An empirical study of food demand in Vietnam.

I. Introduction

Food consumption has been a subject of research all over the world. It is especially meaningful in developing countries where food expenditures account for a relatively large share of household income. Studies of food consumption shed light on food-related nutritional policies. They provide estimates of how food consumption is affected by changes in prices, income, and taxation policies (Dunne and Edkins 2005). Food consumption in Vietnam has been an important issue, not only because it is related to poverty, food security, but also because it is highly correlated with living standards and household resources. Vietnam has undergone a shift in the population's diet structure with a decrease in carbohydrates and rice consumption and an increase in protein sources, fruit and vegetables (FAO 1999). The demand for non-food goods and services grows faster than the demand for food. The demand for vegetables, fruit, and meat rises more quickly than the demand for grains and tubers. So studying food consumption can give a better understanding how the demand for food responds to shocks in prices, income, and policies. Given the importance of this issue, there has been surprisingly little research on the demand for food in Vietnam. In a recent contribution, Thang and Popkin (2004) study the patterns of food consumption in Vietnam and its effects on socioeconomic groups, while Hop et al. (2003) focuses particularly on the trends in food productions for the last twenty years. Figuie (2003) analyses vegetable consumption behaviour in Vietnam, while FAO (1999) considers the relationship between food consumption and nutrition.

This paper uses the Almost Ideal Demand System (AIDS) and the extended AIDS models to investigate food consumptions in Vietnam by using the iterative seemingly unrelated regressions (SUR) and the Ordinary Least Square (OLS) methods. In particular, AIDS models are estimated to calculate income and price elasticities for three different categories of food: rice food, non-rice food, and meat/fish.

The AIDS models have been applied to estimate demands for food and other categories of consumption in different countries. A survey can be found in Dunne and Edkins (2005), Karagiannis et al. (2000), LaFrance and Beatty (2001), Ederton et al. (1996). Some studies focus on specific products. For example, Gibson (1995) focuses on urban demand for food, beverages, betel nut and tobacco; Caswell (1995) studies demand for beef; Eales and Unnevehr (1988) analyzes beef and chicken products.

The rest of this paper is organized as follows: Section II presents model specifications and estimation methods. These models will be used to estimate coefficients of the AIDS model and calculate price and income elasticities of demand for each category of food consumed in Vietnam. Section III provides some general characteristics of food consumption at household level in Vietnam. It also describes the survey for this analysis. In section IV, we present the results and interpretations. Section V gives some discussion and concludes the paper.

II. Model Specification

Estimation of demand for food and services has attracted the attention of economists. Very dense literature is available. Linear Expenditure System (LES) of Stone (1954) is the seminal work of this literature. Some limitations of LES such as proportional income and price elasticities, and no complementary relationship among goods opened doors to the development of other models. Rotterdam model (Theil 1965) and Trandlog model (Christensen et al. 1975) are more flexible models. More recently, Deaton and Muellbauer (1980) have proposed an alternative model which has been named as Almost Ideal Demand System (AIDS).