Is rice becoming an inferior good? Food demand in the Philippines

-nd~~~~~~ External As Za Policy, Research, and External Affairs WORKING PAPERS International Trade International Economics Department The World Bank July 1991 WPS 722 Is Rice Becoming an Inferior Good? Food Demand in the Philippines Merlinda D. Ingco Improving pricing and trade policy for cereals, feed, and meat in the Philippines may be more cost-effective in improving the balance of food supply and demand than further investments in rice irrigation - particularly if wheat consumption increasingly substitutes for rice consumption. The Policy. Research, and External Affairs Complex distributes PRI Working Papers to dissemunate the findings of work in progress and to encourage the cxchange of ideas among Bank staff and aU othcrs interested in development issues. 'hese papers carry the names of the authors, reflect only their views, and should be used and cited accordingly. The findings, interpretations, and conclusions arc the authors' own. They should not be attributed to the World Banki, its Board of Directors, its managemcnt, or any of its member countries.  Policy, llesearch, and External Affairs l. k ,bzr11 1 Itrational Trade WPS 722 This paper- a product of the International Trade Division, International Economics Department- is part of a larger effort in PRE to understand the changes in food markets in developing countries. Copies are available free from the World Bank, 1818 H Street NW, Washington DC 20433. Please contact Pauline Kokila, room S7-040, extension 33716 (25 pages, with tables). Using time-series data, Ingco estimates a de- * Rice, corn, wheat, au. A meat are net substi- mand system model for food - including rice, tutes. Rice, fish, and fruits and vegetables are corn, wheat, meat, fish, and fruits and vegetables net complements. Wheat is a net substitute for - for the Philippines. She finds that: rice, com, fish, and fruits and vegetables - but a net complement to meat (partly because of Food demand is responsive to reladve price Ltirbanization and the proliferation of fast-food changes. Most of the other food products are outlets in recent years). parLicularly responsive to changes in rice prices, an important variable in agricultural policy in the * Urbanization increases the consumption of Philippines. A marked change in rice prices wheat, fish, and fruits and vegetables - and relative to other food prices would have impor- slightly decreases the conswnption of rice. tant policy implications because of its relatively large share in food budgets and the relatively * Consumption of rice and wheat can be great response of other foods to changes in rice expected to grow. Per capita consumption of prices. corn should decline slightly. * In particular, as wheat prices decline, wheat * These trends should be considered in consumption should increase - resulting in evaluating the costs and benefits of further some substitution away from rice - because irrigation investments in the rice sector. An wheat and rice are net substitutes. improved pricing and trade policy in the cereals, feed, and livestock-meat sectors may be more * The demand for wheat, meat, and fruits and cost-effective in improving the balance of food vegetables is more responsive to own-price supply and demand than more investments in changes than are the staple foodstuffs. rice irrigation. The PRE Working Paper Series disseminates the findings of work under way in the Bank's Policy, Research, and Extemal AffairsComplex. Anobjectiveofthcseries is to getthese findingsout quickly. even if presentations are lessthanfully polished. The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the PRE Dissemination Center  IS RICE BECOMING AN INFERIOR GOOD? FOOD DEMAND IN THE PHILIPPINES Table of Contents I. Introduction 1 II. Food Demand System for the Philippines 3 H.1 Demand Model 3 II.2 Data Description 4 11.3 Estimation Procedure 4 III. Empirical Results 5 III.1 Food Demand System 5 III.2 Testing of Restrictions 7 iII.3 Price and Income Elasticities 9 IIIA Comparisons 14 IV. Conclusions and Policy Implications 15 References 25 Tables in the Text Table 1. Parameter Estimates for the LA/AIDS Model. 6 Table 2. Test Statistics for Homogeneity and Symmetry 8 Table 3. Income Elasticities of Demand, 1965-90. 11 Table 4. Marshallian Elasticity Matrix. 13 Table 5. Hicksian Elasticity Matrix. 13 Table 6. Income Elasticities from Other Studies. 14  Appendix Tables Table Al. Income Elasticities of Demand, 1965-1990. 18 Table A2. Rice Demand: Marshalian Own and Cross-Price Elasticities. 19 Table A3. Corn Demand: Marshallian Own and Cross-Price Elasticities. 20 Table A4. Wheat Demand: Marshallian Own and Cross-Price Elasticities. 21 Table AS. Meat Demand: Marshalilan Own and Cross-Price Elasticities. 22 Table A6. Fish Demand: Marshallian Own and Cross-Price Elasticities. Table A7. Fruits & Vegetables Demand: Marshallian Own and Cross-Price Elasticities, 1965-1988. 23  IS RICE BECOMING AN INFERIOR GOOD? FOOD DEMAND IN THE PHILIPPINES I. Introduction 1.1 What are the prospects for demand for the main foodstuffs, particularly rice, in the Philippines? Countries which have traditionally consumed rice as the basic staple such as Japan, the Republic of Korea, and Taiwan are eating more wheat and wheat products. There is also a shift towards increased consumption of meats, dairy products, vegetable oils, and fruits and vegetables. A recent study found rice to be an inferior good in Japan, Taiwan, Malaysia, Singapore, Thailand, and Nepal (Ito, et al 1989). Estimates of income elasticities for rice give conflicting results, but the study by Ito has caused a controversy. The income elasticity estimates by Ito et al based on time-series analysis are very low for 14 Asian countries and show significant decline from 1961 to 1984. Based on these results, Ito et al argue that there may be potential for an excess supply of rice in Asia. 1.2 Bouis (1989) has challenged these findings and argued that changing economic and production structures, such as the increasing commercializatioi' of production, and not changing income may have resulted in declining rice consumption. Bouis argues that failure to account for the effects on demand for staples of decisions by semi-subsistence farmers to produce staples will lead to a downward bias in income elasticities. 1.3 Huang (1990) has also argued that Ito et al's results are generally implausible and noted statistical problems which may have led to underestimation of the income elasticities. Huang's study of demand for cereals in Asia using time-series data shows that for Asia as a whole, total cereal grains are a normal good with an income elasticity of 0.32. Among the nine countries studied, rice is found to be a normal good except in Japan and Thailand. The rice income elasticity estimates are as follows': India (0.527), Pakistan (0.486), Indonesia (0.471), Republic of Korea (0.456), China (0.427), Bangladesh (0.379), Taiwan (0.258), Philippines (0.247), Japan (-0.214), and Thailand (-0.136). 1.4 However, the manner in which food demand parameters have been estimated for most developing countries makes them of limited use for policy analysis for the following reasons. First, most of these estimates are based on ad-hoc demand models, and hence lack consistency with consumer demand theory making comparability difficult. Second, most studies estimate only the demand parameters for staple cereals. No systematic linkages to demand of other foods, such as meat, fish, and other major food items in the diet have been estimated. 1.5 Changing goverment policies is usually a delicate task and therefore it is desirable for policy makers to anticipate correctly responses to any changes. In most developing countries, the part of the populations most affected by agricultural policy changes are usually near subsistence and, moreover, many consumers of food are also producers. An important set of parameters that determines the effects of policy changes is the matrix of consumer demand elasticities. Consistent measures of income, own-price, and cross-price elasticities for major food item are necessary, therefore, to evaluate adequately the effects of changes in agricultural and incomes policies. I Hunng's. udy asumed week sparabiy of th cera goup and pied twoap budgein uang de inea Ahnost Idead Demud Synstm (LA/ADS) by Deao nd MesUbar (1980). Then lelsdits we product of nomme elstcit of oa emas deived In thdo fstg and tde Incom eadeity of rlic with opedt to cereal expafditur In te aeond.t.  Pap 2 1.6 In the case of the Philippines, previous studies of food demand using the systems approach have assumed separability of the group2 (e.g. cereals). For instance, a cereal group defined to include rice, wheat, and maize or a meat group consisting of beef, poultry, and pork. In the case of the cereal group, the separability assumption implies that the demand for cereals is independent of the demand for other food in the diet and other goods outside the cereal group. In other words, the prices of other foods are assumed to be either (a) uncorrelated with cereal prices and cereal expenditure or (b) irrelevant in the cereal demand equation. However, while it simplifies empirical analysis, this assumption has important implications for the elasticity estimates, economic interpretation, predictive performance of demand equations and hypothesis testing. 1.7 Whether or not it is appropriate to assume separability is an empirical question. If the separability assumption is rejected, then the elasticities derived from the system estimated under this assumption will be biased. Intuitively, the marginal rate of substitution between rice and corn is not independent of the level of other major foods in the diet, such as meat or fish and vice-versa. Hence, the true structure of demand may violate the assumption of strong or even weak separability. 1.8 In this paper, the demand for cereals in the Philippines is analyzed. Instead of assuming separability, an alternative specification of the linear Almost Ideal Demand System (LA/AIDS) which includes rice, wheat, maize, meat, fish, fruits and vegetables, other foods, and non-food commodities for the Philippines is estimated using time-series data from 1961 to 1988.3 The effects of urbanization and dynamic factors such as habit formation in consumption are also considered in the empirical analysis. This study complements Bouis's (1989) study which includes the same set of commodities but uses cross-section data from the 1978 Philippine nationwide nutrition survey. The effects of semi-subsistence production of cereals on consumption as described in Bouis's analysis of household cross-section data is also considered. 1.9 To test separability, non-nested hypothesis testing procedures are used. Alternative specifications of the demand system are evaluated based on goodness- of-fit, predictive performance, and bias in elasticity estimates. Since elasticities may not be constant over time, estimates over the period of the study are presented, in addition to the estimates based on the means of the observations. 1.10 The outline of the paper is as follows: the demand system is described in the next section. The various specifications are estimated using aggregated annual time-series and the tests are implemented. Then using the estimated parameters, the demand and income elasticities are estimated over the sample period. The parameters are used to generate baseline projections of cereal demand to 2000. Some policy implications and concluding remarks are given in the final section. 2 Data lmiato often prevent explicMiy including all prices In a demand model, so the asumpdon of weak serablity Is used to reduce the number of prices which must be Included In empirical analysis. 3 Changes In food expenditurs by ioam group or by regio are not accounted for In this study, given the aggregated dme- seri data. The effects of Income levels and other demoographic variables on consumption are analyzed In the cross-section study by BouDs (1989). which used the 1978 Nutrition Survey. For the purpose of evaluating aggrgate demand pospects, aggregated time- series analysis b adequate. his study also adms to lnvestigate the pattuns In income and demand elasticitis over time, thus r8quirIng dth use of time-sries data.  II. Food Demand System for the Philiroines 11.1 Demand Model 2.1 Among the empirical demand systems applied in the literature, the Almost Ideal DemAnd System (AIDS) (Deaton and Muellbauer, 1980) has provided the most robust estimates. This model combines the best of the theoretical features of both the translog and Rotterdam models. Food consumption behavior in the Philippines is analyzed by estimating a complete food demand system using a linear version of the AIDS model. The resulting p2rameter estimates are used to derive expenditure or income elasticities, own-price, and cross-price elasticities. The theoretical restrictions of adding-up, homogeneity, and Slutsky symmetry are imposed and tested. 2.2 An estimable variant of the Almost Ideal Demand System with the addition of dynamic factors can be specified as wit - + Eil ln Ps, + Pi ln (Yt/Pt) + 8$ ln Z, + Uit (1) i,j - 1,.., n commodities where wit is the average budget share of the ith commodity in time t, Pit is the jth commodity price at time t, Yt is per capita expenditure, Pt is an aggregate price index, and Z represents dynamic factors. a, r, P, and 8 are parameters to be estimated. The aggregate price (Pt), used to normalize nominal per capita expenditure (Yj), is defined as, ln P, - 4, +