LSM - 82 JULY 1991 Living Standards Measurement Study Working Paper No S2 Does Undernutrition Respond to Incomes and Prices? Dominance Tests for Indonesia Martin Ravallion LSMS Working Papers No. 11 Three Essays on a Sri Lanka Household Survey No. 12 The ECIEL Study of Household Income and Consumption in Urban Latin America: An Analytical History No. 13 Nutrition and Health Status Indicators: Suggestions for Surveys of the Standard of Living in Developing Countries No. 14 Child Schooling and the Measurement of Living Standards No. 15 Measuring Health as a Component of Living Standards No. 16 Procedures for Collecting and Analyzing Mortality Data in LSMS No. 17 The Labor Market and Social Accounting: A Framework of Data Presentation No. 18 Time Use Data and the Living Standards Measurement Study No. 19 The Conceptual Basis of Measures of Household Welfare and Their Implied Survey Data Requirements No. 20 Statistical Experimentation for Household Surveys: Two Case Studies of Hong Kong No. 21 The Collection of Price Data for the Measurement of Living Standards No. 22 Household Expenditure Surveys: Some Methodological Issues No. 23 Collecting Panel Data in Developing Countries: Does It Make Sense? No. 24 Measuring and Analyzing Levels of Living in Developing Countries: An Annotated Questionnaire No. 25 The Demand for Urban Housing in the Ivory Coast No. 26 The C6te d'Ivoire Living Standards Survey: Design and Implementation No. 27 The Role of Employment and Earnings in Analyzing Levels of Living: A General Methodology with Applications to Malaysia and Thailand No. 28 Analysis of Household Expenditures No. 29 The Distribution of Welfare in COte d'Ivoire in 1985 No. 30 Quality, Quantity, and Spatial Variation of Price: Estimating Price Elasticities from Cross-Sectional Data No. 31 Financing the Health Sector in Peru No.32 Informal Sector, Labor Markets, and Returns to Education in Peru No. 33 Wage Determinants in COte d'Ivoire No. 34 Guidelines for Adapting the LSMS Living Standards Questionnaires to Local Conditions No.35 The Demand for Medical Care in Developing Countries: Quantity Rationing in Rural COte d'Ivoire No. 36 Labor Market Activity in C6te d'Ivoire and Peru No. 37 Health Care Financing and the Demand for Medical Care No. 38 Wage Determinants and School Attainment among Men in Peru No. 39 The Allocation of Goods within the Household: Adults, Children, and Gender No. 40 The Effects of Household and Community Characteristics on the Nutrition of Preschool Children: Evidence from Rural C6te d'Ivoire No. 41 Public-Private Sector Wage Differentials in Peru, 1985-86 No. 42 The Distribution of Welfare in Peru in 1985-86 No. 43 Profits from Self-Employment: A Case Study of C6te d'Ivoire No. 44 The Living Standards Survey and Price Policy Reform: A Study of Cocoa and Coffee Production in C6te d'Ivoire No. 45 Measuring the Willingness to Pay for Social Services in Developing Countries No. 46 Nonagricultural Family Enterprises in C6te d'Ivoire: A Descriptive Analysis (List continues on the inside back cover) Does Undernutrition Respond to Incomes and Prices? Dominance Tests for Indonesia The Living Standards Measurement Study The Living Standards Measurement Study (Lmss) was established by the World Bank in 1980 to explore ways of improving the type and quality of house- hold data collected by statistical offices in developing countries. Its goal is to foster increased use of household data as a basis for policy decisionmaking. Specifically, the LIS is working to develop new methods to monitor progress in raising levels of living, to identify the consequences for households of past and proposed gov- ernment policies, and to improve communications between survey statisticians, an- alysts, and policymakers. The LSMS Working Paper series was started to disseminate intermediate prod- ucts from the LSs. Publications in the series include critical surveys covering dif- ferent aspects of the Lss data collection program and reports on improved methodologies for using Living Standards Survey (Lss) data. More recent publica- tions recommend specific survey, questionnaire, and data processing designs, and demonstrate the breadth of policy analysis that can be carried out using Ls data. LSMS Working Paper Number 82 Does Undernutrition Respond to Incomes and Prices? Dominance Tests for Indonesia Martin Ravallion The World Bank Washington, D.C. Copyright 0 1991 The International Bank for Reconstruction and Development/IE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing July 1991 To present the results of the Living Standards Measurement Study with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. Any maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning the delimitation of its boundaries or its national affiliation. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to Director, Publications Department, at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to photocopy portions for classroom use is not required, though notification of such use having been made will be appreciated. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Publications Sales Unit, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, USA., or from Publications, The World Bank, 66, avenue d16na, 75116 Paris, France. ISSN: 0253-4517 Martin Ravallion is a senior economist in the Welfare and Human Resources Division, Population and Human Resources Department of the World Bank. Library of Congress Cataloging-in-Publication Data Ravallion, Martin. Does undernutrition respond to incomes and prices?: dominance tests for Indonesia / Martin Ravallion. p. cm. - (ISMS working paper, ISSN 0253-4517; no. 82) Includes bibliographical references (p.). ISBN 0-8213-1883-7 1. Malnutrition-Indonesia-Econometric models. I. Title. IL Series. RA645.N87R38 1991 363.8-dc2O 91-14825 CIP V ABSTRACT Recent evidence suggests that food energy intakes are less responsive to incomes of the poor than was once thought. However, it is not intakes per se that we are concerned about, but undernutrition. Two facts have long confounded assessments of impacts on undernutrition: individual nutrient requirements vary in a generally unobserved way, and intakes are observed with error. By modelling observed intake distributions econometrically, straightforward stochastic dominance tests can permit robust qualitative inferences. An application to Indonesia in the mid-1980s indicates that regional differences in energy intake distributions are influenced by average income levels, intra-regional inequalities, and local prices of staple food- grains, all with unambiguous effects on undernutrition. The results suggest that any adverse effects on inequality of a growth process would need to be large to outweigh the desirable effect on undernutrition. Plausible effects on rural incomes are insufficient to outweigh adverse effects on undernutrition of higher rice prices. vi ACXNOWLEDGEMENTS The author is grateful to Monika Huppi for her help in setting up the data set used in the paper from the 1984 and 1987 data tapes of Indonesia's national socio-economic survey, to Indonesia's Central Bureau of Statistics for providing those tapes, and to Alok Bhargava for his help with the calculations needed to implement the Bhargava-Sargan dynamic random effects etimator used. Helpful comments on the paper were made by Shubham Chaudhuri, The World Bank Economic Review referees, and by seminar participants from the Department of Economics, Cornell University, and the Eastern Economics Association Meeting, Pittsburgh. vii TABLES OF CONTENTS 1. Introduction.................... ... . . . . . 1 2. Approaches to Assessing Changes in Undernutrition....... . . . . 5 3. An Implementation for Indonesia........ ..... . . . . . 11 4. Further Implications of the Empirical Results... . . . . . . . . 21 5. Conclusions................... .... . . . . . 28 Notes..................... ..... . . . . . 31 References..................... .... . . . 33 LIST OF TABLES AND FIGURES Table 1 Average Percentage Changes in the Proportion below Selected Calorie Cut-Off Points for 52 Regions of Indonesia, 1984-1987 15 Table 2 Income and Expenditure Elasticities of Energy Intake Frequencies by Region of Indonesia, 1984-1987.... . . . . . 15 Table 3 Econometric Models of the Determinants of Energy Intake Distribution Across Regions of Indonesia, 1987.... . . . . . 19 Table 4 Estimated Elasticities of Cumulative Frequencies of Energy Intake at Selected Points, Indonesia, 1987...... . . . . . 22 Table 5 Displacements of Cumulative Frequencies of Energy Intakes Due to Changes in Incomes and Prices, Indonesia, 1987 . . . . . . 23 Table 6 Compensating Equivalent Changes in Inequality for Various Income Growth Rates, Indonesia 1987...... .... . . . 25 Figure 1 Distribution of Caloric Intake Growth, Indonesia 1984-87 . . . 13 Figure 2 Reductions in Caloric Undernutriton and Income Growth, Indonesia, 1984-87.......... ..... . . . . . . . 16 Figure 3 Income Elasticity of the Distribution of Caloric Intake . . . . 17 Figure 4 Effects of Income and Prices Changes on Intake Distribution............... .... . . . . . 24 1. itouto The attainment of adequate nutrition is an important criterion for evaluating the success of development policies. However, such evaluations have often been hampered by the fact that the measurement of undernutrition is fraught with both conceptual and technical problems. Most importantly, the nutritional requirements needed for good health vary across individuals and over time in generally unknown ways, and their intakes are typically also measured with error. Can we still say anything useful about how the instruments of development policies influence the extent of undernutrition? Development policies have often emphasized the role of various economic variables, particularly incomes and prices, in reducing undernutrition. Recent household level studies have thrown light on the effectiveness of these instruments in improving nutrition levels. There is mounting empirical evidence suggesting that the income elasticity of individual energy intake in developing countries is lower than had been thought 10 or so years ago. (For recent surveys see Behrman 1988, Behrman et al., 1988, Bouis and Haddad 1988, and Alderman 1989). There are reasons why the methodologies of the earlier studies would have led to some over- estimation of that elasticity, particularly associated with the level of aqqrecation across persons and goods in the early studies.' And the downward revision that seems to be called for is far from negligible; whereas 10 years ago an income elasticity of energy intake for the poor of about 1.0 would probably not have been seriously questioned, elasticities a good deal less that 0.5 would be considered more in keeping with recent estimates. (See, for example, Behrman and Deolalikar 1987, Bouis and Haddad 1988, Ravallion 1990, Strauss and Thomas 1989, Bhargava 1991). 2 This substantial downward revision in our priors about the income response of nutrition intake has some potentially profound implications for development policies. The fight against hunger has been one of the strongest motivations for development, and raising incomes of the poor (through both the. growth process and policies aimed at reducing inequalities of income) has long been seen as the main weapon in that fight. The recent evidence has led some observers to suggest that this weapon may well be quite blunt, or, indeed, virtually useless. Does this aspect of our approach to development policy need a major revision? The recent empirical evidence on determinants of individual nutrition has also led some to question the role of incomes and prices in the causation of famines, a role which has been stressed in recent literature on famines (Sen, 1981; Ravallion, 1987). Do we also need to rethink this approach to understanding transient food insecurity? This paper takes a further look at the question of whether aggregate undernutrition is responsive to incomes and prices. The point of departure for this study is the realization that it is not impacts on nutrient intakes per se that one cares about in this context, but it is the impacts on the adequacy of those intakes relative to needs which matters. While this is obvious at a conceptual level, it presents difficult measurement problems if the insight is to be put into practice. The methodology proposed here is potentially far more robust to the inevitable errors and unknowns in measuring attainment relative to needs. The approach starts with an econometric model of how nutrient intake distributions vary across regions or sectors of the economy, and then uses theoretical results on stochastic dominance to infer the effects of changes in 3 incomes and prices on undernutrition. The usefulness of stochastic dominance theory for ranking distributions in terms of some objective function has been known for over 20 years, though the relevance to poverty and nutrition analysis has only come to be appreciated quite recently.2 In assessing impacts on undernutrition, the dominance approach has the advantages over past methods that it uses all of the information available on the distribution of nutrient intakes, and it places far fewer ad hoc restrictions on the unknown distribution of individual nutrient requirements. It can also allow inferences which are more robust to intake measurement errors, and to arbitrary choices about the specific measure of undernutrition. The theoretical results needed for this approach are outlined in Section 2. The methodological contribution of the present paper lies in combining the dominance approach with an econometric model of intake distributions, thus permitting dominance tests of the comparative static effects on undernutrition of changes in the explanatory variables of the intake model. The methodology is used to explore further the results of Ravallion and Huppi (1991) concerning Indonesia's progress in reducing undernutrition during the mid-1980s. The regional dimensions of that progress are examined, and some explanations are offered. New empirical evidence for Indonesia is presented in Section 3, implementing the dominance approach. The most popular question from past work is re-examined, namely that of what effects (if any) on aggregate undernutrition can be expected from economic growth and contraction. The role of food-staple prices is also examined, with and without associated rural income effects. The present approach also allows us to explore quite directly other questions of interest concerning the distribution of the benefits of growth: 4 How responsive is undernutrition to changes in income inequality? Will a growth process which is associated with an increase in inequality, lead to a reduction in undernutrition? If equalizing income redistributions (such as through targeted transfers) entail a "growth trade-off" (such as through adverse effects on aggregate savings), how severe would that trade-off have to be to wipe-out the effect of greater equity on undernutrition? These questions are addressed in Section 3. 5 2. Approaches to Assessing Changes in Undernutrition Recent approaches have been based on econometric estimates from household level data of how energy intakes are related to incomes, prices and other variables.3 This does not, however, directly address the issue of primary interest, namely that of the response of undernutrition to changes in incomes and prices. It is not intakes per se that matter, rather it is the adequacy of those intakes relative to individual needs. This raises a number of difficult further issues concerning the assessment of individual nutritional well-being and aggregation across individuals. The biggest single problem is that we typically do not know the nutrient requirements that surveyed individuals need to maintain good health, allowing normal physiological functions without symptoms of deficiency. Requirements may vary widely, reflecting differences in the human body's metabolic rate at rest and differences in activity levels. There has been a great deal of controversy concerning the implications of variability in requirements for estimating the extent of undernutrition (for good recent surveys of the issues see Dasgupta and Ray, 1990, and Osmani, 1987). Two sources of variability can be distinguished: inter-personal variability (genotypic variations around the assessed requirements of some reference person) and inter-temporal variability for a given person, which has been interpreted as the outcome of physiological regulatory mechanisms influencing energy utilization in the human body (as in, for example, Sukhatme, 1978, and Srinivasan, 1981).4 However, for the purpose of the present discussion, we need only imagine that there exists some distribution of person-specific requirements at the survey date. This could reflect either source of variability. 6 Uncertainty about requirements is not the only problem. Nutrient intakes (though more readily observed than requirements) are typically measured with error. In effect, this will also mean that we are generally using the wrong individual requirements (which should ideally be adjusted for the intake measurement error). This is another reason for treating requirements as a random variable. By making explicit assumptions about the inter-personal distribution of nutrient requirements, measures of undernutrition can still be readily constructed from household or (preferably) individual survey data. Also, with an appropriate econometric model of intake determination, income and price effects can then be simulated (Ravallion, 1990). But there is still an uncomfortable arbitrariness in the assumptions made about requirements. And those assumptions can readily matter to the conclusions drawn. For example, it is not difficult to imagine the existence of an underlying distribution of requirements such that even a purely random change in food energy intakes (uncorrelated with intake levels) could have a substantial impact on aggregate energy undernutrition. Similarly, while there may be little or no sign of a correlation between intakes and incomes, there can exist requirement distributions such that both the incidence and severity of undernutrition are highly correlated with poverty. Fortunately, for many purposes, one is mainly interested in the qualitative effects of changes in incomes and prices on undernutrition. Is some policy combination, interpretable as a set of price and income changes, moving us in the right direction? Then there is an alternative approach which places far fewer restrictions on the unknown distribution of individual requirements. In fact each person may have a different requirement, and the distribution need not have any recognizable form (such as the commonly assumed 7 normal distribution). The still popular assumption of a single cut-off point - a degenerate distribution - is an extreme limiting case. However, while this approach allows a far more general class of possible distributions of requirements, it still imposes two potentially important assumptions about how requirements are distributed. And, like all assumptions, neither of these need hold in reality. First, it is assumed that intakes and requirements are independently distributed. This is a common assumption, following Sukhartme (1961). On a priori grounds, one would suspect that, because of the existence of common factors influencing both variables (such as age, weight, and activity levels), they would tend to be positively correlated. Nonetheless, there is some evidence suggesting that there is little or no correlation between intakes and requirements (Sukhartme, 1961). And Kakwani (1989) has found that estimates of the proportion of the population of India deemed to be undernourished are quite insensitive to the assumption one makes about the correlation between intakes and requirements. Second, it is assumed that the distribution of requirements does not change (or, in the specific cases of interest here, independent of incomes and prices). Whether that holds in reality will depend on how the changes in incomes and prices come about. For example, if intakes improve as a result of greater work effort leading to higher incomes then undernutrition need not improve. There appears to be little more that can be said in general, though the qualification should be kept in mind when interpreting this paper's empirical results. Under these assumptions, one can invoke well-known theoretical results on stochastic dominance to attain at least a partial ordering of intake distributions in terms of any well behaved measures of undernutrition. To see how, suppose the intake distribution shifts from that of state A to 8 state B, as a result of (say) changes in income distribution. For example, state B may be reached after economic growth in state A. If one finds that intakes increase for all individuals in the move from A to B, and there is no change in any individual's requirement, then undernutrition must fall. But that is an unnecessarily strong test. A more useful test can be constructed as follows. Let Fj(z) denote the proportion of the population who fails to reach a given intake level z in the state i=A,B. In the vocabulary of poverty measurement, Fi(z) is the "headcount index" of undernutrition when a single requirement cut-off point is set at z. As one allows z to vary over its entire range, F,(z) traces out the cumulative distribution function of intakes. If PA(s) is below FB(z) at all intake levels z (or, more precisely, FA(M) is nowhere above FB(z), and at least somewhere below), then the proportion of the population who are undernourished will be lower in A than B. This is called the first-order dominance test for comparing undernutrition or poverty in two states. First-order dominance can be an extremely useful test for determining whether there is more poverty in one state than another for any unknown but fixed poverty line (Atkinson, 1987; Foster and Shorrocks, 1988). Furthermore, the underlying theoretical result which supports this test can be readily generalized to accommodate any fixed distribution of poverty lines, or (as in the present application) nutrient requirements. It is easy to see why this is so once one notes that our best estimate of the proportion of the population who are undernourished for any distribution of requirements is simply the expected value of Fj(z), where the expectation is taken over that distribution of requirements. If two intake distributions have the same requirement distribution and first-order dominance holds then clearly the expected value 9 of PA(s) must exceed FB(z) when both expectations are evaluated over the distribution of requirements. Undernutrition is higher in state A than B. First-order dominance of one intake distribution over another also implies an unambiguous ranking of the two distributions in terms of a broader class of undernutrition indicators than the simple headcount index. Let the level of undernutrition of a person with intake x and requirement z be u(x,z) which is positive for x