of the parameters. The model presented here is frequently called the
Linear Approximate Almost Ideal Demand System (LA/AIDS) and is commonly employed to
obtain estimates for elasticities (Cozzarin and Gilmour (1998))

Nearly all of the households consumed ordinary rice. But for other food items, the
incidence of consumption was less than 50% of households. Because not all households consume
all commodities, a problem arises in how to appropriately treat zero household expenditures for
some commodities. Using household data to estimate demand parameters for commodities
requires procedures that account for the bias that results when linear estimators are applied to
truncated, or censored, data. Heckman (1979) derives a simple, two-stage estimation technique
as a solution to the problem of estimating a single equation system. I will use an extension of the
Heckman sample selectivity approach is applied to a system of demand equations prescribed by
the model of the almost ideal demand systems. Heien and Wessells (1990) applied such an
approach to the estimation of food demand systems, focussing on dairy products. Saha, Capps
and Byrne (1997) show that when the Heckman correction is applied to single equation systems,
the resulting marginal effects must also take into account changes in the inverse Mills Ratio.
When applying the procedure to a system of equations, the changes in the inverse Mills Ratio
hardly alter the calculation of the marginal effects.

Following Heien and Wessells (1990), the Heckman procedure for a system of equations
can be formalized as follows. Consider the task of estimating the parameters, β , for the linear
model
Yh = β ′ X h + ε h for h = 1,.., H

We assume that at some price the household will consume a non-zero amount of the
commodity. Because negative amounts cannot be consumed the distribution of the error term is
truncated.
Yh = β ′ X h + ε h for Yh > 0 otherwise Yh = 0
which implies
ε h > − β ′ X h for Yh > 0 otherwise Yh = 0

OLS estimates of the parameters will be biased. Consider instead estimation of the
sample that only includes non-zero amounts of the dependent variable.
E ( yh | yh > 0) = β ′ X h + E (εh | εh > − β ′ X h )
φ ( β ′X h σ )
E ( yh | yh > 0) = β ′ X h + σ
Φ( β ′ X h σ )
where use has been made of the fact that

4
E (ε h | ε h > − β ′ X h ) = σ E (εh σ | εh σ > − β ′ Xh σ )
1
=σ E (ε h σ )
Prob( ε h σ > − β ′ X h σ )
σ
= φ (β ′X h σ )
Φ( β ′X h σ )

In general, the appropriate estimation procedure requires a non-linear approach.


Alternatively, the Heckman two-step procedure also produces consistent estimates. Consider the
consumer’s choice problem of
I h = β ′ X h + ε h where
I h = 1 if Yh > 0
I h = 0 otherwise
Applying the probit model on the latent variable, I h , produces estimates of β / σ , which
φ ( β ′X h σ )
can be used in constructing the explanatory variable , the Mills ratio. Inserting this
Φ( β ′ X h σ )
explanatory variable into the uncensored regression produces unbiased parameter estimates when
there are no restrictions on the parameter estimates. Thus, in this paper I incorporate the Mills
ratio in all the budget share equations, except for those that are associated with a high incidence
of consumption.

Table 1 presents the 22 food items that are used in the analysis. The third column
identifies two specific variables for each food item. According to the dictates of the model, the
variables include the natural logarithm of price in Vietnamese Dong and the budget share, which
is the fraction of the group expenditures that is spent on the food item. Note that ordinary rice
represents the largest food expenditure, using more than 54% of the budget. Pork also represents
a prominent consumption good with a budget share of 17%. Thus, pork and rice represent nearly
three-quarters of expenditures. The last row of Table 1 presents household expenditures in
Vietnamese Dong which have been deflated by the Stone price index. Space considerations
prevent the presentation of the Mills ratios, which are also used as variables for most of the
equations of . The regressions that were used to construct the Mills ratios included the price and
income variables shown in Table 1, plus two additional variables-household size and income
quintile ranking for the household. Mills ratios were not constructed for 6 food items (ordinary
rice, pork, spinach, fish sauce, salt and monosodium glutamate) which had nearly complete
representation in the household data because, in those cases, selectivity bias is negligible.

Table 1. Variable descriptions

Variable Description Variable name Mean St. Dev.

Ordinary rice Natural log of lporice 1.2388 0.1479


price (Dong)
Budget share worice 0.5403 0.1721
Sticky rice Natural log of lpsrice 1.5438 0.1736
price (Dong)
Budget share wsrice 0.0124 0.0278

5
Corn Natural log of lpcorn 0.9000 0.2915
price (Dong)
Budget share wcorn 0.0045 0.0239
Cassava Natural log of lpcass -0.0312 0.5137
price (Dong)
Budget share wcass 0.0026 0.0189
Potato Natural log of lppot 0.2971 0.4066
price (Dong)
Budget share wpot 0.0055 0.0123
Mien (Cassava Natural log of lpmien 2.2696 0.2816
vermicelli) price (Dong)
Budget share wmien 0.0030 0.0120
Pork Natural log of lppork 3.0406 0.1969
price (Dong)
Budget share wpork 0.1747 0.1166
Beef Natural log of lpbeef 3.4227 0.1955
price (Dong)
Budget share wbeef 0.0131 0.0326
Chicken Natural log of lpchick 3.0268 0.1571
price (Dong)
Budget share wchick 0.0431 0.0489
Duck Natural log of lpduck 2.5119 0.2288
price (Dong)
Budget share wduck 0.0112 0.0213
Eggs Natural log of lpeggs -0.0203 0.1240
price (Dong)
Budget share weggs 0.0219 0.0283
Tofu Natural log of lptofu 1.3159 0.2911
price (Dong)
Budget share wtofu 0.0206 0.0264
Spinach Natural log of lpspin 0.1138 0.4312
price (Dong)
Budget share wspin 0.0182 0.0186
Cabbage Natural log of lpcabb 0.8247 0.4329
price (Dong)
Budget share wcabb 0.0079 0.0125
Tomato Natural log of lptom 1.3132 0.3821
price (Dong)
Budget share wtom 0.0116 0.0145
Orange Natural log of lporang 1.8971 0.3172
price (Dong
Budget share worang 0.0093 0.0213
Banana Natural log of lpban 0.7315 0.3370
price (Dong)
Budget share wban 0.0172 0.0233
Mango Natural log of lpmang 2.5206 0.3390
price (Dong)
Budget share wmang 0.0049 0.0121
Fish Sauce Natural log of lpsauc 1.5530 0.4561
price (Dong)
Budget share wsauc 0.0231 0.0175

6
Salt Natural log of lpsalt -0.0003 0.2966
price (Dong)
Budget share wsalt 0.0058 0.0049
Monosodium Natural log of lpmsg 2.5141 0.0865
Glutamate price (Dong)
Budget share wmsg 0.0316 0.0201
Tea Natural log of lptea 3.4068 0.3363
price (Dong)
Budget share wtea 0.0172 0.0240
Deflated Natural log of lexp 6.5134 0.4874
expenditure deflated
expenditure
(Dong)

3. The results

Tables 2a-2c present the results of the regressions of budget share equations (equations ).
Note that no restrictions on the parameters are imposed on the model. Because the model allows
the household to have zero expenditures for some food items and because the data includes many
instances of zero expenditures, the model restrictions that emerge from interior solutions are not
strictly applicable to this system.2 Consequently, Tables 2a-2c does not exhibit Slutsky
symmetry, which is frequently imposed on LA/AIDS models.

Most of the coefficients in the model are statistically significant. In particular, virtually
all of the coefficients on the Mills ratios are significant and mainly negative, indicating the
necessity of correcting for selectivity bias in the data. Using these estimates and the relations in ,
we can obtain price and expenditure elasticities, which represent the parameters of interest.

Table 2a. Seemingly unrelated regression results

worice wsrice wcorn wcass wpot wmien wpork

const 1.478 0.054 -0.016 0.029 -0.014 0.023 -0.414


17.482 3.587 1.207 2.776 1.387 2.558 6.760
lporice 0.112 0.002 -0.019 -0.004 0.006 -0.002 -0.008
6.326 0.577 6.680 1.662 4.016 1.373 0.621
lpsrice -0.195 0.005 -0.010 0.007 -0.005 -0.001 0.092
13.521 1.517 4.696 3.615 4.747 0.386 8.820
lpcorn 0.017 0.001 0.016 0.012 -0.004 0.000 -0.005
1.960 0.346 5.731 7.763 4.535 0.299 0.853
lpcass -0.056 -0.004 0.001 0.002 0.003 0.001 0.025
11.452 4.126 1.095 2.228 6.490 1.374 6.972
lppot 0.001 0.002 0.009 0.005 -0.009 -0.002 -0.006
0.132 1.666 8.899 5.669 9.052 3.069 1.242

2
Deaton (1997) states, “Unless we build different models for the different ‘regimes’ comprising
different combinations of purchased goods, and then link these regimes within a general statistical model of
selection, we cannot follow the parameters of the theory through into the empirical specification. As a
result, even though the utility restrictions hold for a single agent who buys all of the goods, they will not
hold in the survey data.” (p. 310)

7
lpmien 0.028 -0.003 0.002 0.009 -0.003 0.002 -0.027
3.101 1.624 1.667 6.955 4.271 2.514 4.103
lppork -0.077 -0.030 -0.004 0.040 0.000 -0.003 0.070
3.998 6.100 1.320 12.177 0.165 1.479 4.977
lpbeef -0.126 0.016 -0.004 -0.007 -0.010 -0.003 0.060
8.485 4.569 1.399 3.586 7.594 2.488 5.565
lpchick -0.089 -0.001 -0.004 -0.020 0.009 -0.001 0.019
5.130 0.465 1.310 8.354 5.314 0.721 1.484
lpduck -0.011 0.003 0.011 0.006 0.007 0.002 0.001
1.048 1.661 6.085 4.020 5.975 2.286 0.163
lpeggs -0.024 0.001 -0.009 -0.021 -0.006 -0.003 0.046
1.199 0.154 2.498 6.050 3.484 1.561 3.179
lptofu 0.064 -0.009 0.011 0.004 -0.004 0.001 -0.002
7.659 5.157 7.657 3.403 4.866 1.025 0.379
lpspin -0.018 -0.006 0.002 -0.001 -0.004 0.001 0.011
3.143 3.926 1.949 1.948 7.188 1.742 2.606
lpcabb -0.023 -0.005 -0.001 -0.002 -0.003 -0.001 0.010
3.619 4.127 1.330 2.859 4.869 1.671 2.176
lptom 0.002 -0.004 -0.005 -0.006 0.004 0.001 -0.008
0.359 3.960 4.107 6.430 6.517 0.832 1.640
lporang -0.027 0.012 0.001 -0.004 0.002 0.002 0.002
3.799 8.247 0.518 3.479 2.742 2.444 0.292
lpban -0.023 0.001 -0.013 -0.008 0.016 -0.002 0.015
3.478 0.707 8.956 9.017 8.290 2.733 3.165
lpmang -0.041 0.001 0.012 0.005 -0.002 0.001 0.003
6.230 0.490 9.502 5.516 3.353 1.359 0.727
lpsauc -0.035 -0.003 -0.006 -0.002 0.003 0.000 0.005
7.014 3.169 6.343 3.587 5.502 0.727 1.366
lpsalt -0.028 0.001 0.009 -0.009 -0.002 0.000 0.005
3.767 0.489 7.843 7.228 3.205 0.367 0.923
lpmsg 0.229 -0.009 0.041 0.000 -0.018 -0.001 -0.101
7.662 1.593 8.479 0.046 5.335 0.250 4.649
lptea -0.035 -0.002 -0.004 -0.008 0.008 0.003 0.011
4.787 1.441 3.609 9.027 6.990 3.308 2.140
lpexp -0.030 0.002 -0.005 -0.006 0.002 -0.002 0.037
7.323 1.997 6.997 10.351 3.889 3.820 12.502
Mills 0.002 -0.021 -0.026 0.018 -0.003
Ratio 0.731 10.215 11.448 5.785 2.851
Note: T-statistics in italics.

Table 2b. Seemingly unrelated regression results

wbeef wchick wduck weggs wtofu wspin wcabb

const -0.105 -0.212 0.037 0.000 0.017 0.067 0.030


5.758 7.462 2.874 0.002 1.186 6.724 3.786
lporice 0.000 -0.010 -0.006 -0.007 -0.020 -0.007 0.002
0.105 1.910 2.377 2.193 6.546 3.431 1.669
lpsrice 0.013 0.023 -0.007 0.010 0.018 0.008 0.001
4.265 4.962 3.486 3.731 6.715 4.903 1.184

8
lpcorn -0.006 0.010 0.004 -0.001 0.000 -0.003 0.000
3.258 3.549 2.847 0.837 0.038 2.916 0.388
lpcass 0.007 -0.004 -0.001 0.007 0.002 0.006 0.002
7.586 2.566 1.486 7.485 2.572 10.718 3.701
lppot 0.005 -0.008 -0.001 -0.002 0.004 -0.002 0.000
3.281 4.143 0.861 2.229 3.294 2.689 0.064
lpmien -0.005 0.003 0.001 -0.003 -0.001 -0.004 0.001
2.485 1.181 0.825 1.738 0.744 3.884 1.810
lppork 0.002 0.023 0.006 -0.003 -0.008 0.003 0.001
0.417 3.785 2.385 0.982 2.454 1.114 0.287
lpbeef 0.002 0.012 0.007 0.011 0.013 0.005 0.003
0.688 2.598 3.357 4.139 4.862 2.771 2.289
lpchick 0.017 0.011 0.006 0.013 0.004 0.005 0.006
4.823 2.000 2.426 4.291 1.371 2.217 4.499
lpduck 0.003 -0.007 -0.003 0.003 0.001 0.002 -0.001
1.309 2.229 1.618 1.771 0.522 1.195 1.043
lpeggs -0.008 -0.008 -0.001 0.005 -0.007 0.004 -0.006
1.928 1.277 0.472 1.491 1.961 1.594 3.329
lptofu -0.007 -0.011 0.005 -0.007 -0.008 -0.004 0.000
4.172 4.346 3.869 4.893 5.221 3.955 0.131
lpspin 0.000 0.004 0.001 0.004 0.001 0.002 0.000
0.411 2.459 1.503 3.535 0.563 3.008 0.489
lpcabb 0.006 0.007 0.002 0.001 -0.003 0.004 0.000
4.752 3.508 1.931 0.964 2.512 5.369 0.520
lptom 0.000 -0.003 -0.002 -0.002 -0.001 0.002 0.000
0.104 1.754 1.799 1.778 0.949 2.084 0.760
lporang -0.002 0.002 0.002 0.001 0.002 -0.005 0.000
1.380 0.848 1.837 0.405 1.809 6.345 0.742
lpban -0.001 -0.004 -0.003 -0.001 0.002 0.000 -0.002
0.391 2.119 3.746 1.169 1.424 0.611 3.517
lpmang 0.008 0.010 0.000 0.006 0.002 0.003 0.000
6.133 4.917 0.404 4.976 1.527 3.545 0.350
lpsauc 0.007 -0.004 -0.002 0.005 0.004 0.003 0.001
6.377 2.631 3.547 5.897 4.189 5.545 2.463
lpsalt 0.003 0.000 0.000 0.005 0.003 0.000 0.002
2.220 0.116 0.270 3.887 2.108 0.508 3.371
lpmsg -0.010 0.020 -0.025 -0.009 -0.005 -0.012 -0.012
1.651 2.185 5.486 1.623 0.982 3.401 4.482
lptea 0.000 0.004 0.003 -0.002 0.002 -0.004 -0.002
0.215 1.964 3.015 1.250 1.208 4.924 2.931
lpexp 0.008 0.005 -0.001 -0.004 -0.004 -0.006 -0.002
7.817 3.188 0.652 5.233 4.949 12.646 6.273
Mills -0.005 -0.025 -0.012 -0.021 -0.005 -0.006
Ratio 4.822 8.981 5.667 11.292 2.339 6.058
Note: T-statistics in italics.

Table 2c. Seemingly unrelated regression results

wtom worang wban wmang wsauc wsalt wmsg wtea

9
const 0.008 -0.013 0.075 -0.029 0.077 0.030 0.093 0.011
0.983 1.087 5.906 3.783 8.120 11.635 8.664 0.828
lporice -0.002 0.005 0.004 -0.011 -0.005 -0.007 -0.016 -0.007
1.149 2.245 1.334 7.861 2.420 12.693 6.911 2.136
lpsrice -0.005 0.009 -0.004 0.008 -0.002 0.000 -0.005 0.004
3.730 4.378 1.662 5.149 1.265 1.085 2.947 1.839
lpcorn -0.004 -0.004 -0.002 0.002 -0.005 0.001 0.003 0.001
5.466 3.310 1.153 3.521 5.084 3.416 2.918 1.032
lpcass 0.003 0.006 0.005 0.001 0.002 0.000 -0.005 0.001
7.702 7.823 5.540 3.587 3.410 1.995 8.681 0.679
lppot -0.001 -0.001 0.000 0.002 0.002 0.001 0.002 -0.004
1.905 0.641 0.427 3.285 3.144 2.952 2.714 3.794
lpmien 0.000 -0.001 -0.002 0.008 -0.001 0.002 0.003 -0.003
0.122 0.567 1.698 9.099 0.648 5.599 2.387 2.335
lppork 0.009 -0.006 0.003 0.000 0.002 0.000 -0.014 -0.003
4.774 2.420 0.937 0.320 0.999 0.231 5.721 1.166
lpbeef 0.007 0.009 0.002 0.006 0.002 -0.001 0.003 -0.007
5.143 4.435 0.908 4.825 1.294 1.725 1.769 2.997
lpchick 0.001 0.014 -0.001 0.007 -0.003 -0.001 0.007 0.008
0.647 6.037 0.272 5.213 1.640 1.240 3.075 2.968
lpduck -0.001 0.000 0.001 0.001 0.000 0.001 -0.008 0.007
1.076 0.227 0.486 1.656 0.355 1.686 6.064 3.984
lpeggs 0.004 0.002 -0.003 -0.008 -0.010 -0.001 -0.002 -0.004
2.298 0.724 0.829 5.066 4.432 2.014 0.616 1.364
lptofu -0.001 -0.007 -0.004 -0.006 0.001 0.002 0.003 -0.004
1.267 6.161 3.527 8.664 0.656 7.653 2.779 2.621
lpspin 0.003 0.004 0.001 0.000 -0.001 0.000 0.000 -0.001
4.979 4.819 1.613 0.257 1.086 0.434 0.354 1.112
lpcabb 0.000 0.002 0.001 0.006 0.002 -0.001 -0.002 0.000
0.482 2.197 1.223 8.806 3.199 2.649 2.354 0.137
lptom 0.002 -0.001 0.002 0.000 0.006 0.000 0.004 -0.003
3.569 1.037 1.639 0.228 8.422 0.838 4.302 2.577
lporang 0.000 0.000 -0.001 -0.002 -0.003 0.000 0.000 0.001
0.727 0.080 1.230 4.034 3.721 2.181 0.361 0.826
lpban -0.001 0.002 -0.001 0.002 0.003 -0.001 -0.001 0.002
1.350 1.852 1.288 4.710 4.432 2.908 1.037 1.688
lpmang 0.003 0.004 0.004 -0.004 0.001 -0.001 -0.003 0.003
4.857 4.995 4.148 5.435 1.563 2.709 3.049 2.470
lpsauc 0.001 0.004 0.001 0.001 0.002 -0.001 0.000 0.001
3.193 5.198 1.048 2.654 3.862 6.516 0.065 1.052
lpsalt 0.004 0.002 0.003 0.003 -0.002 0.000 0.003 0.000
5.261 2.187 2.404 5.709 2.143 0.551 2.975 0.094
lpmsg -0.008 -0.024 -0.007 -0.016 -0.014 0.001 0.011 0.004
2.696 5.818 1.427 6.505 4.018 0.622 2.890 0.808
lptea -0.001 0.005 0.002 0.004 0.005 -0.001 0.004 0.006
1.284 4.739 1.476 6.026 6.330 2.816 3.912 5.226
lpexp -0.003 -0.002 -0.009 0.001 -0.006 -0.003 -0.008 -0.003
7.675 2.865 13.088 4.059 13.599 21.559 15.294 2.581
Mills -0.008 -0.002 -0.016 0.001 -0.012
Ratio 8.664 2.905 4.867 1.756 4.362
Note: T-statistics in italics.

10
Tables 3a-3c presents the price elasticity of food demand for households in Vietnam. All
of the own-price elasticities have the correct magnitude and sign, except for corn. In addition,
certain groups of food items exhibit expected price relationships. For example, the price
elasticities within the sub-group of meats shows the substitutability of one type of meat for
another. Specifically, an increase in the price of pork in Table 3a increases the demand for beef,
chicken and duck. Similarly, an increase in the price of any of the meat items increases the
demand of all the other meat items, demonstrating the substitutability of one meat item for
another.

Moreover, the elasticities in Tables 3a-3c show the relative magnitudes of meat demand
increases due to a price increase in a substitute meat. In particular, note that increases in the price
of pork in Table 3a produces relatively large increases in the demand for chicken and duck but a
relatively small increase in the demand for beef, indicating that the market for beef in Vietnam
will hardly be affected by changes in policies related to pork, like further liberalization in the
trade of pork. Also, an increase in the price of beef increases the demand for pork, chicken and
duck in roughly the same magnitudes. However, a change in the price of chicken has a much
larger impact on the demand for beef than for the other meats. Apart from the elasticity of
demand for beef with respect to the price of chicken, the elasticities of this group have reasonable
magnitudes such that the own price effect is larger than the cross price effect. The elasticities of
the meat group provide meaningful estimates of the magnitude of potential shifts in household
consumption patterns.

The group of food items that represents starchy products-ordinary rice, sticky rice,
cassava and potatoes-cannot be considered pure substitutes. Consider an increase in the price of
ordinary rice in Table 3a. Interestingly, ordinary rice and cassava appear to be complements,
such that a price increase in one results in a reduction in the demand of the other. But potatoes
and cassava appear to be substitutes, such that a price increase in one results in an increase in the
demand of the other. Thus, within the group of starchy household food products, little prediction
can be made about possible shifts in household consumption patterns.

Table 3a. Cross price elasticities


Percentage change given a 1% change in the price of . . .
in the demand of ordinary sticky corn cassava potato mien pork beef
rice rice

ordinary rice -0.76 -0.36 0.03 -0.10 0.00 0.05 -0.13 -0.23
sticky rice 0.06 -0.64 0.04 -0.29 0.15 -0.21 -2.42 1.26
corn -3.49 -2.23 2.55 0.21 1.98 0.52 -0.66 -0.84
cassava -0.10 2.51 4.52 -0.41 1.75 3.33 15.67 -2.44
potato 0.93 -0.98 -0.64 0.59 -2.64 -0.61 -0.02 -1.83
mien -0.36 -0.17 0.10 0.19 -0.51 -0.40 -0.97 -0.99
pork -0.16 0.53 -0.03 0.14 -0.03 -0.16 -0.64 0.34
beef -0.36 0.97 -0.44 0.57 0.35 -0.35 0.02 -0.85
chicken -0.31 0.52 0.22 -0.09 -0.19 0.08 0.50 0.28
duck -0.48 -0.61 0.33 -0.10 -0.07 0.09 0.58 0.60
eggs -0.20 0.44 -0.06 0.30 -0.11 -0.13 -0.12 0.50
tofu -0.85 0.88 0.00 0.10 0.18 -0.05 -0.35 0.61
spinach -0.22 0.47 -0.16 0.34 -0.11 -0.23 0.20 0.27
cabbage 0.48 0.19 -0.03 0.19 0.00 0.18 0.13 0.35
tomato -0.02 -0.43 -0.36 0.30 -0.09 0.01 0.82 0.59

11
orange 0.68 1.01 -0.41 0.61 -0.06 -0.07 -0.64 0.96
banana 0.49 -0.20 -0.09 0.28 0.03 -0.14 0.25 0.12
mango -2.49 1.55 0.47 0.27 0.37 1.63 -0.15 1.27
sauce -0.06 -0.09 -0.21 0.08 0.10 -0.03 0.14 0.10
salt -0.93 0.09 0.16 -0.05 0.10 0.27 0.10 -0.13
msg -0.36 -0.17 0.10 -0.17 0.07 0.09 -0.40 0.11
tea -0.29 0.26 0.08 0.03 -0.22 -0.19 -0.17 -0.39

The group of vegetables (spinach, cabbage and tomatoes) shows a limited degree of
substitutability. An increase in the price of spinach in Table 3b increases both the demand of
cabbage and the demand of tomatoes. Similarly, an increase in the price of either cabbage or
tomatoes increases the demand for spinach, indicating substitutability. However, an increase in
the price of tomatoes reduces the demand for cabbage just as it reduces the demand for tomatoes.
Within the group of vegetables, the relative magnitudes of the elasticities conform to expectations
in that the own price elasticity is larger than the cross-price elasticities.

Table 3b. Cross price elasticities


Percentage change given a 1% change in the price of . . .
in the demand of chicken duck eggs tofu spinach cabbage Tomato

Ordinary rice -0.16 -0.02 -0.04 0.12 -0.03 -0.04 0.00


sticky rice -0.13 0.26 0.04 -0.76 -0.46 -0.38 -0.36
corn -0.73 2.35 -2.06 2.45 0.39 -0.28 -1.05
cassava -7.54 2.34 -7.78 1.47 -0.48 -0.86 -2.17
potato 1.65 1.19 -1.11 -0.69 -0.79 -0.55 0.74
mien -0.34 0.68 -0.93 0.26 0.31 -0.30 0.21
pork 0.10 0.00 0.26 -0.02 0.06 0.06 -0.05
beef 1.24 0.21 -0.60 -0.55 0.02 0.46 0.00
chicken -0.76 -0.17 -0.19 -0.26 0.10 0.16 -0.08
duck 0.50 -1.24 -0.12 0.42 0.11 0.15 -0.16
eggs 0.60 0.16 -0.76 -0.32 0.17 0.05 -0.09
tofu 0.20 0.05 -0.32 -1.37 0.03 -0.15 -0.05
spinach 0.26 0.09 0.22 -0.21 -0.88 0.23 0.09
cabbage 0.81 -0.11 -0.70 0.00 0.04 -1.05 -0.05
tomato 0.10 -0.09 0.36 -0.08 0.23 0.03 -0.82
orange 1.51 0.04 0.21 -0.73 0.40 0.20 -0.09
banana -0.02 0.06 -0.14 -0.24 0.09 0.08 0.10
mango 1.49 0.28 -1.75 -1.14 0.02 1.24 0.02
sauce -0.13 -0.02 -0.43 0.03 -0.03 0.10 0.27
salt -0.09 0.10 -0.20 0.34 0.00 -0.09 0.03
msg 0.23 -0.26 -0.04 0.10 0.01 -0.06 0.11
tea 0.47 0.39 -0.25 -0.22 -0.06 0.01 -0.15

Consider flavor enhancers- fish sauce, salt and MSG. Note that some combinations of
these flavor enhancers demonstrate substitutability, while other combinations demonstrate
complementarity. For example, an increase in the price of fish sauce in Table 3c increases the
demand for MSG, but decreases the demand for salt, indicating that fish sauce and salt are
complements in Vietnamese household consumption but that fish sauce and MSG are substitutes

12
in Vietnamese household consumption. An increase in the price of salt has a symmetric impact
such that the demand for fish sauce decreases, indicating complementarity, and the demand for
MSG increases, indicating substitutability. Also, an increase in the price of MSG increases the
demand for salt and decreases the demand for fish sauce. As expected, the magnitude of the own
price elasticities is greater than the magnitude of the cross price elasticities. Moreover, the own
price elasticities of the flavor enhancers indicates that each is price inelastic. The data show how
households adjust the consumption of flavor enhancers in response to changes in the prices of the
flavor enhancers.

With regard to fruits (oranges, bananas and mangoes), bananas and mangoes appear to be
unambiguous substitutes. However, it’s not clear if oranges should be regarded as substitutes or
complements with bananas or mangoes. An increase in the price of oranges in Table 3c causes
decreases in the demand for each of the fruits. But increases in the price of either bananas or
mangoes increase the demand for oranges. The magnitude of the own price elasticities is again
greater than the magnitude of the cross price elasticities. However, unlike flavor enhancers, the
demand for each of the fruits is price elastic, indicating that households are more responsive to
price changes in fruits than in price changes in flavor enhancers.

Table 3c. Cross price elasticities


Percentage change Given a 1% change in the price of . . .
In the demand of orange banana mango sauce salt msg tea

ordinary rice -0.05 -0.04 -0.08 -0.06 -0.05 0.43 -0.06


sticky rice 1.01 0.07 0.05 -0.23 0.06 -0.73 -0.15
corn 0.16 -2.76 2.57 -1.20 2.05 9.05 -0.90
cassava -1.42 -2.98 1.79 -0.81 -3.47 0.01 -3.09
potato 0.31 2.79 -0.42 0.62 -0.35 -3.28 1.40
mien 0.50 -0.57 0.25 0.12 0.08 -0.18 0.89
pork 0.01 0.08 0.02 0.02 0.03 -0.58 0.06
beef -0.16 -0.05 0.61 0.51 0.25 -0.78 0.01
chicken 0.04 -0.10 0.23 -0.10 0.01 0.47 0.10
duck 0.16 -0.31 -0.03 -0.21 0.03 -2.21 0.27
eggs 0.02 -0.06 0.26 0.24 0.23 -0.38 -0.07
tofu 0.12 0.08 0.08 0.20 0.13 -0.25 0.08
spinach -0.29 -0.02 0.15 0.19 0.03 -0.65 -0.23
cabbage -0.05 -0.23 -0.02 0.13 0.26 -1.55 -0.23
tomato 0.04 -0.07 0.25 0.13 0.31 -0.69 -0.07
orange -1.01 0.19 0.48 0.40 0.23 -2.63 0.49
banana -0.08 -1.07 0.24 0.07 0.16 -0.38 0.10
mango -0.46 0.51 -1.81 0.20 0.64 -3.25 0.71
sauce -0.13 0.15 0.05 -0.90 -0.08 -0.58 0.23
salt 0.09 -0.09 -0.09 -0.16 -0.98 0.11 -0.10
msg -0.01 -0.02 -0.08 0.00 0.09 -0.64 0.12
tea 0.06 0.12 0.15 0.05 0.01 0.22 -0.63

The price elasticities in Tables 3a-3c provide quantitative measures for consumption
relationships that are demonstrated by contemporary Vietnamese households. In particular, the
measures provide a means for evaluating the quality of substitutability or complementarity, which
are relevant for both public and private agents who are interested in the markets for household
food items in Vietnam.

13
In the context of public policy, the values in Tables 3a-3c provide the data to make a first
guess at the effects of a number of food policy alternatives. For example, one might wonder
about the effects on household consumption of reduced barriers in the trade of pork with China,
which would likely reduce the price of pork to Vietnamese households. The values in Table 3a
provide an answer. For that matter, one might wonder about the effect of price reductions among
a set of foods that are produced at lower cost in China. Also, institutional reforms among State
enterprises, or the increase of competition with State enterprises, within a certain food sector, like
tea production, are known to reduce domestic prices. The effect of such an institutional shift on
household consumption patterns can be estimated with the values in Tables 3a-3c. In addition,
the development of sector-specific technology, which reduces production costs throughout the
sector, like the application of high-yielding or cost-reducing, genetically-modified crops, is
expected to reduce the market price of the commodity. Tables 3a-3c provide parameters that can
shed light on the effect of the adoption of such a technology on household consumption patterns.

Private sector agents can also take advantage of the predictive power of the calculated
elasticities. Private sector business plans should account for the effects of government policy
when those policies affect related markets of the business.

Table 4 presents the expenditure elasticity measures for the 22 food items. Note that the
expenditure elasticities are positive with only two exceptions. The values reported in Table 4
cannot be interpreted as total income elasticities because the model has supposed that households
purchase the goods in the second stage according to a budget that is less than total household
income. Therefore, the elasticities of Table 4 represent demand sensitivities given an increase in
the budget allocated for the purchase of the group of food items listed in Table 4. The most
elastic demands may or may not be luxury goods, goods which are elastic with respect to total
income. The relative magnitudes of the values in Table 4 reflect the nature of the shift in
household consumption due to increases in income.

Table 4. Calculated expenditure elasticities

Food item Expenditure


elasticity
ordinary rice 0.95
sticky rice 1.18
corn -0.08
cassava -1.37
potato 1.40
mien 0.48
pork 1.21
beef 1.61
chicken 1.12
duck 0.94
eggs 0.80
tofu 0.81
spinach 0.67
cabbage 0.69
tomato 0.74
orange 0.82
banana 0.50
Mango 1.30

14
Sauce 0.73
salt 0.54
msg 0.75
tea 0.82

As household income continues to increase in Vietnam, significant changes in household


consumption should be expected. As has been observed in a host of other developing countries,
meat demand in Vietnam will rise substantially as household income increases. Meat
consumption will increase much faster than vegetables, in particular, and faster than all the other
food items, in general. As development in Vietnam progresses, households will have more
income available for food consumption and a greater portion of the food consumption budget will
be devoted to meat consumption. Thus, meat products, in general, are expected to be the fastest
growing market among household food consumption products.

4. Summary

The LA/AIDS model, applied to recent household data, provided useful estimates for
elasticity measures that shed light on potential shifts in contemporary household consumption
patterns. The price elasticity measures provide a means of evaluating the ways in which
contemporary Vietnamese households substitute for other goods as relative prices change. In the
context of public policy, such measures permit reasonable predictions about the effects of a wide
range of policies, including trade liberalization, domestic institutional reforms and adoption of
cost-reducing technology.

Private sector agents can also utilize the calculated elasticities to estimate changes in the
market for household food products given the general trend in economic development that is
expected to continue. Recent data from Vietnam confirms the observed trend throughout the
developing world that the demand for meat products increases at a faster pace than other food
products, resulting in an increasing share of meat in the household’s budget. In addition, because
private sector agents can be affected by public policy for specific food sectors, the price
elasticities can be useful for private sector agents to assess the impact of a sector-specific policy
on the agent’s market.

References

Cozzarin, B. P., and Gilmour, B. W. 1998. A methodological evaluation of


empirical demand systems research. Canadian Journal of Agricultural Economics
46(3):297-316.
Deaton, A. 1997. Analysis of household surveys: a microeconometric approach to
development policy. Johns Hopkins University Press, Baltimore. 479 pp.
Deaton, A., and Muellbauer, J. 1980a. Economics and Consumer Behavior.
Cambridge University Press, New York. 450 pp.
Deaton, A., and Muellbauer, J. 1980b. An almost ideal demand system. American
Economic Review 70(3):312-326.
Green, R. D., and Alston, J. M. 1990. Elasticities in AIDS models. American
Journal of Agricultural Economics 72(2):442-445.

15
Heckman, J. J. 1979. Sample Selection Bias as a Specification Error.
Econometrica 47(1):153-161.
Heien, D., and Wessells, C. R. 1990. Demand systems estimation with microdata:
a censored regression approach. Journal of Business and Economic Statistics 8(3):365-
371.
Litvack, J. I., and Rondinelli, D. A. 1999. Market reform in Vietnam: building
institutions for development. Quorum, Westport, Conn. 203 pp.
Minot, N., and Goletti, F. 2000. Rice market liberalization and poverty in Viet
Nam. IFPRI, Washington, D.C.. 128 pp.
Rae, A. N. 1998. The effects of expenditure growth and urbanisation on food
consumption in East Asia: a note on animal products. Agricultural Economics 18(3):291-
299.
Saha, A., Capps, O., and Byrne, P. J. 1997. Calculating marginal effects in models
for zero expenditures in household budgets using a Heckman-type correction. Applied
Economics 29(10):1311-1316.
Teklu, T., and Johnson, S. R. 1988. Demand systems from cross-section data: an
application to Indonesia. Canadian Journal of Agricultural Economics 36(1):83-101.
Viet Nam. 1998. Vietnam living standards survey 1997-1998. General Statistical
Office, Ha Noi.
Zellner, A. 1962. An efficient method of estimating seemingly unrelated
regressions and tests of aggregation bias. Journal of the American Statistical Association
57(298):348-368.

16

Vous aimerez peut-être aussi