Malnutrition and Poverty Magnitude and Policy Options Swomo 19maka5ir~~~~~~~~~~~~~~~~~~ $k1tt~~~~ RUtXtzmaR World Bank Staff Occasional Papers No. 1. Herman G. van der Tak, The Economic Choice between Hy- droelectric and Thermal Power Developments. No. 2. Jan de Weille, Quantification of Road User Savings. No. 3. Barend A. de Vries, The Export Experience of Developing Countries (out of print). No. 4. Hans A. Adler, Sector and Project Planning in Transportation. No. 5. A. A. Walters, The Economics of Road User Charges. No. 6. Benjamin B. King, Notes on the Mechanics of Growth and Debt. No. 7. Herman G. van der Tak and Jan de Weille, Reappraisal of a Road Project in Iran. No. 8. Jack Baranson, Automotive Industries in Developing Coun- tries. No. 9. Ayhan Cilingiroglu, Manufacture of Heavy Electrical Equip- ment in Developing Countries. No. 10. Shlomo Reutlinger, Techniques for Project Appraisal under Uncertainty. No. 11. Louis Y. Pouliquen, Risk Analysis in Project Appraisal. No. 12. George C. Zaidan, The Costs and Benefits of Family Planning Programs. No. 13. Herman G. van der Tak and Anandarup Ray, The Economic Benefits of Road Transport Projects. No. 14. Hans Heinrich Thias and Martin Carnoy, Cost-Benefit Analy- sis in Education: A Case Study of Kenya. No. 15. Anthony Churchill, Road User Charges in Central America. No. 16. Deepak Lal, Methods of Project Analysis: A Review. No. 17. Kenji Takeuchi, Tropical Hardwood Trade in the Asia-Pacific Region. No. 18. Jean-Pierre Jallade, Public Expenditures on Education and In- come Distribution in Colombia. No. 19. Enzo R. Grilli, The Future for Hard Fibers and Competition from Synthetics. No. 20. Alvin C. Egbert and Hyung M. Kim, A Development Model for the Agricultural Sector of Portugal. No. 21. Manuel Zymelman, The Economic Evaluation of Vocational Training Programs. No. 22. Shamsher Singh and others, Coffee, Tea, and Cocoa: Market Prospects and Development Lending. No. 23. Shlomo Reutlinger and Marcelo Selowsky, Malnutrition and Poverty: Magnitude and Policy Options. - , . ; -' f L ' WORLD BANK STAFF OCCASIONAL PAPERS NUMBER TWENTY-THREE The views and interpretations in this paper are those of the authors and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. Shlomo Reutlinger Marcelo Selowsky Malnutrition and Poverty Magnitude and Policy Options PUBLISHED FOR THE WORLD BANK The Johns Hopkins University Press Baltimore and London COPYRIGHT © 1976 THE INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT / THE WORLD BANK ALL RIGHTS RESERVED Library of Congress Cataloging in Publication Data Reutlinger, Shlomo. Malnutrition and poverty. (World Bank staff occasional papers; no. 23) Includes bibliographical references. 1. Nutrition policy. 2. Underdeveloped areas- Nutrition. I. Selowsky, Marcelo, joint author. II. Title. III. Series. TX359.R48 362.5 76-17240 ISBN 0-8018-1868-0 pbk. The most recent editions of Catalog of Publications, describing the full range of World Bank publications, and World Bank Research Program, describing each of the continuing research programs of the Bank, are available without charge from: WORLD BANK PUBLICATIONS UNIT 1818 H STREET, N.W. WASHINGTON, D.C. 20433 U.S.A. Foreword I would like to explain why the World Bank does research work and why this research is published. We feel an obligation to look beyond the projects that we help finance toward the whole resource allocation of an economy and the effectiveness of the use of those resources. Our major concern, in dealings with member countries, is that all scarce resources-including capital, skilled labor, enterprise, and know-how-should be used to their best advantage. We want to see policies that encourage appropriate increases in the supply of savings, whether domestic or international. Finally, we are required by our Articles, as well as by inclination, to use objective economic criteria in all our judgments. These are our preoccupations, and these, one way or another, are the subjects of most of our research work. Clearly, they are also the proper concern of anyone who is interested in promoting develop- ment, and so we seek to make our research papers widely available. In doing so, we have to take the risk of being misunderstood. Although these studies are published by the Bank, the views ex- pressed and the methods explored should not necessarily be con- sidered to represent the Bank's views or policies. Rather, they are offered as a modest contribution to the great discussion on how to advance the economic development of the underdeveloped world. ROBERT S. MCNAMARA President The World Bank Contents Preface xi 1. Introduction and summary I Magnitude of the world calorie deficit 2 Breastfeeding and infant malnutrition 4 Cost effectiveness of alternative policies 5 Conclusion 7 2. The nature and extent of malnutrition 8 Calories versus specific nutrients 9 Populationwide evidence on calorie deficits 10 Estimating calorie consumption by income groups 11 Projections 26 Deficits by age groups: infant malnutrition 29 3. Cost effectiveness of some policy options 39 A general (urban) food price subsidy 40 Food subsidies and their effectiveness relative to particular nutrients 44 4. Programs oriented to target groups 46 A food stamp program 47 Income transfers 48 Food price subsidies 49 Summary of cost-effectiveness comparisons 49 Feasibility and cost of actual programs 51 viii CONTENTS Appendix A. Basic country data 53 Appendix B. Income distribution data 56 Method of estimation 56 Data and results 58 Appendix C. Estimated calorie consumption functions based on cross-country data 71 Appendix D. Projected calorie deficits 75 Appendix E. Cost of programs oriented to target groups 79 Cost of an income transfer 80 Cost of a price subsidy 81 Cost of a food stamp program 81 Tables 1. Daily calorie deficits in developing countries, 1965 3 2. Calorie consumption and deficit by income groups, Brazil, 1960 11 3. Daily per capita calorie consumption and requirements in developing countries, by regions, 1965 13 4. Income distribution in developing countries, by regions, 1965 13 5. Estimated equations of per capita calorie consumption from cross-country data, by regions, 1965 15 6. Analysis of calorie consumption equations using statistically estimated coefficients of consumption propensity, 1965 16 7. Analysis of calorie consumption equations using assumed elasticities of 0.15 and 0.30 at level of calorie requirements, 1965 19 8. Number of people consuming insufficient calories, and calorie deficits in developing countries, by regions, 1965 24 9. Estimates of total daily calorie deficits in developing countries, 1965, based on regional, country, and income group averages 25 10. Calorie deficit in relation to regional consumption and GNP, 1965 26 Contents ix 11. Annual percentage growth rates of per capita income (p) and calorie consumption from other sources (A) used in three projection alternatives (A, B, and C) 27 12. Projected per capita calorie consumption by income groups, 1975 and 1990 28 13. Implied (compounded) annual percentage growth rates in mean regional per capita calorie consumption, projections A, B, and C 29 14. Size of undernourished population, 1975 and 1990, with alternative projections 30 15. Daily calorie deficit, 1975 and 1990, with alternative projections 32 16. Mean intake of nutrients in the poorest one-third of Calcutta families, by age groups 33 17. Expenditure elasticities (,B) of different nutrients, by age groups and sex, Calcutta 34 18. Expenditure and fraction of vegetarians' elasticities obtained by multiple regressions 36 19. Required marginal propensity to spend on milk to maintain nutritional status of children, Calcutta 37 20. Cost effectiveness and efficiency loss of a food price subsidy, expressed as a ratio of the price per unit 42 21. Fiscal cost and efficiency loss of alternative price subsidy policies for cereals, as a percentage of GNP 43 22. Fiscal cost of supplying an extra unit of food to the target group, expressed as a ratio of the price per unit 50 23. Cost effectiveness of target-group-oriented programs and a general food price subsidy program with different demand and supply elasticities, as a ratio of the price per unit 51 A.1. Population distribution, per capita income, and per capita calorie consumption, by region and country, 1965 53 B.1. Africa: income distribution data, selected countries, various years 60 B.2. Asia: income distribution data, selected countries, various years 62 B.3. Middle East: income distribution data, selected countries, various years 64 B.4. Latin America: income distribution data, selected countries, various years 65 B.5. Estimates of the parameters of the Lorenz curve 68 x CONTENTS B.6. Estimates of the country distribution of population by per capita income ranges: Africa, Asia, and Middle East 69 B.7. Estimates of the country distribution of population by per capita income ranges: Latin America 70 C.1. Average per capita private consumption expenditure and income elasticities of demand for food, selected regions and countries, 1964-66 72 C.2. Cross-country data, 1965, based on regressions of daily per capita calorie consumption (C) on per capita income (X) 73 D.1. Projected population by income group and region, 1965, 1975, and 1990 76 D.2. Estimated daily calorie deficit, by income group and region, 1965 77 D.3. Estimated daily per capita calorie deficit in affected populatiGon, by income group and region, 1975 and 1990 78 Figures 1. Calorie consumption by income groups, Latin America, 1965, with calorie-income elasticity equal to 0.15 20 2. Calorie consumption by income groups, Asia, 1965, with calorie-income elasticity equal to 0.15 20 3. Calorie consumption by income groups, Middle East, 1965, with calorie-income elasticity equal to 0.15 21 4. Calorie consumption by income groups, Africa, 1965, with calorie-income elasticity equal to 0.15 21 5. Calorie consumption by income groups, Latin America, 1965, with calorie-income elasticity equal to 0.30 22 6. Calorie consumption by income groups, Asia, 1965, with calorie-income elasticity equal to 0.30 22 7. Calorie consumption by income groups, Middle East, 1965, with calorie-income elasticity equal to 0.30 23 8. Calorie consumption by income groups, Africa, with calorie- income elasticity equal to 0.30 23 9. Estimated undernourished population and total population, 1965, and three alternative projections, 1975 and 1990 31 10. Analysis of the cost of a food stamp program 47 11. Analysis of the cost of a food price subsidy 48 B.1. A typical Lorenz curve relation 57 E.1. Analysis of the food market relevant for the target group 80 Preface Only in recent years has the problem of malnutrition been viewed as a development problem-as not only a consequence of underdevelopment but a contributing factor to it, a drag on the potential from which better nutrition might be provided. Since 1973 the World Bank has been examining nutrition in the context of its own programs and has taken preliminary action on several fronts, including operational activities. Malnutrition and Poverty presents the conclusions of a first large research effort in the World Bank to determine the global dimension of malnutrition in low-income countries and to investigate the eco- nomic dimensions of certain policy interventions by governments. The authors take explicit account of the close association between maldistribution of existing and projected food supplies and the incidence of malnutrition. In doing so, they have been able to improve on existing estimates of the dimension of the problem, and they demonstrate conclusively that the roots of malnutrition and policy palliatives cannot be assessed without explicit consideration of the distribution of food among different groups in the population. In short, the authors conclude that malnutrition "is not likely to go away in the normal course of development"; on the contrary, the situation may under some circumstances get worse. Similarly, although there are strong links between the food problem (which has received much attention of late) and the malnutrition problem (which has received considerably less), the authors demonstrate that the malnutrition problem is not going to be solved simply by solving the aggregate food problem. Deliberate policies and programs are called for, especially among nutritionally vulnerable groups. Of the alternatives examined, food Xui PREFACE programs oriented to target groups are regarded as more cost effective than general food price subsidies and outright income redistribution. The authors appreciate the difficulties that must be surmounted in implementing such programs, but conclude "they may offer the only hope for meeting the challenge of hunger amid plenty." Although they recognize undeniable need for expansion of the total food supply, the authors point out that the food deficit is small relative to aggregate food availability. The nutrition deficit and the distribution needs this implies is a more difficult issue with which to deal. Unlike certain other recent statements on the food problem, theirs is not a pessimistic prognosis-except in the sense that the policies proposed to solve malnutrition problems are not now being adequately pursued. This volume is published in the hope that it will help bring about a better understanding of the problem and of what is required to mreet it. The original version of this research was one of several studies undertaken by a World Bank task force on urban poverty. In the process of improving on earlier drafts, the authors received useful comments by Graham Pyatt and Lance Taylor; they are especially grateful to Nanak Kakwani for his suggestions about the estimation of the income distribution data used in the study. The final manu- script was edited by Julia McGraw. ALAN BERG Senior Nutrition Adviser Washington, D.C. Agriculture and Rural June 1976 Development Department Malnutrition and Poverty Magnitude and Policy Options 1 Introduction and Summary MALNuTRrrioN is but one of several manifestations of poverty, yet its effects, as well as the policy instruments available to improve the nutritional status of populations, have distinctive features that deserve separate analysis. First, nutrition is one of the main determinants of health, and health is now regarded as a desirable end in itself and distinct from the general objective of improving economic welfare. In this view the allocation of food and other health-preserving measures need not conform to the criteria normally used for allocating income and wealth in general. A society may have no minimum standard for income yet strive for minimum standards for health and nutrition for all its citizens. Similarly, a society may generally respect the sovereignty of family units over the allocation of their earnings but still take deliberate measures to bias allocations in favor of better nutrition for all or some members of the family. Second, health and nutrition interventions have an impact on human capital formation, with implications for the future earnings of individuals and the growth in gross national product. The nutri- tional status of infants is perhaps the most important policy-induced determinant of the individual's initial physical condition, which in turn determines the effectiveness of further investment in human capital. Certain types of malnutrition during the working years appear to have a crucial influence on an individual's level of productivity. On the policy side, the effectiveness of nutrition interventions depends heavily on the behavior of households in reaction to such policies. Households make the ultimate decision concerning nutri- 1 2 MALNUTRITION AND POVERTY tional intake, and it is in this context that public policies must be designed. The problem of nutrition also has unique features in relation to supply decisions. The greater part of infrastructure services is usually provided directly by the government, yet in making public nutrition policies in most mixed economies it must be taken into account that the private sector will ultimately be the main source of food supply. In other words, nutrition policies face a more complex situation in that private decisionmaking-as regards both supply and demand-plays a more dominant role than in other types of public intervention. That undernutrition is a function of absolute poverty is self- evident. But estimates of the global magnitude of the calorie deficits in developing countries have usually been made by refer- ence to highly aggregated per capita data. Moreover, policies to curtail undernutrition often employ instruments-such as those which increase per capita food production and consumption-that do not differentiate among beneficiaries. This study consists of two parts: an assessment of the nature and magnitude of undernutrition in the developing countries, with spe- cial reference to income distribution; and a theoretical analysis of the cost effectiveness of some policy instruments to reduce under- nutrition of target groups in urban areas. MAGNITUDE OF THE WORLD CALoRIE DEFICIT Chapter 2 discusses the nature and extent of undernutrition. It reviews the extent of undernutrition in the developing countries, as measured by the gap, or deficit, between the intake and the physiological requirements of calories. The need of deriving such deficit estimates by reference to income groups is emphasized inas- much as national averages underestimate the relevant deficit by pooling calorie surpluses and deficits. An attempt has been made here to compute deficits by income groups by using income distribu- tion data for different regions and applying plausible estimates of the calorie-income relation. Based on average calorie consumption data in the mid-1960s, it is estimated that 56 percent of the population in developing coun- tries (some 840 million people) had calorie-deficient diets in excess of 250 calories a day. Another 19 percent (some 290 million people) had deficits of less than 250 calories a day. It is noteworthy that the above estimates are not especially Introduction and Summary 3 sensitive to recommended levels of calorie requirements. This is important, since the particular level of calorie requirement used in making estimates in this and most other studies has recently come into question as overstating the true requirements. Although a lower requirement level-say, by 10 percent-would indeed wipe out the undernutrition problem when undernutrition is measured as the difference between total regionwide calorie availability and requirements, such a reduction in the level of requirements would reduce only marginally the size of the undernourished population when measured by the method used in this study. The first two columns of Table 1 show earlier estimates of calorie deficits based on regional and country averages. The third column contains the estimates by income groups derived in this study. Estimates based on regional averages are unrealistically low be- cause they contain the implicit assumption that surpluses in some countries compensate for deficits in other countries-and, similarly, that within countries, surpluses of some income groups compensate for deficits of others. The latter assumption is also implicit in esti- mates obtained by summing the average deficits of all deficit countries. Total deficit estimates based on country averages there- fore also underestimate the true deficit. From Table 1 it may be concluded that estimates of world calorie deficits given here (by the use of a mid-figure of the daily deficit of 419 thousand million calories) are one-third higher than the ones obtained through existing country average data. A better ap- preciation of a calorie deficit of this order of magnitude may be con- veyed by noting that 400 thousand million calories a day are equiva- lent to approximately 38 million metric tons of food grain a year- a figure equal to 4 percent of the world production of cereals in the mid-1960s. By the use of the same analytical framework for estimation, the Table 1. Daily Calorie Deficits in Developing Countries, 1965 (Thousand millions of calories) Regional Country Income group Region averages averages averages Latin America 0 19 32-74 Asia 202 213 225-283 Middle East 19 32 32-45 Africa 48 50 61-86 Total 269 314 350-488 Source: Table 9. 4 MALNUTRITION AND POVERTY calorie deficit for 1965 was projected forward to 1975 and 1990. With per capita income in all income groups growing at approxi- mately the current rate of overall per capita income growth and per capita food supply growing just enough to meet the demand gen- erated by the income growth, the total calorie deficit would remain virtually unchanged by 1990. The absolute number of people suffering from undemutrition would be higher, but-their proportion of the total population would decline. The per capita daily deficit in the afflicted population would decline from approximately 350 calories in 1965 to approximately 240 calories in 1990. Were income distribution to remain unchanged, extremely high rates of growth in the demand for and supply of food would be required to achieve per capita growth rates of food consumption that would eliminate the calorie deficits in the lowest income classes. The necessary growth rates in per capita income needed to sustain such growth in demand are not likely to be achieved. Furthermore, it would require a rate of growth in food production and consumption that could be achieved only if large subsidies were paid to maintain high incentive prices to farmers and low food prices to consumers. A realistic solution to the problem therefore must lie in target-group-oriented programs specifically designed to allow people of low income to achieve minimal standards of ade- quate nutrition. BREASTFEEDING AND INFANT MALNUTRITION Chapter 2 also reviews briefly some aspects of infant undemutri- tion, with particular reference to changes in income and breast- feeding. An analysis of a comprehensive nutrition survey of low- income families in Calcutta shows that the nutrition of all age groups improves with rising income. - Although the nutrient-income elasticities for young infants are higher than for adults, this does not mean that infant undemutrition is resolved with improvement in income. In fact, the opposite can be true if it is assumed that higher incomes are achieved partly through the mother obtaining employment, with the consequent partial sacrifice of breastfeeding. The same Calcutta data show that the marginal propensity of spend- ing on infants' diets from additional family income is extremely low, on the order of 5 percent. A low marginal propensity is con- sistent with a high income elasticity out of family income if, as is the case, only a small fraction of the total expenditure is devoted to childfeeding in the first place. Yet calculations about the loss of Introduction and Summary 5 breastfeeding and the cost of replacing the equivalent nutrients suggest that about 50 percent of the mother's earnings would need to be spent on the infant for the sheer maintenance of its nutritional health. Clearly, higher per capita incomes not only may fail to reduce but, on the contrary, may increase infant undernutrition. COST EFFECTIVENESS OF ALTERNATIVE POLICIES Chapters 3 and 4 contain an analysis of the cost effectiveness of various policies aimed at improving the nutritional status of the urban target population. Policies may be divided between country- wide and target-group-oriented policies. The countrywide policy considered is A GENERAL PRICE SUBSIDY on a particular food for the entire urban population. Target-group-oriented programs con- sidered are A FOOD PRICE SUBSIDY, A FOOD STAMP PROGRAM, and A STRAIGHTFORWARD INCONME TRANSFER for the undernourished popula- tion. Cost effectiveness is defined as the fiscal cost per additional unit of food or specific nutrient consumed by the undernourished group. For every policy, two scenarios were considered: first, that the addi- tional food is obtained at a constant price, as in the case of a country that imports food; and, second, that the additional food require- ment causes its price to rise. Cost effectiveness in terms of a nutri- tion objective for the target groups is obviously not the only cri- terion whereby a preferred policy is established. A general food price subsidy may be justified as an instrument for keeping down wages and wage-push inflation. Food price subsidies may also be advocated to increase farm income. Target-group-oriented pro- grams may have multiple objectives: to increase food consumption and to increase the consumption of other commodities and services. In the final analysis, political feasibility is important in selecting the optimal policy. As might be expected, a general food price subsidy is the least cost-effective policy for eliminating undernutrition in the target population. On the assumption that the target population's share of total consumption of the food is 20 percent and its demand elasticity is 1.0, it will cost approximately $5 to provide them with $1 worth of food-and this is probably the most favorable case.' If the target group's demand elasticity is 0.5, it will cost approximately $12 to provide the same amount of food. The program will be even 1. The dollars referred to throughout the book are U.S. dollars. 6 MALNUTRITION AND POVERTY more expensive if the additional food requirement leads to a higher food price. If the supply elasticity is 1.0 and the demand elasticity of the nontarget population is 0.5, it will cost $9 in the first case and $18 in the second. These illustrative but not unrepresentative figures should give second thoughts to those who seek the answer to undernutrition through general food price subsidies, either in the form of a direct subsidy or of subsidized agricultural inputs. Target-group-oriented policies are generally more cost effective. But only a food stamp program that provides participants with a claim for food in excess of their current consumption-at a cost equivalent to their food expenditure without the program-could be expected to provide $1 of food consumption per $1 program cost. The cost effectiveness of a price subsidy to the target group is approximately inversely proportional to the group's demand elas- ticity for the food: that is, it would cost about $1.33 if the elasticity were 0.75. Similarly, the cost effectiveness of a straight income transfer is approximately inversely proportional to the target group's marginal propensity to spend on the food: that is, the cost would be about $2 if the marginal propensity were 0.5. When the addi- tional food requirements for the program cause the price of food to rise, all target-group-oriented programs become only slightly less cost effective. The additional food requirement of such a program is not likely to be large (as it is in general food price subsidies), and to an extent will be compensated by a decline in food consumption by the nontarget population. In any case, the actual cost effectiveness of target-group-oriented programs may be less than is implied by the actual calculations. In reality, several kinds of leakages occur. A great deal of ingenuity is required to assure that all the benefits from a target-oriented food program reach only the intended beneficiaries. Middlemen and administrators may divert the subsidized food to the regular market. Some of the concessionary food may be purchased by people outside the target group. Where there is a food price sub- sidy, the participants may resell some food at the unsubsidized price. If this were to happen, the food price subsidy program would become only as effective as an income transfer. Actual food assistance programs in developing countries often cannot be classified as belonging strictly to one particular policy as analyzed in this study. For instance, a general food price subsidy may be applied to a commodity that is in fact primarily consumed by the target group. Public stores, selling food rations at a subsi- dized price (a method popular in several Asian countries), may Introduction and Summary 7 cater primarily but not exclusively to the target group. The cost effectiveness of food rations would depend on the size of the rations. If the ration supplies less than the amount of food participants would be consuming without the program, the cost effectiveness is no more than could be achieved with a straight income transfer. If the ration supplies all the food the participants would want to consume at the subsidized price, the cost effectiveness is the same as with a price subsidy. CONCLUSION Malnutrition is unlikely to disappear in the normal course of development: that is, in the course of normal per capita income growth, even with greater emphasis on expansion of food produc- tion-barring, of course, unusual technological breakthroughs. On the contrary the situation may worsen if present higher energy cost, leading to higher cost of food production, is not fully compensated by higher agricultural productivity. Only policies deliberately de- signed to reallocate food or income can eliminate undernutrition. Target-group-oriented food programs in urban areas and programs to assist low-income farm families to increase and stabilize pro- duction of food for their own consumption can be more cost effec- tive than outright income distribution. Although there must be no illusions about the difficulties to be surmounted in implementing such programs, they may offer the only hope for meeting the challenge of hunger amid plenty. 2 The Nature and Extent of Malnutrition JUST AS THE EDUCATIONAL LEVEL of a population should ideally be measured by its educational achievements and not by its exposure to, and use of, educational inputs, so malnutrition should ideally be defined by its consequences, such as health status, rather than by nutrient intake. In practice, it is difficult to define objec- tive indicators of consequences, and it is even more difficult to collect and interpret relevant data. The consequences of undernutrition are, first, poor bodily and mental health, which in turn causes physical suffering and mental anguish; and, second, low productivity, with effects on private and national levels of consumption and on accumulation of wealth. Such indexes of underachievement clearly are not easily defined or measured objectively. It is possible to measure mortality rates and the incidence of some diseases for children, as well as certain in- dexes of bodily and mental growth, and to attribute these with fair certainty to inadequate nutrition. In some cases, productivity data can be obtained and low productivity traced to inadequate nutrition.1 But by and large the concepts are not developed, and 1. For attempts to associate productivity with infant malnutrition and iron deficiency in workers, see, respectively, Marcelo Selowsky and Lance Taylor, "The Economics of Malnourished Children: An Example in Disinvestment in Human Capital," Economic Development and Cultural Change, 22 (October 1973): 17-30; and S. S. Basta and A. Churchill, "Iron Deficiency Anemia and the Productivity of Adult Males in Indonesia," World Bank Staff Working Paper no. 175 (Washington, D.C.: World Bank, 1974). 8 The Nature and Extent of Malnutrition 9 data are not available to provide a global picture of undernutrition on the basis of indexes of nutrition-related achievement. It is even more difficult to deduce from indexes of nutrition- related underachievement the particular explanatory variables that can be manipulated by policy measures and to isolate their effect from effects of other environmental variables conditioning those indexes of underachievement. With these reservations as background, we turn to the definition and measurement of deficiencies in the intake of nutrients as indicators of malnutrition. CALORIES VERSUS SPECIFIc NUTRIENTS The most promising indicator for measuring the extent of nutri- tional inadequacies in large population groupings is the deficit in dietary energy measured in calories. Calorie requirement standards can be determined with fair precision for specified populations. In addition to the direct ill effects of calorie-deficient diets on physi- cal and mental health and the ability to perform normal activity, calorie deficiency also often signals insufficient intake of specific nutrients such as proteins, minerals, and vitamins. This is par- ticularly true when the theory is accepted that in such cases part of the protein is used as energy. In nutrition surveys based on 7,000 households in four states of India, about 50 percent of those persons with calorie deficiencies also had inadequate protein intake. By contrast, only 5 percent of the households without calorie de- ficiencies had inadequate intake of protein.2 Protein consumption is an unreliable indicator of malnutrition because generally applicable standards of requirements are more difficult to define. Protein utilization depends on many specific conditions such as environmental health, amino acid composition of the protein, and whether part of the protein is used as calories in calorie-deficient diets. Diseases of undernutrition as a result of inadequacies of specific nutrients are more often a local or age- specific problem than is the widespread phenomenon of large popu- lation groupings' consuming insufficient calories. 2. P. V. Sukhatmne, "Incidence of Protein Deficiency in Relation to Different Diets in India," British Journal of Nutrition 24 (1970): 447-87. 10 MALNUTRITION AND POVERTY POPULATIONWIDE EVMENCE ON CALORIE DEFICITS How many people in the developing countries today suffer from a deficit in calorie intake? This seems a simple question given the vast amount of data on food consumption that has been col- lected around the world. Yet existing data and analyses remain highly unsatisfactory. Estimates of calorie deficiencies are usually based on two major data sources: surveys of household budget and food consumption and aggregate national food consumption data. Household food consumption surveys rarely cover all segments of the population in sufficient detail to derive aggregate estimates that are reconcilable with national food consumption data. Global estimates of calorie deficits and of the number of people afflicted are therefore made on the basis of comparisons between aggregate per capita calorie consumption and requirements in countries or regions. Such analyses clearly underestimate the true magnitude of the problem because of the unequal distribution of food among regions and countries and among people within countries. Such calculations imply-incorrectly-that calories consumed in excess of physio- logical requirements can be counted against the aggregate calorie requirements.' As an illustration of the order of magnitude of the actual nutri- tional inadequacies when derived on the basis of income groups rather than country averages, it is instructive to review the case of Brazil. Based on family budget surveys in the early 1960s on a national sample and a reconciliation of these data with national food consumption data, the national average daily per capita con- sumption amounted to 2,566 calories, or a calculated excess over per capita requirements of about 116 calories. Yet analysis by income groups shows that approximately 44 percent of the total population 3. For example, the Food and Agriculture Organization (FAO) estimated average calorie consumption as a percent of requirements in 1970 as follows: 105 percent in Latin America, 94 percent in Asia, 93 percent in the Middle East, 92 percent in Africa, and 95 percent in all developing regions. (Source: United Nations, Economic and Social Council, Preparatory Committee of the World Food Conference, "Assessment of Present Food Situation" [E/ Conf./Prep./61 [Rome, 1974].) Based on the average of all developing regions, the total annual calorie deficit would seem to be approximately 180 thousand million units, or the equivalent of 20 million metric tons of grain. Alternatively, when the calorie deficits in deficit countries are added up, the total annual deficit amounts to approximately 260 thousand million calories. The Nature and Extent of Malnutrition 11 Table 2. Calorie Consumption and Deficit by Income Groups, Brazil, 1960 ' Daily calorie Daily calorie Population consumption deficit Annual family Number Total income (thou- Percent Amount Percent Per (mil- (new cruzeiros) sands) of total (millions) of total capita lions) Under 100 3,583 5.05 5,172 2.87 1,006 3,604 100-149 4,873 6.87 8,847 4.91 . 634 3,089 150-249 12,235 17.25 25,940 14.41 330 4,037 250-349 10,197 14.37 23,378 12.98 157 1,601 350-499 11,145 15.71 28,293 15.71 500-799 12,884 18.16 34,958 19.41 800-1,199 7,198. 10.14 22,689 12.60 1,200-2,499 6,840 9.65 23,022 12.78 2,500 and over 1,986 2.80 7,800 4.33 Total 70,941 100.00 180,099 100.00 12,331 Source: Fundacao Getuilio Vargas, Food Consumption in Brazil: Family Budget Survey in Early 1960x (Jerusalem: Israel Program for Scientific Translations, 1970), Tables 4, A.2, A.3, and A.4. a. Deficits are defined as the difference between daily calorie requirements (2,450 calories) and actual consumption. were calorie deficient. Brazil's daily calorie deficiency amounted to some 12 thousand million calories, or 7 percent of its population's actual calorie consumption. Without regard to distribution, Brazil's level of consumption could have been judged as more than adequate (see Table 2). ESTIMATING CALORIE CONSUMPTION BY INCOME GROUPS Any attempt to make a reasonable estimate of calorie deficits in the developing countries clearly must take explicit account of the unequal distribution of calorie consumption among income groups. In the absence of direct data on the number of people whose calorie intake is short by a known amount of their requirements, an indirect estimation procedure based on existing data might serve as a reason- able approximation of the true magnitudes involved.4 The basic strategy of the analysis consists of allocating the total 4. Household food consumption sample surveys for a few urban and rural populations are of course available. The data generated in these surveys, how- ever, cannot be used directly for assessing the global dimension of food con- sumption by income groups without first reconciling the extrapolated data from such surveys and available data about national total food consumption. 12 MALNUTRITION AND POVERTY known amount of calories consumed in each major region among eight income groups. For each region, observed per capita calorie consumption (CO) consists of the weighted sum of per capita con- sumption by income groups (Ci); that is: ( 1) Co = :wiCi* where the weights (wi) are the shares of the region's population in each income group. The sum of the weights is 1. The relation between consumption and income (Xi) is specified to take on the semilogarithmic form: (2) Ci= a+bLnX,. Equation (2) implies appropriately that the calorie income elas- ticity (/A) is a declining function of consumption (and income); that is: (3) us= b/Ct. Substituting equation (2) into equation (1) yields equation (4), which is used for estimating the per capita calorie consumption by income groups consistent with total calorie consumption: (4) Co= a+b 1wLnXi. Observed per capita calorie consumption (CO) and requirements (Cr) by major regions are shown in Table 3. Per capita regional consumption estimates were derived by aggregating each country's per capita calorie consumption, weighted by its share in the region's total population.5 If the country estimates are not biased in one direction, the regional aggregates should be fairly precise, in spite of large errors in the country estimates of calorie consumption and population. The per capita calorie requirements data are regional estimates used by the Food and Agriculture Organization (FAO) of the United Nations. The distribution of population shares receiving specified levels of per capita income, reported in Table 4, were estimated on the basis of income distribution data from thirty countries, accounting for 900 million people, or about 60 percent of the population. The sources of the data and the method used for transforming income share distributions into distributions of shares of the population falling within specified ranges of per capita income are given in Appendix B. Although these data are clearly of questionable pre- 5. The basic country data are provided in Appendix A. The Nature and Extent of Malnutrition 13 Table 3. Daily per Capita Calorie Consumption and Requirements in Developing Countries, by Regions, 1965 Standard Calorie Calorie consumption requirements Region (C°) (C') Latin America 2,472 2,390 Asia 1,980 2,210 Middle East 2,315 2,450 Africa 2,154 2,350 Source: Appendix A. cisiori, the numbers are sufficiently close to other estimates to have validity in the context of the present analysis. Finally, estimates of the parameters in equation (4) are required. An estimate of b (the calorie income propensity) at once implies a value for a, since the object is to ensure that the equation is consistent with allocating the actually observed mean level of calorie consumption, (CO); that is: (5) a= C°-blwiLnXi. Table 4. Income Distribution in Developing Countries, by Regions, 1965 Per capita income Percentage of population Mean Latin Middle Class b Range (X) America Asia East Africa I Under 50 35 6.6 19.6 13.4 29.7 II 51-100 75 16.0 43.3 20.4 31.8 I1I 101-150 125 13.0 19.3 17.7 15.8 IV 151-200 175 10.7 8.3 11.4 8.4 V 201-250 225 8.6 4.0 8.1 4.9 VI 251-300 275 6.9 2.1 5.8 3.0 VII 301-350 325 5.6 1.3 4.3 1.9 VIII Over 350 1,131 32.5 0 0 0 1,119 0 2.1 0 0 717 0 0 18.9 0 908 0 0 0 4.5 Per capita income (X) 474 120 246 136 Population (millions) 244 896 144 247 Source: Appendix B. a. Income is in 1972 constant U.S. dollars. b. For classes I to VII, the means are assumed values. The mean of the highest income group was calculated to satisfy the equation XtviXs = X, i = I . .. VIII. 14 MALNUTRITION AND POVERTY As a first approach, the parameter b was estimated for each region with cross-country per capita calorie consumption and income data; that is, the equation estimated was: (6) Cj= a+bLnXj+.E, where Cj and Xj are country per capita calorie consumption and income. The results are presented in Table 5. The data and regres- sion results obtained with alternative mathematical specifications are shown in Appendix C. Several observations may be made about the appropriateness of using the parameter estimates obtained from cross-country statisti- cal analysis. Less than 50 percent of the variability in calorie con- sumption is explained by income in all regions. As a direct conse- quence and the low number of observations, the precision of the b estimates is not very high, although in three of the four regions the coefficient is statistically significant. Moreover, the low predic- tive power of the estimating equation and a priori theory suggests that income alone cannot explain all the differences in consump- tion. Intercountry variation in income distribution, urbanization, relative food price, and sociocultural factors are additional ex- planatory variables.6 To the extent that any of these omitted 6. Some attempt was made to take intracountry income distribution explicitly into account in estimating the relation between calorie consumption and income from cross-country data. The number of countries for which income distribution data are available, however, limits the number of observations. For the same fourteen countries in Latin America and ten countries in Asia and the Middle East combined, the comparable results were as follows: Latin America Cj=530+311 (zw,,LnX.j) R'=0.73 (3.7) C,= -417+452 LnX, R2=0.81 (4.8) Asia and the Middle East C,=1388+163 (2;w,,LnX,) R2=0.50 (1.6) Cj =1538+119 LnXj R2=0.38 (1.2) Because neither the precision nor the bias of the estimates was in a clear direction and the number of countries with explicit income distribution data is small, the specification of equation (6) in the text was considered preferable. The Nature and Extent of Malnutrition 15 Table 5. Estimated Equations of per Capita Calorie Consumption from Cross-country Data, by Regions, 1965 Elasticity at Coefficient region's ob- of deter- served level mination of calorie Region Equation' R2 consumption Latin America C=-256+417 LnX 0.44 0.168 (4.1) Asia C= 1191+188 LnX 0.46 0.095 (3.5) Middle East C= 1700+ 96 LnX 0.15 0.041 (2.4) Africa C= 1854+ 71 LnX 0.04 0.033 (0.6) a. Figures in parentheses are t statistics. variables are correlated with income, the calorie-income coefficient estimates obtained by the oversimplified specification could be biased. The net effect of such biases cannot be ascertained, how- ever, without examination of additional data that are not readily available. Another disadvantage of using cross-country data for estimating the parameters relevant to the estimation of calorie consumption by income groups is the limited range of variation in countries' per capita income as compared with variation in the income of indi- viduals. Even if the estimating equation adequately reflects the relation between consumption and income over the range of country per capita incomes, it does not directly follow that the equation is also appropriate for per capita incomes outside this range. The plausibility of using the statistically estimated b coefficients was further tested by using them in conjunction with the observed calorie consumption and income distribution data in the respec- tive regions. The resulting equations, implied calorie consumption levels at low- and high-income levels and corresponding elasticities, as well as the implied level of income needed for the consumption of minimum calorie requirements, are shown in Table 6. There are several nonstatistical criteria for judging the plausibil- ity of calorie consumption functions. An acceptable function should predict a level of calorie consumption consistent with the critical minimum calorie requirements needed to sustain life at the lowest prevailing income levels and also consistent with the purchasing power of the corresponding lowest income. Calorie consumption Table 6. Analysis of Calorie Consumption Equations Using Statistically Estimated Coefficients of Consumption Propensity, 1965 Income Per capita income Per capita income needed to US$25 US$3,000 meet FAO - ~~~~~~~~~~~~calorie Calorie Calorie requirements Region Equation consumption Elasticity consumption Elasticity (U.S. dollars) Latin America C= 147+417LnX 1,489 0.28 3,486 0.12 217 Asia C=1139+188 LnX 1,744 0.11 2,644 0.07 298 Middle East C=1830+ 96 LnX 2,139 0.04 2,599 0.04 638 Africa C=1836+ 71 LnX 2,065 0.03 2,404 0.03 1,393 The Nature and Extent of Malnutrition 17 data collected in a large number of household surveys provide empirical evidence that low-income groups often subsist on as little as 50 percent of the recommended FAO calorie requirement levels., The estimated low levels of calorie consumption predicted at $25 for Latin America and Asia are therefore plausible.8 On the other hand, a diet containing more than 2,000 calories per day could be expected to cost nearly or in excess of $25 a year, and therefore the levels of calorie consumption implied by the Middle East and Africa equations are likely to be too high.9 Similarly, for high levels of income a plausible equation should predict calorie intakes that are not in excess or short of calorie levels implied by the food consumption pattern of high-income classes. The equation for Latin America seems more plausible based on that criterion than the equations for the other regions. Another test of plausibility is the consistency of the implied elasticities with food income elasticity estimates provided in many food demand studies. Certainly the very low elasticities implied by the equations for the Middle East and Africa are not plausible. It should, of course, be expected that calorie elasticities be lower than food expenditure elasticities; as incomes rise, people turn to more expensive sources of calories-that is, foods containing more and higher quality nutrients per calorie and foods more palatable and convenient to prepare."0 Nevertheless, for very low levels of 7. See, for instance, U.S., Agency for International Development, A Study of Food Habits in Calcutta (Calcutta: HinduLstan Thompson Associates, 1972); and Fundacao Getulio Vargas, Food Consumption in Brazil. 8. The difference between Latin America and Asia may be a direct conse- quence of the larger share of the population in Latin America residing in urban areas. 9. For instance, if all calories are supplied by a cereal, the daily grain equivalent wotuld be approximately 600 grams. At a retail cost of 20 cents a kilogram, the annual cost of the diet would be $44. 10. Let calories (C) and the cost of calories (P) be a function of income, X, that is: (1) C=C (X) and (2) P= P(X). The respective income elasticities are then: (3) E,=C(X) C(X) and CpP(X (X) (4)~~~~~~~~~PX Food expenditure (Y) can be expressed as the cost per calorie, P(X), times the number of calories, C(X); that is: 18 MALNUTIUTION AND POVERTY income, the calorie income elasticity should be expected to be sig- nificantly higher than zero. Several analyses of data obtained through household food consumption surveys have confirmed that calorie income elasticities in the range of 0.10 to 0.30 are consistent with observed behavior.11 A final test of the plausibility of the consumption equation is the implied level of income associated with the recommended level of calorie requirements. The statistically fitted equations for the Middle East and Africa clearly imply income levels far higher than would seem plausible. In view of all the statistical and nonstatistical shortcomings of equations based on the b parameter estimated from cross-country data, two sets of calorie consumption functions were specified that, in our judgment, bracket the full range of functions consistent with data and observations from other studies. The equations and implied levels of consumption at low- and high-income levels are presented in Table 7. The two sets of functions were specified with the property of having implicit calorie-income elasticities of 0.15 and 0.30, respec- tively, at each region's level of calorie requirements. Given that the expression of the elasticity is b/C the b coefficients vary among regions in direct proportion to their different levels of calorie re- quirements. It is reasonable to expect that people have a higher pro- pensity to spend for calories at any given level of income when their physiological requirements are higher. A lower elasticity than 0.15 was ruled out simply because there is much empirical evidence from food market and household consumption studies indicating that the income elasticity should be significantly positive. A higher elasticity than 0.30 was seen to be inconsistent with the data, inas- much as a higher elasticity would imply that large low-income (5) Y==P(X) C(X). The food expenditure income elasticity is: (6) E,= [P'(X) C(X)+C'(X) P(X)1 P(X)C(X) Substituting equations (3) and (4) into (6) gives: (7) E1,=Ep+E,. The last expression shows that the calorie income elasticity is the food expendi- ture elasticity minus the cost of calorie elasticity. 11. See, for example, Fundacao Getulio Vargas, Pesquisa Sobre Consumo Alimentos (Instituto Brasileiro de Economia, 1975), pp. 192-95; D. Cherni- chovsky, "The Demand for Nutrition-an Economic Interpretation," (Washing- ton, D.C.: World Bank, 1975; processed); and Table 17 in this book. The Nature and Extent of Malnutrition 19 Table 7. Analysis of Calorie Consumption Equations Using Assumed Elasticities of 0.15 and 0.30 at Level of Calorie Requirements, 1965 Income needed to Implied per capita meet FAO calorie consumption at calorie per capita income require- of ment (U.S. Region Equation US$25 US$3,000 dollars) Elasticity at level of requirements: 0.15 ' Latin America C=471+359 LnX 1,627 3,345 210 Asia C=491+332 LnX 1,560 3,149 177 Middle East C=455+368 LnX 1,640 3,401 226 Africa C=574+353 LnX 1,710 3,400 153 Elasticity at level of requirements: 0.30 b Latin America C= -1524+717 LnX 784 4,217 235 Asia C=-997+663 LnX 1,137 4,311 126 Middle East C=-1399+735 LnX 967 4,486 188 Africa C=-1002+705 LnX 1,267 4,643 116 a. The elasticity implied at US$25 per capita income is approximately 0.22 in all regions. b. The elasticity implied at US$25 per capita income varies from 0.55 in Africa to 0.91 in Latin America. segments of the population could subsist on consumption levels too low to sustain life. Calorie consumption by income groups and population shares with the two consumption functions are shown in Figures 1 through 8. Total calorie deficit is represented by the shaded deficit area in each graph, multiplied by the region's population. Table 8 contains a summary of the estimates of the number of people consuming insufficient calories and the total calorie deficit in the developing countries, corresponding with the two sets of consumption functions. Within the limits bracketed by the two sets of functions, the estimated size of the undernourished popula- tion does not vary significantly. The size of the population with a daily per capita calorie deficit in excess of 250 calories is practically unchanged.12 As might be expected, the higher income elasticity implies lower levels of consumption for the lowest income groups and higher levels of consumption for people with higher incomes. 12. Notice that such a group could be defined as the deficitarian population if calorie requirement were to be 250 calories a day lower than the ones used through our exercise. Figure 1. Calorie Consumption by Income Groups, Latin America, 1965, with Calorie-Income Elasticity Equal to 0.15 3,500 - ......... ..Per capita requieretnent - - - Mean per capita consumption 3,000 _ g 2,500 2,472 % wI T .5s e.*....................................... ,.... .... _ __. _. ........................2,9 2390 2,000 1,500 1,000 _ _ _ _ 7 10 13 11 9 7 5 32 Shares of total population (percent) Figure 2. Calorie Consumption by Income Groups, Asia, 1965, with Calorie-Income Elasticity Equal to 0.15 3,500 --..- --. Per capita reqsiremnent -- - Afean per capita conssumption a 3,000 2,500 2,210 E !~~~~~~~~.'.x. -!. -S. 5.-S ... .. _.;:5_I...21 § 2 000 ,:::.:.:::.......... ::::: .: :.::......::. -- ,8 ,,,-,,,..,.,,.,:e ..- rt ...,:,..- .-.:..:.. ....... 2,000- 1.980 1,500 1 000 20 43 20 8 4212 Shares of total population (percent) Figure 3. Calorie Consumptioni by Income Groups, Middle East, 1965, with Calorie-Income Elasticity Equal to 0.15 3,500 --...--. Per capita rc(qtsirement -- - Afean per capita consuimption 3,000 - a-_ w C n ..l.....a. .....15 2,315 21,000 1,500 1,0-13 21 18 11 8 ff '4 19 Shares of total popuilation (percent) Figure 4. Caloric Consumption by Income Groups, Africa, 1965, with Caloric-Income Elasticity Equal to 0.15 3,500- ...-... Per capita requirement - - - Mean per capita consumption ? 3,000 - 4 2,500 2,350 ii . . . R R .: 1 _ :-: .:::: _ r . . 2,154 ~'2,000 .... 1,500 1' ,000 .- . -:: : 30 32 16 8 5 3 2 4 Shares of total population (percent) Figure 5. Calorie Consumption by Income Groups, Latin America, 1965, with Calorie-Income Elasticity Equal to 0.30 4,000 ......... Per capita requiirement --- Afean per capita consumption 3,500 ; 3,000 2,500 2,472 r ..s...>.r ............. ..... ..... .. ...... .. 2,390 2,000 1,500 1,000 - - :.:::: : ::: . 7 113 13 11 9 7 5 32 Shares of total population (percenit) Figure 6. Calorie Consumption by Itncome Groups, Asia, 1965, with Calorie-Income Elasticity Equal to 0.30 4,000 --------- Per capita requirement - - - Mean per capita consumption 3,500 r a 3,000 2,500 Sxares of tota0 :opuaton (pecnt) . . . . . . . . . . .-..-.,.,...'.,.'-.....- I,S0 . .... ... 1,000 __ . 20 43 20 8 4 21 2 Shares of total poptilation (percent) Figure 7. Calorie Consumption by Income Groups, Middle East, 1965, with Calorie-Income Elasticity Equal to 0.30 3,500 - ......... .Per capita requirement -Meas per capita coLsu,nption 3,000 - a 2500 - 2,450 :'"S'''... .'' .'.' '-.......-t_..._.......... .. .. .. .... . .. . . . E ---~~.:.--............ -.:::: ... __ __2,315 r, .:...':,.. . ...'.. :' :' :-. -..::- 2,000 .4 . . .... ...... 1,500 1000 *-:-:-'::-'-': _ __ 13 21 18 11 8 6 4 19 Shares of total population (percent) Figure 8. Calorie Consumption by Income Groups, Africa, with Calorie-Income Elasticity Equal to 0.30 4,000- ......... Per capita requirement, --- Afean per capita consumption 3,500 3,000 S 2,500 t o 230 .....::::::......... .... ... .. . . t3 - .::. ::':::' .-. :. .......:::..: 30 32 16 8 5 32 4 Shares of total poptilation (percent) 24 MALNUTRITION AND POVERTY Table 8. Number of People Consuming Insufficient Calories, and Calorie Deficits in Developing Countries, by Regions, 1965 Population with daily calorie deficits More than Fewer than Total daily 250 calories 250 calories calorie deficit Millions Average Millions Average (thousand Region of people deficit of people deficit millions) Calorie income elasticity: 0.15 Latin America 55 450 58 131 32 Asia 563 364 173 116 225 Middle East 75 407 16 94 32 Africa 151 380 39 72 61 Total 844 286 350 Calorie income elasticity: 0.30 Latin America 87 783 26 211 74 Asia 563 503 0 0 283 Middle East 48 906 25 60 45 Africa 151 570 0 0 86 Total 849 51 488 This means higher deficits for the lowest income groups but elimi- nation of deficits for higher income groups. The net effect is a larger total calories deficit. Table 9 presents comparisons of the total daily calorie deficit estimated on the basis of average consumption levels by income groups with earlier estimates based on country or regional average levels of consumption.13 Estimates based on regional averages are too low because they imply unrealistically that surplus consumption in some countries compensates for the deficits in other countries. Similarly, when the total deficit is calculated on the basis of country averages, surplus consumption by the higher income groups is assumed to compensate for the deficits in the low-income groups. 13. Estimates of the calorie deficit from regional averages (D') and country averages (D') were made on the basis of country data presented in Appendix A, as follows: D'=N(C'-2(nj/N)Cj') and D'=N,(C'-X(mk/N.)Ck), where C' is per capita calorie requirement, nj and C,° are, respectively, the population and per capita calorie consumption in each country of the region, and nk and C.' are, respectively, the population and per capita calorie consump- tion in each country with a calorie deficit. N is the total population of the re- gion, and N1 is the population in all countries of the region with a per capita calorie deficit. See also Table D.2. The Nature and Extent of Malnutrition 25 Table 9. Estimates of Total Daily Calorie Deficits in Developing Countries, 1965, Based on Regional, Country, and Income Group Averages (Thousand millions of calories) Income group averages Regional Country Elasticity Elasticity Region averages averages =0.15 =0.30 Latin America 0 19 32 74 Asia 202 213 225 283 Middle East 19 32 32 45 Africa 48 50 61 86 Total 269 314 350 488 Sources: Regional averages, computed by multiplying the region's population by the mean per capita deficit-the difference between per capita requirements and the mean per capita consumption-of the region, country averages, computed by adding each country's deficit, which is obtained by multiplying its population by the difference between per capita requirements and the per capita consumption of the country. At this point a few thoughts and figures may help to put into some perspective the magnitudes discussed so far. The estimated daily deficit of 350 thousand million calories amounted to approxi- mately 11 percent of total calorie consumption in the developing countries and to approximately 4 percent of worldwide calorie con- sumption in 1965. An estimated daily deficit of 350 thousand million calories is equivalent to approximately 100,000 metric tons of cere- als." On an annual basis, this deficit is equivalent to 36.5 million metric tons, equal to 3.8 percent of the world production of cereals and to 16.9 percent of cereal consumption in developing countries. The cost of this calorie deficit is approximately $7 thousand million, or 2.4 percent of total GNP in the developing countries and 0.3 percent of worldwide GNP in 1965. The calorie deficit in relation to regional magnitudes is shown in Table 10. Providing for additional food and resources on these orders of magnitude would have constituted no small order. In reality, much larger quantities of food would have been needed, for it is hard to conceive of a program that would eliminate the deficit in the affected population without also increasing food consumption of the population already consuming an adequate supply of calories. The full cost of eliminating calorie deficits can be severalfold the cost of the food needed in target-group-oriented programs (see Chapter 4). 14. It is assumed that one metric ton of cereals has 3.5 million calories. 26 MALNUTEUTION AND POVERTY Table 10. Calorie Deficit in Relation to Regional Consumption and GNP, 1965 Calorie deficit as a percentage of Cost of calorie deficit as a Total calorie Total cereal percentage of Region consumption consumption GNP b Latin America 5 12 0.5 Asia 13 18 4.2 Middle East 10 13 1.6 Africa 11 19 3.9 a. Annual cereal consumption is 28, 128, 26, and 33 million metric tons in, respectively, Latin America, Asia, the Middle East, and Africa. (FAO, Agricultural Commodity Projec- tions, 1970-1980 [Rome, 1971].) b. This cost estimate assumes that the calorie deficit would be made up by cereals at a retail cost of $200 a metric ton. PROJECTIONS The consumption functions developed on the basis of the 1965 data can be used to project the number of people consuming insufficient calories and the total calorie deficit in a later period. These pro- jections are made with the assumption of alternative growth rates of per capita income and per capita calorie consumption, but no change in income distribution. It is assumed further that the shares of the population in each income group will remain as in 1965, while the absolute number of people and per capita income in each income group will grow at the same rate as in the total popula- tion. The population projections by income groups in each region are given in Table D.1 in Appendix D. Per capita calorie consumption at any year T (T=O is the base year) may be conveniently represented by equation (7): (7) CT= (a+bLnXT)eXT. Letting XT=XOe T, equation (7) becomes: (8) CT=(a+bLnXO+bcpT)eXT or (9) CT= (C, + boT)e, where C. is per capita consumption in the base period, p is the growth rate of per capita income, and k is the growth rate in con- sumption as a result of factors other than the growth in per capita income. X=0 implies that per capita calorie supply increases just enough to meet the demand generated by the growth in per capita income. X7#O implies variation over time in demand or supply con- ditions. Greater awareness of the importance of good nutrition may be operative on the demand side. Most important, however, is the The Nature and Extent of Malnutrition 27 Table 11. Annual Percentage Growth Rates of per Capita Income (r/,) and Calorie Consumption from Other Sources (X) Used in Three Projection Alternatives (A, B, and C) Alternatives A B C Region ' X 0 x , X Latin America 2.0 0 3.0 0 3.0 0.25 Asia 1.0 0 2.0 0 2.0 0.25 Middle East 3.0 0 6.0 0 6.0 0.25 Africa 1.0 0 2.0 0 2.0 0.25 fact that consumption growth for any given level of income is likely to be the consequence of accelerated food production and reduction in the relative price of food. In Table 11 the annual growth rates are given of per capita in- come (4) and per capita calorie consumption as the result of fac- tors other than income (X) in three alternative projections of calorie deficits for the years 1975 and 1990.1' Throughout the "Pro- jections" section, the consumption level of each income group in 1965 is assumed to correspond with the values estimated on the basis of the lower income elasticity: that is, 0.15 at the level of the region's requirements. The estimated calorie consumption by in- come groups derived on the basis of equation (9) and the param- eter estimates presented in Table 11 are given in Table 12. The implied growth rates in the regions' overall calorie con- sumption for the three alternative projections are shown in Table 13. The three alternative projections bracket the growth rates that might realistically be expected to occur. The annual growth rate of per capita food production in the developing market economies has been estimated at 0.2 percent during 1962-72 and at 0.7 per- cent during the prior decade. For 1975 a realistic assessment would therefore be most appropriately consistent with projection alterna- tive A. To achieve calorie consumption on the levels projected under alternative C, extraordinary efforts and success in expanding food production would be required, particularly because it is doubtful that per capita income will grow at the higher rate. 15. In calculating calorie consumption levels for 1975 and 1990 by equation (9), instantaneous annual growth was assumed. The consumption levels would have been slightly less had they been estimated on the assumption of discrete annual growth rates; alternatively, these estimates imply slightly higher annual growth rates than those given in Table 11. 28 MALNUTRITION AND POVERTY Table 12. Projected Per Capita Calorie Consumption by Income Groups, 1975 and 1990 Alternative growth paths A B C Income group _ _ _ by region 1965 1975 1990 1975 1990 1975 1990 Latin America I 1,747 1,819 1,927 1,855 2,016 1,901 2,145 II 2,021 2,093 2,201 2,129 2,290 2,182 2,437 III 2,204 2,276 2,384 2,312 2,473 2,370 2,631 IV 2,325 2,397 2,505 2,433 2,594 2,494 2,760 V 2,415 2,487 2,595 2,523 '2,684 2,586 2,856 VI 2,487 2,559 2,667 2,595 2,756 2,660 2,932 VII 2,547 2,619 2,727 2,655 2,816 2,721 2,996 VIII 2,995 3,067 3,175 3,103 3,264 3,181 3,473 Mean 2,472 2,544 2,652 2,581 2,741 2,644 2,917 Asia I 1,671 1,704 1,754 1,737 1,837 1,781 1,955 II 1,924 1,957 2,007 1,990 2,090 2,040 2,224 III 2,094 2,127 2,177 2,160 2,260 2,214 2,405 IV 2,206 2,239 2,289 2,272 2,372 2,329 2,524 V 2,289 2,322 2,372 2,355 2,455 2,414 2,612 VI 2,356 2,389 2,439 2,422 2,522 2,483 2,683 VII 2,411 2,444 2,494 2,477 2,577 2,539 2,742 VIII 2,822 2,855 2,905 2,888 2,988 2,960 3,179 Mean 1,980 2,013 2,063 2,446 2,146 2,097 2,283 Middle East I 1,763 1,873 2,039 1,983 2,315 2,033 2,463 II 2,044 2,154 2,320 2,264 2,596 2,321 2,762 III 2,232 2,342 2,508 2,452 2,784 2,513 2,962 IV 2,356 2,466 2,632 2,576 2,908 2,640 3,094 V 2,448 2,558 2,724 2,668 3,000 2,735 3,192 VI 2,522 2,632 2,798 2,742 3,074 2,811 3,271 VII 2,583 2,693 2,859 2,803 3,135 2,873 3,336 VIII 2,875, 2,983 3,151 3,095 3,427 3,172 3,646 Mean 2,315 2,426 2,592 2,536 2,868 2,599 3,051 Africa I 1,829 1,864 1,917 1,899 1,995 1,946 2,123 II 2,098 2,133 2,186 2,168 2,264 2,222 2,409 III 2,278 2,313 2,366 2,348 2,444 2,407 2,600 IV 2,397 2,432 2,485 2,467 2,563 2,529 2,727 V 2,486 2,521 2,574 2,556 2,652 2,620 2,822 VI 2,557 2,592 2,645 2,627 2,723 2,693 2,897 VII 2,616 2,651 2,704 2,686 2,782 2,753 2,960 VIII 2,978 3,013 3,066 3,048 3,144 3,124 3,345 Mean 2,154 2,189 2,242 2,244 2,320 2,279 2,468 The Nature and Extent of Malnutrition 29 Table 13. Implied (Compounded) Annual Percentage Growth Rates in Mean Regional per Capita Calorie Consumption, Projections A, B, and C Growth rates (percent) Region A B C Latin America 0.3 0.4 0.7 Asia 0.2 0.3 0.6 Middle East 0.5 0.9 1.1 Africa 0.2 0.3 0.5 The daily per capita calorie deficit estimates in the affected popu- lation groups by 1975 and 1990 are given in Table D.3 in Ap- pendix D. The estimated size of the undernourished population in 1975 and 1990 under the alternative projections is shown in Table 14 and Figure 9. The total calorie deficits are estimated under the three alternative projection assumptions and are shown in Table 15. The above projections warrant several conclusions. Per capita growth in food consumption that is just sufficient to meet the de- mand generated by the growth in per capita income, without in- come redistribution, would make little contribution to the reduction of calorie undernutrition. By contrast, high rates of growth in per capita food consumption (based on high rates of food availability and lower food prices) could substantially reduce undernutrition by 1990, provided that income and food consumption in the lowest income groups grow at the same rate as in the higher income groups. Under projection Alternative C, the proportion of the severely undernourished (in excess of 250 calories below require- ments) in the total population could be expected to decline from 55 percent in 1965 to 12 percent by 1990. Given realistic assumptions, however, about per capita income growth rates and possibilities for accelerating food production at prevailing production costs, we believe that only government inter- vention specifically designed to subsidize food production or to assist the poor to achieve minimally satisfactory levels of food con- sumption could lead to the elimination of undernutrition. Various program options directed to this goal are discussed in the final chapter. DEFICITS BY AGE GROUPS: INFANT MALNUTRITION Average per capita nutrition data do not address the problem of the distribution of nutrition intake within the family. Although per Table 14. Size of Undernourished Population, 1975 and 1990, with Alternative Projections (Population in millions) Population with calorie deficit Population with calorie deficits in excess of 250 calories below requirements below requirements Region A B C A B C 1975 Latin America 112 (0.36)' 112 (0.36) 112 (0.36) 71 (0.23) 71 (0.23) 21 (0.07) Asia 924 (0.82) 924 (0.82) 707 (0.63) 707 (0.63) 707 (0.63) 221 (0.20) Middle East 94 (0.51) 61 (0.33) 61 (0.33) 61 (0.33) 24 (0.13) 24 (0.13) Africa 243 (0.77) 193 (0.61) 193 (0.61) 93 (0.61) 93 (0.61) 93 (0.61) Total 1,373 (0.71) 1,290 (0.66) 1,073 (0.55) 932 (0.48) 895 (0.42) 359 (0.19) 1990 Latin America 102 (0.23) 102 (0.23) 30 (0.07) 30 (0.07) 30 (0.07) 30 (0.07) Asia 1,299 (0.82) 994 (0.63) 994 (0.63) 311 (0.20) 311 (0.20) 311 (0.20) Middle East 89 (0.34) 35 (0.13) 0 35 (0.13) 0 0 Africa 280 (0.62) 280 (0.62) 135 (0.30) 135 (0.30) 135 (0.30) 0 Total 1,770 (0.64) 1,411 (0.51) 1,159 (0.42) 511 (0.19) 476 (0.17) 341 (0.12) a. Figures in parentheses represent percentage of total population. The Nature and Extent of Malnutrition 31 Figure 9. Estimated Undernourished Population and Total Population, 1965, and Three Alternative Projections, 1975 and 1990 2,800 \ Population with deficit of more than 250 calories [Z:::::. Population with deficit of 2400 tless than 250 calories 2,4000 2,000_ *! 1,600 ........ F ................... ... ........ .. E 1,200 ....... ........ 800i0t A B C A B C 1965 1975 1990 Year and projection 32 MALNUTRITION AND POVERTY Table 15. Daily Calorie Deficit, 1975 and 1990, with Alternative Projections (Thousand millions of calories) 1975 1990 Region A B C A B C Latin America 32 27 21 27 18 7 Asia 283 253 222 291 198 79 Middle East 28 18 15 21 5 - Africa 69 60 50 82 60 31 Total 412 358 308 421 281 117 capita figures might well show an intake above requirements, some family members-particularly classified by age groups-experience a strong deficit of that particular nutrient. Protein-calorie deficiency in infants and young children appears to be one of the main features of the overall nutrition problem in developing countries. The mag- nitude of the problem appears not only in its rate of incidence in low-income families of these countries but also in its potential effect on the physical and mental development of the individual. In an attempt to evaluate the effect of income on the nutrient intake by age groups, it is useful to review data for Calcutta and to examine some of the evidence on the decline in breastfeeding, one of the principal factors in infant undernutrition in urban areas of developing countries. Data on the intake of nutrients by age groups are extremely difficult and expensive to generate, and few studies report on such data out of food surveys. Perhaps the best source of such data, and worth examining as an illustrative case, is the Calcutta Food Survey undertaken in 1969.1' A summary is given of the mean intake of different nutrients (as a percentage of the requirements) by selected age groups for the poorest one-third of the families in Calcutta (see Table 16). They correspond to all families with a monthly per capita expenditure of less than 40 rupees a month. A strong difference is evident between the deficit in calories and in proteins; although protein consumption is relatively close to requirements, the deficit in calories is substan- tial for all age groups. The implications of such a calorie deficit, particularly for young children, are of prime importance. When the body lacks calories, it uses up protein as a source of energy. Proteins, therefore, are not 16. U.S., Agency for International Development, Food Habits in Calcutta. The Nature and Extent of Malnutrition 33 Table 16. Mean Intake of Nutrients in the Poorest One-Third of Calcutta Families, by Age Groups (Percentage of requirements) Age groutp Calories Proteins Calcium Vitamin A 2-4 years 55 105 70 75 12-16 years 56 85 69 65 22-56 years 59 95 94 69 Source: U.S., Agency for International Development, A Study of Food Habits in Calcutta (Calcutta: Hindustan Thompson Associates, 1972). available for other uses such as cell maintenance and cell formation. To determine the effect of income on nutrition, the intake of each nutrient by age group was regressed on the per capita expenditure of the family. The results are given in Table 17, where ,/ represents the expenditure elasticity of the nutrient in question. The regression results do not appear significant for iron except for children younger than one year. For all other nutrients the'p results are highly significant except for infants younger than six months. There are three principal results. First, all elasticities are smaller than 1 and most are smaller than one-half; this is simply Engel's law. Although expressions allowing for a variable elasticity were attempted, the constant_Plasticity4-form gave th best fit. i Second, it is possible to observe certain patterns of the elasticity across age groups. Young children appear to have a higher elasticity te5 l than older children and adults. Therefore, although nutrition intake does not grow at an equal rate with income, there is a positive redistribution toward young children. Third, the elasticities for calories and proteins are extremely or close, suggesting that both are derived from the same staple. With 5 increased income, consumption of that staple is simply increased without major shifts toward other types of foods. Finally, because the survey provides data on the fraction of vegetarians in each expenditure bracket, that information is in- cluded as an additional explanatory variable (see Table 18). Results of Table 18 show that the fraction-of-vegetarians variable is significant in explaining nutritional intakes of infants younger than six months; in addition, it increases substantially the coefficient and significance of the expenditure variable. Holding constant food habits as determined by vegetarian practices, the effect of income on infant nutrition becomes quite strong. The coefficient of the fraction-of-vegetarians variable is negative: that is, when income is held constant, vegetarian habits have a 34 MALNUTRITION AND POVERTY Table 17. Expenditure Elasticities (,8) of Different Nutrients, by Age Groups and Sex, Calcutta Calories Protein Vitamin A Iron Calcium Age , R 2 R 2 ' R3 R2' p R2 Younger than 0.44 0.19 0.44 0.11 0.18 0 0.57 0.61 0.49 0.24 6 months (1.68) (1.42) (0.31) (3.65) (1.87) Younger than 0.26 0.43 0.33 0.60 0.55 0.82 0.35 0.46 0.37 0.49 1 year (3.03) (4.17) (7.24) (3.19) (3.42) 1-2 years 0.31 0.94 0.38 0.92 0.63 0.81 0.05 0.14 0.61 0.91 (13.66) (11.89) (7.21) (1.73) (11.13) 2-3 years 0.32 0.90 0.32 0.85 0.55 0.85 0.04 0 0.50 0.83 (10.64) (8.59) (8.58) (0.38) (8.06) 3-4 years 0.33 0.94 0.29 0.93 0.43 0.82 0.03 0 0.55 0.88 (14.84) (12.94) (7.86) (0.65) (10.01) 4-5 years 0.21 0.68 0.19 0.68 0.41 0.76 -0.08 0.04 0.59 0.86 (5.30) (5.31) (6.54) (1.21) (9.02) 5-6 years 0.26 0.86 0.24 0.73 0.39 0.60 0.06 0.08 0.56 0.94 (8.90) (6.02) (4.49) (1.45) (13.70) 7-11 years 0.24 0.86 0.19 0.65 0.39 0.70 0.05 0.27 0.39 0.55 (8.87) (5.00) (5.55) (2.42) (4.10) 12-16 years 0.18 0.84 0.12 0.33 0.25 0.55 0.04 0.02 0.24 0.27 (male) (8.40) (2.72) (4.10) (1.12) (2.43) 12-16 years 0.24 0.86 0.24 0.75 0.34 0.52 0.05 0.11 0.46 0.78 (female) (9.04) (6.27) (3.85) (1.62) (6.87) 17-21 years 0.17 0.85 0.17 0.68 0.32 0.55 0.02 0 0.36 0.87 (male) (8.51) (5.30) (4.11) (0.52) (9.27) 17-21 years 0.16 0.63 0.14 0.52 0.14 0.13 -.001 0 0.35 0.82 (female) (4.85) (3.88) (1.71) (0.02) (7.64) 22-56 years 0.20 0.95 0.19 0.94 0.27 0.52 0.05 0.51 0.42 0.94 (male) (15.16) (14.17) (3.84) (3.80) (13.91) 22-56 years 0.20 0.94 0.17 0.87 0.34 0.80 -0.02 0.01 0.41 0.94 (female) (13.90) (°.40) (7.17) (1.06) (13.80) 57 years+ 0.23 0.72 0.23 0.78 0.30 0.43 0.08 0.14 0.48 0.90 (male) (5.86) (6.86) (3.27) (1.76) (10.63) Note: Figures in parentheses are t-statistics. a. Coefficient of determination. negative effect on the child's nutrition status. The earlier, low figure of the elasticity of expenditure was therefore probably a combina- tion of the positive income effect and the negative vegetarian effect because a higher fraction of vegetarians was found in the higher income groups. The analysis of the Calcutta data might be interpreted super- The Nature and Extent of Malnutrition 35 ficially to mean that income growth improves nutrition in all age groups. It must be noted, however, that the data pertained only to purchased food. The decline in breastfeeding practices is per- haps the most distinctive feature behind infant undernutrition in urban areas; urbanization tends to shorten the breastfeeding period. In evidence summarized by Berg for countries such as Guatemala, Indonesia, Taiwan, and Zambia, a strong difference in the fraction of mothers breastfeeding is found between urban and rural areas." Jelliffe reviewed similar evidence for Jamaica, Brazil, Chile, Panama, and Guatemala."' Specific urban-rural comparisons for the poorer segments of the population in Jamaica and Guatemala show the same pattem; in the case of Guatemala, rural Indians breastfeed two and a half years, compared with less than half a year for low-income urban groups. Although among lower income moth- ers in Singapore in 1951, 62 percent of babies were breastfed at six months of age, the figure declined to 27 percent in 1960. Of several determining factors in the decline in breastfeeding practices in low-income urban groups, some are sociological, such as imitation of other socioeconomic groups;-9 others are economic, perhaps the most important of which is the increased participation of women in the labor force. Infant malnutrition as a result of absence from the home of lactating mothers participating in the labor force is an interesting case of income redistribution within the family in the face of urbanization: although the family's real monetary income rises because of the mother's work, the child suffers a "negative income effect." Unless sufficient food for the infant to compensate for nutrients lost by reduced breastfeeding is provided from that in- creased family income, that negative income effect remains.20 This will basically depend on the marginal propensity to spend on food for the infant. A rough estimate has been made, for urban India, of the required marginal propensity to spend on the infant so as to maintain con- stant his nutritional status. Assume that the mother substitutes commercially sold cow milk (M.), purchased at price (PO), for 17. Alan Berg, The Nutrition Factor: Its Role in National Development (Cambridge, Mass.: M.I.T. Press, 1973). 18. D. Jelliffe, Infant Nutrition in the Subtropics and Tropics (Geneva: World Health Organization, 1955). 19. Ibid. 20. Breastfeeding provides other benefits: for example, it extends the spac- ing of births and diminishes exposure of the infant to health hazards in unsani- tary environments. Table 18. Expenditure and Fraction-of-Vegetarians' Elasticities Obtained by Multiple Regressions Calories Proteins Vitamin A Expendi- Vegetar- Expendi- Vegetar- Expendi- Vegetar- Age group ture ians R2 ture ians R ture ians R2 Younger than 0.5 year 1.12 -11.85 0.70 1.28 -14.68 0.73 1.73 -72.20 0.64 (3.71) (2.85) (3.97) (3.31) (2.83) (3.22) 0.5-1 year 0.24 0.49 a 0.96 (3.34) (5.00) 4-5 years 0.22 0.37 0.89 (3.24) (3.41) 22-56 years 0.22 0.80 0.97 (15.42) (2.62) 3 Note: Figures in parentheses are t-statistics. ' a. Indicates the coefficient of the vegetarian variable in its natural form. 0 t' The Nature and Extent of Malnutrition 37 a fraction of breast milk (MB). When she goes to work and eams an income (AY), the required marginal propensity can be calcu- lated by the expression: P. MBM1 where a represents a conversion factor between both types of milk to correct for different nutrient contents.2' In calculating the value of m, it was assumed that because the mother goes to work the infant receives only 30 percent of her potential breast milk: that is, the required marginal propensity out of the increased income of the family to spend on the infant's milk must be sufficiently high to substitute for the loss of 70 percent of the mother's breast milk. The monthly cost of replacing 70 per- cent of breast milk at different ages of the infant is shown in row 4, Table 19. The comparison of such figures with a rough estimate of the monthly income of an unskilled woman in urban India gives the implicit required marginal propensity to spend in commercial milk for the child's purpose. It amounts to 0.53 in the first six months of the infant's life (see row 6, Table 19). The figure of 0.53 may be compared with the marginal propensity 21. The amount of cow milk required is M, MB. The equivalent marginal propensity times the income change equals the expenditure on milk: that is, mnAY=P M-. Combining the two equations yields the expression in the text. Table 19. Required Marginal Propensity to Spend on Milk to Maintain Nutritional Status of Children, Calcutta Intfants' ages (months) Item 0-6 7-12 13-18 19-24 0-24 1. Monthly potential breast milk (liters) 25.9 15.2 15.2 6.1 15.6 2. Cow milk equivalent (liters) 30.5 17.5 17.5 7.0 18.2 3. Cost of milk equivalent (rupees) 91.5 52.5 52.5 21.0 54.6 4. 70 percent of row 3, above 64.0 36.7 36.7 14.7 38.2 5. Monthly income of unskilled woman, 1973 (rupees) 120 120 120 120 120 6. Required marginal propensity to spend on milk 0.53 0.30 0.30 0,12 0.32 Sources: rows I and 2, Alan Berg, The Nutrition Factor: Its Role in National Develop- ,nent (Cambridge, Mass.: M.I.T. Press, 1973); row 3, using a price of 3 rupees a liter (price as of October 1972 in Bombay and Madras); row 5, corresponds approximately to the monthly wage of a woman in domestic service in a large urban area. 38 MALNUTRITION AND POVERTY to spend on food observed in the Calcutta Food Survey. That ob- served value, for children of younger than one year, is below 0.06, almost one-tenth of the required value. This strong difference between the observed and required marginal propensity gives sup- port to the conclusion that, even under optimistic assumptions, a decline in breastfeeding as a result of the mother's participation in the labor force could have an important negative effect on the nutritional status of the child. 3 Cost Effectiveness of Some Policy Options EVALUATION OF POLICY OPTIONS directed toward increasing the nutritional intake of a target group would require answers to the following questions: (a) What are the variables explaining the behavior of the typical low-income urban household in relation to its intake (and distribution within the household) of different nutrients? Such variables would be both economic-that is, in- come and the relative cost of different nutrients-as well as cultural determinants of food habits. (b) Which of these variables can be affected by policy instruments in the short and long run? (c) What is the cost effectiveness of inducing changes in different varia- bles: that is, the cost incurred by that policy per unit change in nutrient consumption by the target group? This study does not address this full set of questions; rather, it concentrates basically on (c), the cost effectiveness of alternative policies. For such a discussion it is useful to distinguish between two large typologies of policies (and programs): "countrywide" and "target-group oriented." The former includes those programs which reach the target group in the process of reaching all segments of the population. Given that these programs require a subsidy or transfer payment, nontarget groups also get subsidized. Target- oriented programs, on the other hand, are those in which particular target groups (by income categories, by age groups, and so forth) may be reached without subsidizing nontarget groups. Most of the policy options discussed here are applicable only to urban populations because many of the target-oriented programs are almost impossible to institutionalize in a rural environment. This 39 40 MALNUTRITION AND POVERTY makes the discussion easier inasmuch as there is no need to trace the nutrition impact of income effects derived from changes in rural production and incomes, as induced by alternative food policies. In the first section below the cost effectiveness of a general (urban) food price subsidy is examined. The modifications needed to analyze the cost effectiveness of a policy that aims at nutrient- specific objectives is described in the second section. In Chapter 4 the cost effectiveness of target-group-oriented programs is con- sidered, specifically a food price subsidy, an income transfer, and a food stamp program. A GENERAL (URBAN) FOOD PRICE SUBSIDY Assume two groups of urban consumers of a particular food com- modity: the target, or poorer, group (group p) and the remaining, or richer, consumers (group r). We begin by defining: (10) qr= aqp+ (1-a) -,, where 7,r is the (absolute) plice elasticity of the demand for the commodity expressed as the weighted average of the elasticities of both groups; and a represents the share of consumption of the poorer group. The object is to increase the consumption of this food in the target group by a fraction equal to A by reducing the price to consumers through a price subsidy. If the initial price is equal to 1, two equations result: (11) Target A= qpdpd and (12) Aggregate equilibrium 'qTdpd= edp9, where dpd and dp8 are the percentage changes in the demand and supply prices, and where e represents the price elasticity of supply. Solving for dpd and dp8 gives: (13) dpd = A and (14) dp =(rT Denoting S the fiscal cost of the subsidy as a fraction of the initial (urban) expenditure in the commodity, it is possible to write: (15) S=(dpd+dps)(1+17Tdpd) and ( 16 ) S =- 1 + -AT)1+X A-A . Cost Effectiveness of Some Policy Options 41 Equation (16), when evaluated for a situation where the com- modity is partly imported (f = X ), must be interpreted as the cost of a general consumption subsidy that does not discriminate be- tween imports and domestic production. Domestic producers con- tinue to receive the world price of the commodity. An altemative, quite common policy is to subsidize only the imported fraction of the commodity. Although this policy reduces the fiscal cost, it also induces a fall in farm income and a production efficiency loss because imports replace cheaper domestic production. Under such a policy the value of S becomes: ( 17) ,1p [ +x1V ) ( p where 7r is the initial fraction of domestic production in total con- sumption, and e represents the domestic supply elasticity. Corre- spondingly, the efficiency loss of the policy, expressed as a fraction of the initial expenditure on the commodity, may be written as:' (18) L= (-) ,r C. An expression of the cost effectiveness of this policy, the cost per additional unit of food consumed by the target group as a ratio of the food's preprogram price, may be written as: (19) s= ) 1. The efficiency loss is derived by assuming a linear supply function of do- mestic producers. It is of interest to compare the tradeoff between the fiscal gain and the (production) efficiency loss in using this alternative policy. The fiscal gain is the difference between cquation (16), evaluated with C= 00, and equation (17) where e must be interpreted as the elasticity of domestic supply: (1') Fiscal gain=FG=- (1+XA77 )--2[(1±x-)- X(i-,)] or 77D np ?PI 7p X7 ) , (2') =r-(1 C ). 77p 71P The fiscal gain will be larger than the efficiency loss, FG > L, when: (3') ( -x- > 3.' X- or (4') x -< (,3). The fiscal gain will be greater as long as (x e )is less than two-thirds. ?P 42 MALNUTRITION AND POVERTY Table 20. Cost Effectiveness and Efficiency Loss of a Food Price Subsidy, Expressed as a Ratio of the Price per Unit a=0.2 ac=0.7 Program q,=0.5 m77=1.0 7p,=0.5 i7P=l.0 General consumption fe=1.0 18.0 8.9 5.1 3.1 subsidy 0=oO 12.0 5.6 3.4 1.7 '7r=0.25 10.5 4.6 3.0 1.4 (0.5) (0.13) (0.14) (0.04) Subsidy on imports 7r=0.50 9.0 3.6 2.6 1.1 (1.0) (0.25) (0.28) (0.07) 7r=0.75 7.5 2.6 2.1 0.8 (1.5) (0.38) (0.43) (0.11) Note: ?2 = 0.2; 7,7 = 0.5. Figures in parentheses show the efficiency loss (L/Xa). Equation (19) presents the value of s in terms of the initial price of the commodity, which has the role of a scaling factor.2 The values of s for selected parameters are given in Table 20. It would appear at first that any program in which the cost of deliver- ing a unit of food exceeds the unit price of that food cannot be very cost effective. Yet, as will be shown later, the cost of delivering a unit of additional food directly to the target population will usually also be in excess of the unit price of the food. Both in theory and by hard evidence it has been found impossible to fine-tune food assis- tance programs so that every additional unit of food delivered is converted into a full additional unit of food consumed. Furthermore, direct food assistance programs usually entail high delivery costs in addition to the cost of the food itself. Clearly, when the target population, and hence its share in the total consumption of a given food (a) is small, the fiscal cost of a general subsidy program per added unit of consumption in the target group becomes prohibitively expensive. The application of such a program would be least desirable in a middle-income country with a large share of the population already receiving adequate nutrition and an inelastic food supply. On the contrary, general food subsidy programs might be very cost-effective alternatives to target-group-oriented programs in countries where a large pro- portion of the population is inadequately fed (a is high) and the food supply relatively elastic. 2. Accordingly, the efficiency loss per extra unit of food consumed by the target groups may be defined as L/Xa. Cost Effectiveness of Some Policy Options 43 The higher fiscal cost effectiveness of a general food subsidy pro- gram in a poor country (with a high a) unfortunately does not suggest that such a country is in a better position to eliminate nu- tritional deficiencies than the richer country (with a low a). In equation (16) it can be seen that the total cost is the same for both types of countries (independent of the value of ao) when q,= 'gr=77T. With a demand elasticity of 0.5, the cost is about equal to the country's urban expenditure on that food if the supply elasticity is 0.5 and about half that cost if food supply is infinitely elastic. The cost, however, will be a much higher proportion of GNP for the poorer country and therefore a much greater budgetary burden. To illustrate the point, the total cost of such programs for a par- ticular food commodity, cereals, has been computed. In Table 21 this cost is presented as a fraction of the GNP of the country, for alternative values of per capita income of the country and the world price of cereals. In these calculations the urban population is as- sumed to represent half of the total population. The figures in Table 21 suggest that for a country such as India, with a per capita income of $100, the total cost of such a program becomes prohibitively expensive under existing world prices of cereals. It may be concluded, therefore, that the total cost of the program relative to GNP is not an argument extraneous to the choice of a cost-effective program. Our original observation that a general Table 21. Fiscal Cost and Efficiency Loss of Alternative Price Subsidy Policies for Cereals, as a Percentage of GNP Cost ot policy at different Per capita per ton world prices of income of cereals (percent of GNP) the country Program (U.S. dollars) US$200 US$300 General consumption subsidy 100 5.6 8.3 (e= oo) { 300 1.9 2.8 100 3.8 5.7 Subsidy on imports (0.3) (0.5) (e=1; 7r=0.5) 300 1.3 1.9 (0.1) (0.2) Note: X = =0.2; -, = 0.5; 71, = 0.75. The values in parentheses refer to the efficiency loss. The following expressions are used to compute the figures in the table: Fiscal cost (P/y),..qM.S and Eflicienscy loss (P/y)If q,.L, where (P/y) is the ratio of the world price of cereals with respect to the per capita income of the country, 1. is the share of urban population, and q. is the yearly per capita physical consumption of cereals in urban areas. 1, was assumed to be 0.5 and qu = 0.182 metric tons a year, equivalent to 500 grams a day. 44 MALNUTRITION AND POVERTY food subsidy program is more cost effective in a poor country may need modification if such a country cannot bear the cost. Regardless of a high level of food deficiency affecting a large part of the popula- tion, the country may be forced to scale down its program objec- tives: the percentage of additional food to be given (A) and the share of the population with the most severe deficiencies to be reached (a). In that case such a country would also have to resort to more cost-effective target-group-oriented programs to reduce undemutrition among the worst affected of the population. FOOD SUBSIDIES AND THEIR EFFECTIVENESS RELATIVE TO PARTICULAR NUTRIENTS A rigorous comparison of the cost effectiveness of price subsidies on altemative foods ought to be carried out in relation to the particular nutrient in question (where calories may be considered a nutrient). The proper cost concept is the cost of an incremental unit of a nutrient (consumed by the target group) as a consequence of price subsidies on altemative food commodities. This becomes particu- larly important when the subsidy on a particular food, because of cross effects in consumption, changes the amount of other food commodities consumed. The new concept introduced here is the elasticity of nutrient N with respect to the price of food 1: that is, the percentage increase in the intake of N when the price of that food commodity changes by 1 percent. Of obvious interest here is the elasticity relevant to the particular target group being analyzed. Let us define such elasticity as EN,: (20) ENI=nl ni771+1l n21+2la+ n3)31 . . .nm, where M represents the number of food commodities, and 71, and '7ij the own and cross-price elasticities with respect to the price of food 1 in the target group, respectively; and m4 represents the (base period) share of the total intake of nutrient N derived from consuming food i in the target group. A price subsidy to food 1 will increase the intake of nutrient N only if EN, is negative; such a result is not obvious and will depend on the sign of the Iijs, namely, to what extent other food commodi- ties are (gross) substitutes or complements of food commodity 1. If commodity 1 is a superior good, then n. will always be negative; Cost Efectiveness of Some Policy Options 45 in this case a negative value of EN, is obtained under either of the following conditions described by equations (21) and (22): (21) i>l that is, on the average, other food commodities are complements or, at most, unrelated to food commodity 1, where the nis are the weights in computing such an average, or: '>ni ( 22) In,n j1>°; but--q,,> 2:n, / On the average other food commodities are substitutes, but in ab- solute terms the weighted sum of their elasticities is lower than the own-price elasticity of food commodity 1. With EN, defined, the fiscal cost per unit increment in nutrient N may be computed in a way analogous to that in the preceding section. The only difference is that now k is replaced by AN: that is, the objective is to increase the intake of nutrient N in the target group by a fraction equal to AN: (23) AN= EN1 dpd and (24) ?IT dpd= edps. Solving for dpd and dp8 and substituting into equation (15), the following is obtained: (25) SN= E + 17T ' where SN is the fiscal cost of the subsidy as a fraction of the initial total expenditure in commodity 1, aimed at achieving an increase in the intake of N in the target group by a fraction AN; and i7r is the (own-) price elasticity of demand for commodity 1 by all groups. Equation (25) is quite symmetric with equation (16) developed earlier; A has been replaced by AN and EN1, or the nutrient elasticity, has been substituted for ,p. The cost per unit increment in the intake of nutrient N may now be written: (26) ( n )S where 4,, represents the amount of nutrient N per dollar spent in food commodity 1. o. reflects again the (initial) share of such food consumed by the target group. 4 Programs Oriented to Target Groups PROGRAMS THAT ARE ORIENTED TO TARGET GROUPS may range from straight income transfers and alleviation of unemployment to the delivery of specific foods to target groups. Possible program options include nutrition education, distribution of food at a price discount, food stamp programs for selected families, free food rations for selected families, food rations for children and pregnant mothers delivered to the home, and feeding of selected population groups in schools and factories. The cost-effectiveness features of three target-group-oriented programs-a food price subsidy, a food stamp plan, and a straight income transfer-are considered here. A food price subsidy pro- vides beneficiaries with as much food as they wish to purchase at a reduced price. A food stamp program, widely adopted in the United States, provides participants with the opportunity of pur- chasing a fixed amount of food at a lesser cost than market prices. These three options are examined by reference to a graphical analysis of a consumer's indifference curves in Figures 10 and 11. The budget line X,Y0 in Figure 10 illustrates the opportunities a consumer has for allocating a fixed amount of income to food and nonfood purchases. Given the indifference curve AB, the initial equilibrium of the consumer is P, with an initial consumption of food equal to F. The curve OZ, the income-expenditure line, il- lustrates the consumer's preference for allocating increasing levels of income to food and nonfood purchases when the budget line has the same slope as X,Y0: that is, the price ratio of food and nonfood remains constant. 46 Programs Oriented to Target Groups 47 Figure 10. Analysis of the Cost of a Food Stamp Program z Y. ~A P ~D B 0 F Xi F* XO Food The objective of the target-group-oriented program is to increase the consumer's food consumption from his initial level, F, to a higher level, F4; we now consider the costs of alternative programs to reach this objective. A FOOD STAMP PROGRAM The cost of an optimal, as well as of a suboptimal, cost-effective food stamp program is illustrated in Figure 10. The optimal program would provide the consumer with a claim for food with a market value of F4 and charge him OX1. The consumer will be in equi- librium at C and the cost of the program would be X,F'. Clearly, such a program would be highly cost effective; the beneficiary would spend not only the full income increment contained in the food sub- sidy, but would also transfer to food expenditure some of his own in- come formerly spent on other items.' By reference to the cost effectiveness employed earlier, it is evident that the cost per unit of food reaching the target group would be less than the price of the food. To achieve such cost effec- tiveness, however, would require knowledge of the consumer's in- 1. Notice, however, that with this program the consumer is, at the margin, indifferent between participating or not participating in the food plan because his level of utility does not increase by participating in the program. In a rigorous definition, the optimal program is the one charging him the maxi- mum price for the food coupon consistent with his participation in the program. 48 MALNUTRITION AND POVERTY Figure 11. Analysis of the Cost of a Food Price Subsidy "2 , P i D 0 X, X, F FO XO X4 Food difference map (knowledge of point C). If, for example, con- sumer's preference for food is overestimated and he is charged OX1, he may choose not to participate in the program. A suboptimal but more realistic food stamp plan would set the cost of the food stamps at OF: that is, at the consumer's expenditure for food without the program. The consumer will be in equilibrium at D and the cost of such a suboptimal food plan becomes FF', the cost per unit of additional food consumed by the target group is just equal to the price of the food. A basic condition for implementing a cost-effective food stamp program is differential costing of the stamps for participants in accord with their expenditures on food (and as a proxy of their income levels). Otherwise, those with the greatest need will re- fuse to participate in the program and participants who have higher-than-average incomes will use only a fraction of the subsidy for added food consumption. INCOME TRANSFERS Figure 11 shows that the income transfer required to induce a consumer to increase his food consumption from F to F* is X2F'. What determines the cost effectiveness of an income transfer as a means of increasing food consumption is the slope of PZ; such a Programs Oriented to Target Groups 49 slope is determined by the marginal propensity to consume food in the relevant range of income changes induced by the transfer. Nor- mally, the higher the income level of the target group, the lower the marginal propensity to consume food. Hence, an income trans- fer is relatively less cost effective. It is also noteworthy that the marginal propensity to consume food of desirable nutritional quality is usually lower in urban than in rural areas for a given food-nonfood price ratio and that the relative price of food is usually higher in urban areas. Consequently, income transfers are likely to be most cost ineffective for urban dwellers and for all but the poorest among them. FOOD PRICE SUBSIDIES A food price subsidy is illustrated in Figure 11. by the new budget line Y0X4. The cost of inducing food consumption to increase from F to F° is equal to X8F'. The extent of the price subsidy (that is, the required slope of the new budget line YOX4.) needed to induce food consumption at a level F° depends on the curvature of the indifference map. Because it is safe to assume, however, that added food consumption has positive utility and the price elasticity for food is less than 1, the new price line must cross the F° level of consumption between E and D2. This means that the cost of a food price subsidy, X3F', will be less than the cost of a straight income transfer and more than the cost of a food stamp program. SUMMARY OF COST-EFFECTIVENESS COMPARISONS Cost-effectiveness formulas for the three target-group-oriented pro- grams already described are summarized in Table 22. Such formu- las define the cost of delivering an extra unit of food consumption as a fraction of the (preprogram) price of food. x again repre- sents the aimed increases in such consumption as a fraction of initial consumption. The new parameters included in such formulas are m, or the marginal propensity to spend on food, and ep, the price elasticity of the (excess) supply faced by the target group. Such a supply 2. If the new price line crosses the F* consumption level beneath point D, it means that the consumption of nonfood diminishes with the food price subsidy; that is, nonfood and food are gross substitutes. This can only be con- sistent with price elasticities for food larger than 1. 50 MALNUTRITION AND POVERTY Table 22. Fiscal Cost of Supplying an Extra Unit of Food to the Target Group, Expressed as a Ratio of the Price per Unit Program e,, 0c ep< oo Foodstamp 1 1 _ 1+ Price subsidy 1+X (1+X)( I + I Income transfer 1 rnL e,,] is equal to the total supply of food minus the demand by the (richer) nontarget group; correspondingly, the price elasticity of such excess supply depends not only on the price elasticity of total supply, f, but also on the price elasticity of demand of the nontarget group, ?R.3 When the programs have no effect on the supply price of food (EP= oo) some of the cost comparisons become rather straightfor- ward. The least costly intervention is the suboptimal food stamp program described earlier, because the marginal propensity to spend in food as well as the price elasticity for food are bound to be less than 1;4 the latter is particularly true if such programs are conceived for rather large aggregates of food commodities. Some illustrative estimates of the costs when the programs operate in an environment in which the additional food required raises the cost of food is pro- vided in Table 23. The total supply elasticity has only a minor effect, contrary to what was observed earlier in reference to a gen- eral food price subsidy program. The explanation, of course, is that only relatively small quantities of additional food are needed in target-group-oriented programs. Consumption of the nontarget group not only does not increase, but even decreases, when total supply is not infinitely elastic. For comparison, the cost of a general food price subsidy is shown in the last line of Table 23. These figures clearly illustrate that target-group-oriented programs are potentially much more cost 3. The expression for such elasticity is e,= e + where a is the proportion of total food consumed by the target group. 4. The value of x appears in the expression for the subsidy program, I+X as a result of evaluating the demand elasticity at initial values of prices and quantities. X drops out if evaluated at final magnitudes. For small changes, X-o, the expression tends to 1/n,,. Programs Oriented to Target Groups 51 Table 23. Cost Effectiveness of Target-Group-Oriented Programs and a General Food Price Subsidy Program with Different Demand and Supply Elasticities, as a Ratio of the Price per Unit 6= oo(ep= oo) e=1(ep=6.2) e=0.5(e,=3.7) Program 71p XD VP 1.0 0.5 1.0 0.5 1.0 0.5 Target group Food stamp 1.00 1.00 1.19 1.19 1.32 1,32 Price subsidy 1.20 2.40 1.39 2.59 1.52 2.72 Income transfer 2.00 2.00 2.39 2.19 2.65 2.32 General Price subsidy 5.45 11.40 7.85 15.28 10.25 19.25 a. The values of the other parameters are X = a = 0.2; m = 0.5; and 'iR = 0.3. effective and are probably the only feasible means to eliminate undernutrition in subgroups of the population. The difference in cost to the government is approximately between one to two times the price of the food for target-group-oriented programs and ten times that price for general price subsidies. FEASIBILITY AND COST OF ACTUAL PROGRAMS For target-group-oriented programs to be as effective as indicated in the previous sections, it must be possible to identify a homoge- neous population with the characteristics described. Furthermore, it must be possible to implement the programs in a way that the benefits reach only the intended population. In reality these condi- tions are unlikely to be met. The cost effectiveness will tend to be less than maximum and minimum food consumption targets may not be reached. Clearly, if the target group is fairly homogeneous with respect to income, its relative poverty is the decisive factor in making a choice between the programs. When the group is extremely poor, the price elasticity and the marginal propensity to consume food tend to approach 1, and the three programs are almost equally cost effective. Food stamps are more difficult to implement because participants would have to accumulate sufficient cash to purchase a month's or even a week's supply. If the target group's income and food consumption falls short only by a small fraction of adequate nutrition, the food stamp program might be twice as cost effective as the other programs. 52 MALNUTRITION AND POVERTY If the target population is heterogeneous, some additional con- siderations enter. A price subsidy program aimed at the average income level is bound to lead the better-off members of the target group to consume beyond minimal nutritional requirements and to leave the poorest inadequately fed. A food stamp program in which everyone is charged the same amount for the stamps may have the result of preventing those in greatest need from par- ticipating, whereas it is less than fully cost effective for those with higher incomes. To avoid neglecting poorer members of the target group, charges for food stamps will have to be set liberally, which means the program will be less cost effective. A study by the U.S. Department of Agriculture reports that, on the average, partici- pants of the U.S. Food Stamp Program spend 50 cents of each dollar of subsidy on food. In contrast, however, it is estimated that a direct income transfer would have provided only 20 cents worth of food consumption for every dollar.5 Still other factors must be considered when choosing among alternative programs. First, the foregoing comparisons of cost effec- tiveness do not take into account the value of additional nonfood expenditure induced by the transfer payment. For example, an income transfer might not be as cost effective as a food stamp pro- gram but would give the target population additional means for nonfood expenditure. Additional income might induce better health and lower fertility, both of which are complementary to food in reducing malnutrition. Second, no food assistance can be expected to be totally efficient in the sense that the subsidized food reaches only the target group. Participating beneficiaries may not reveal the full truth of their cir- cumstances and may receive unintended assistance. Middlemen and administrators may take illegal cuts and target-group beneficiaries may resell food intended to augment their own consumption. Thus, political and administrative problems may be as important as minor differences in cost effectiveness in determining the choice of an optimum program. Unless leakages are kept low, these inefficiencies will more than offset advantages perceived on the basis of idealized cost-effectiveness measures. 5. U.S., Department of Agriculture, Economic Research Service, "Bonus Food Stamps and Cash Income Supplements," Marketing Research Report no. 1034 (Washington, D.C., 1974). Appendix A Basic Country Data THIS APPENDIX comprises a single table whose data in- dicate the basic figures for population, GNP per capita, and per capita daily calorie consumption-all from 1965-used throughout the main body of the book. Table A.1. Population Distribution, per Capita Income, and per Capita Calorie Consumption, by Region and Country, 1965 GNP per capita ' Per capita Population (1972 daily calorie (thousands) U.S. dollars) consumption Country (1) (2) (3) Latin America Argentina 22,545 1,069 2,868 Bolivia 4,334 179 1,731 Brazil 80,766 370 2,541 Chile 8,708 721 2,523 Colombia 18,020 343 2,220 Costa Rica 1,490 482 2,234 Cuba 7,631 2,665 Dominican Republic 3,624 360 2,004 Ecuador 5,150 302 1,848 El Salvador 2,917 315 1,877 Guatemala 4,581 365 1,952 Guyana 655 360 2,291 Haiti 4,396 118 1,904 Honduras 2,182 286 1,930 Jamaica 1,791 480 2,243 Mexico 42,689 470 2,623 Nicaragua 1,745 428 2,253 Panama 1,246 648 2,317 Paraguay 2,030 272 2,732 Peru 11,650 480 2,255 53 54 MALNUTRITION AND POVERTY Table A.1 (continued) - GNP per capita a Per capita Population (1972 daily calorie (thousands) U.S. dollars) consumption Country (1) (2) (3) Puerto Rico 2,632 2,531 Surinam 338 566 2,371 Trinidad and Tobago 974 740 2,361 Uruguay 2,715 760 3,037 Venezuela 9,240 1,098 2,392 Total and average 244,049 2,471 Asia Burma 24,732 84 2,011 Sri Lanka 11,164 94 2,219 China, Republic of 12,962 301 2,379 Hong Kong 3,692 663 2,324 India 486,650 106 1,948 Indonesia 105,736 69 1,798 Khmer Republic 6,251 161 2,168 Korea, Republic of 28,377 180 2,421 Laos 2,631 104 2,005 Malaysia 9,422 352 2,255 Nepal 10,103 81 2,218 Pakistan 113,300 114 1,993 Philippines 32,345 188 1,895 Singapore 1,865 686 2,454 South Vietnam 16,124 178 2,134 Ceylon 11,164 94 2,219 Thailand 30,744 164 2,226 Total and average 869,098 1,984 Middle East Afghanistan 15,051 78 2,000 Cyprus 594 742 2,459 Iran 24,813 297 2,019 Iraq 8,140 328 2,055 Jordan 1,910 320 2,430 Lebanon 2,300 594 2,401 Libya 1,667 1,050 2,031 Saudi Arabia 4,500 349 2,077 Sudan 13,540 129 2,088 Syria 5,232 256 2,440 Turkey 31,151 274 2,858 United Arab Republic 29,389 229 2,421 Yemen, Arab Republic of 4,473 1,895 Yemen, People's Democratic Republic of 1.120 184 2,089 Total and average 143,880 2,315 Appendix A E Basic Country Data 55 Table A.1 (continued) GNP per capita a Per capita Population (1972 daily calorie (thousands) U.S. dollars) consumption Country (1) (2) (3) Africa Algeria 11,923 318 1,967 Angola 5,154 214 1,907 Benin 2,365 73 2,230 Burundi 3,210 62 2,017 Cameroon 5,229 153 2,264 Central Africa Republic 1,335 139 2,172 Chad 3,307 93 2,259 Congo, Democratic Republic of 15,627 86 2,036 Congo, People's Republic of 840 268 2,151 Ethiopia 22,699 73 2,152 Gabon 463 548 2,164 Gambia, The 330 118 2,335 Ghana 7,740 257 2,136 Guinea 3,510 2,075 Ivory Coast 4,200 271 2,433 Kenya 9,365 131 2,253 Liberia 1,076 198 2,287 Malagasy Republic 6,059 2,375 Malawi 3,908 72 2,257 Mali 4,480 69 2,159 Mauritania 1,050 154 1,981 Mauritius 761 286 2,343 Morocco 13,323 226 2,091 Mozambique 6,957 212 2,108 Niger 3,513 100 2,211 Nigeria 58,480 102 2,168 Rwanda 3,110 53 1,908 Senegal 3,490 234 2,348 Sierra Leone 2,367 173 2,185 Somali 2,500 72 1,778 Southern Rhodesia 4,480 213 2,551 Tanzania 11,674 96 2,170 Togo 1,638 123 2,232 Tunisia 4,451 269 2,153 Uganda 7,551 85 2,179 Upper Volta 4,708 71 2,060 Zambia 3,710 356 2,237 Total and average 246,583 2,154 Sources: columns (1) and (3), Food and Agriculture Organization of the United Nations, Agrichnltural Commodity Projects 1970-1980 (Rome, 1971); column (2), World Bank, Division of Economic and Social Data, Economic Analysis and Projections Department, average of 1964-66 period. Appendix B Income Distribution Data ESTIMATION OF THE SHARES of the regional population in predetermined (dollar) per capita income ranges was made in two steps. First, the shares of the population in given dollar per capita income ranges were estimated for several countries in each region. Second, the corresponding shares of the region were then com- puted as the weighted average of the countries' shares, where the weights were the shares of the population of each country in the total population of the countries included in the sample. It was further assumed that the distribution of the countries excluded could be approximated with the regional distribution obtained with the countries included. The first step is discussed below; the regional distributions are reported in Table 4 in the text. METHOD OF ESTIMATION To estimate a country's population share in each income range, it is necessary to know the density function of the income distribution for each country. One approach has been to fit a Pareto or log normal function to actual observed data. The shortcoming of such an approach is that these density functions usually do not provide good statistical fits to the data. An alternative is to find an equation of the Lorenz curve that would fit actual data reasonably well. This curve has been estimated with a new technique, developed by Kakwani and Podder, that introduces a new coordinate system for the Lorenz curve and appears to fit actual data quite well.' 1. N. Kakwani and N. Podder, "Efficient Estimation of the Lorenz Curve and Associated Inequality Measures from Grouped Observations," Econo- metrica, 44 (January 1976): 137-48. 56 Appendix B C Income Distribution Data 57 Figure B.1. A Typical Lorenz Curve Relation Fraction of families (F) Figure B.1 represents a typical Lorenz curve relationship de- scribing the percentage of families receiving specific fractions of total income. Families are ordered according to the per capita income of the family. A given point P(F,G) on the Lorenz curve defines the following coordinates: (1) 7r F+G and F-G (2) -F The Lorenz curve may be written in terms of these new coordinates as: (3) f 00, where (4) dq = -X d1; j+X' 58 MALNUTIITION AND POVERTY where , is the mean household per capita income and X is any given per capita income of a family.2 A particular specification of the equation 71=f(wr) is the form: (5) q =a 7r-(V2-r), where the difference between a and /8 reflects the skewness of the Lorenz curve toward each extreme. For this particular specifica- tion equation (4) becomes: (6) a a ar 1 ( VA2-7r) -, aara (V 2- 7)01= For estimation purposes, take logs of equation (5): (7) log q=a'+a log ir+p log (V+/2-7r). Defining Fi as the cumulative percentage of families at the ith per- centile ordered according to per capita household income, and Gi as the percentage of total income perceived by those Fi per- centage of families: (8) Fr= + G, and (9) F,-Gi The estimating equation therefore becomes: (10) log 7 ='+alog 7r,+/log (V2-,rt)+Mx. DATA AND RESULTS The basic income distribution data and data sources are presented in Tables B.1-B.4.3 Equation (10) was estimated for each country 2. Equation (4) is derived as follows: g(X)=density function of X, - F(X)= X g(X)dX=fraction of families below per capita income X, and G(X)= x f Xg(X)dX=fraction of total income of families below per capita income X. Equation (4) is then obtained by making use of the derivatives d7r/dF, d,,/dG, di,/dF, d,,/dG, F'(X), and G'(X). 3. The data were provided by Shail Jain of the Income Distribution Division, Development Research Center, the World Bank. See also Shail Jain, Size Dis- tribution of Income: A Compilation of Data (Washington, D.C.: World Bank, 1976). Appendix B E Income Distribution Data 59 using ordinary least squares. Table B.5 gives the values of the parameters a, a, and /3 for each country. The coefficient of de- termination (R2) of all regressions yielded values over 0.98. Substituting the values of the estimated parameters into equa- tion (6), and given the mean per capita income of the country (A), it is possible to solve for 7r, given X. The value of 7r can then be substituted into (5) so as to derive the corresponding value for ?. By the above procedure the values of F and G for any predeter- mined per capita income X are obtained. Basically we are inter- ested in F given X, the percentage of families with per capita household income below X. The results for thirty countries are reported in Tables B.6 and B.7. Table B.1 Africa: Income Distribution Data, Selected Countries, Various Years' (Cumulative percentages). Gabon, 1968 Chad, 1958 Income Ivory Coast, 1959 Malawi, 1969 Population Income recipients Income Population Income Households Income 60 35 10 1.4 60 30 10 2.3 80 57 20 3.3 80 45 20 5.8 90 70 30 5.8 90 60 30 10.0 95 77 40 8.8 95 80 40 14.9 100 100 50 12.0 100 100 50 20.9 60 16.7 60 28.2 70 23.5 70 36.8 80 32.5 80 46.8 90 45.5 90 60.8 95 54.9 95 70.8 100 100.0 100 100.0 Uganda, 1970 Senegal, 1960 Tanzania, 1967 Adult male Population Income Population Income employees Income 20 3 20 5.0 2.3 0.3 40 10 40 13.5 11.4 2.9 60 20 60 25.0 18.0 5.3 80 36 80 40.0 23.4 7.8 90 52 90 55.0 41.2 17.9 95 64 95 66.0 55.0 27.2 100 100 100 100.0 61.6 32.3 75.4 46.6 84.7 60.1 90.9 71.7 100.0 100.0 Sources: Chad, Gabon, Ivory Coast, and Senegal, Christian Morrisson, La repartition des revenus dans les pays du tiers-monde (Paris: Editions Cuias, 1969): Malawi, Christian Morrisson, "Special Paper on Malawi" (Washington, D.C.: World Bank, Income Distribution Division, 1973; unpublished memo- randum), Tanzania, 'Annual Economic Survey of 1968-69," in Christian Morrisson, "Special Paper on Tanzania" (Washington, D.C.: World Bank, Income Distribution Division, 1973; unpublished mnemorandum); Uganda, Ministry of Planning and Economic Development, Statistics Division, Enumeration of Employees, 1961-1970 (Entebbe: Uganda Government Printer, annual). a. The surveyed categories "Population," "Income recipients," "Households," and so on are variously defined in the sources. Table B.2 Asia: Income Distribution Data, Selected Countries, Various Years ° (Cumulative percentages) Hong Kong, 1971 India, 1964-65 Korea, 1970 Households Income Families Income Families Income 4.8 0.7 10 3.3 20 7 15.3 3.9 20 7.7 40 18 39.3 15.9 30 13.0 60 33 59.0 28.8 40 19.1 80 55 70.3 40.4 50 26.0 90 72 79.8 50.5 60 34.1 95 83 87.1 60.4 70 43.1 100 100 91.8 68.7 80 53.9 95.1 76.1 90 70.0 98.3 87.3 100 100.0 100.0 100.0 Malaysia, 1970 Pakistan, 1970-71 Philippines, 1965 Households Income Households Income Families Income 6.40 0.40 0.4 0.1 20 3.6 9.54 0.88 9.4 3.3 40 11.6 20.53 3.58 32.3 15.4 60 24.4 28.11 6.16 53.4 30.9 80 44.6 38.95 10.88 68.5 45.2 90 60.0 45.33 14.26 79.4 57.8 100 100.0 54.33 19.88 90.5 73.9 59.14 23.34 94.7 81.8 73.63 36.50 98.1 90.4 75.95 39.05 99.0 93.7 84.64 50.56 99.6 96.5 88.63 57.25 99.9 98.4 89.30 58.50 100.0 100.0 92.07 64.21 93.72 68.18 94.00 68.90 94.64 70.65 95.13 72.08 96.11 75.23 96.68 77.25 100.00 100.00 Sources: Hong Kong, Census and Statistics Department, Hong Kong Population and Housing Census 1971, Main Report (Hong Kong: Government Printer, 1973); India, K. R. Ranadive, in Pranab K. Bardhan, "The Pattem of Income Distribution in India: A Review" (Washington, D.C.: World Bank, Develop- ment Research Center 1973: processed); Korea, Bureau of Statistics, Economic Planning Board, Annual Report of the Family Income and Expenditure Survey. (Seoul, 1970) and Korea, Ministry of Agriculture and Forestry, Report on the Results of Farm Household Economy Survey and Production Cost Survey of Agricultural Products (Seoul, 1971); Malaysia, Department of Statistics, "Post Enumeration Survey of 1970 Census" (unpublished report); Pakistan, Ministry of Finance, Planning and Development, Statistical Division Household Income and Expenditure Survey 1970-71 (Karachi: Government Publications Division, n.d.); Phihppines, Department of Commerce and Industry, Bureau of Census and Statistics, "Family Income and Expenditure 1961, 1965, and 1971," BCS Survey of Households Bulletin, nos. 14, 22, and 34 (Manila, 1964, 1968, and 1973). a. The surveyed categories "Households" and "Families" are variously defined in the sources. Table B.3 Middle East: Income Distribution Data, Selected Countries, Various Years' (Cumulative percentages) Egypt, 1964-65 . Iraq, 1956 Libya, 1962 Sudan, 1963 Tunisia, 1970 Households Income Population Income Households Income Households Income Recipients Income 32.6 9.7 60 16 2.1 0.8 10 2.0 10 1.8 63.2 32.5 80 32 11.8 5.5 20 5.2 20 4.1 83.6 57.9 90 49. 26.0 13.9 30 9.5 30 7.2 91.8 72.2 95 66 48.6 30.3 40 14.2 40 11.4 100.0 100.0 100 100 66.7 47.5 50 20.3 50 16.6 81.6 65.3 60 28.0 60 23.4 89.9 77.6 70 37.2 70 32.4 96.2 88.9 80 49.2 80 45.0 100.0 100.0 90 66.8 90 62.4 100 100.0 100 100.0 Sources: Egypt, United Arab Republic Institute, Report on Employment Problems in Rural Areas (Cairo, 1968) and Mostafa H. Nagi, Labor Force and Employment in Egypt: A Demographic and Socioeconomic Analysis (New York: Praeger, 1971); Iraq, Christian Morisson, La Repartition des revenues dans les pays du tiers-monzde (Paris: Editions Cuias, 1969); Libya, Libyan Arab Republic, Ministry of National Economy, Central Statistics Office, Family Budget Survey in Tripoli Town, 1962 (Tripoli, 1963); Sudan, Department of Statistics, Omdurman Household Budget Survey (Khartoum, 1965); Tunisia, Christian Morrisson, "Special Paper on Tunisia," (unpublished memorandum) (Washington, D.C.: World Bank, Income Distribution Division, 1973) and Tunoisia, Premier Ministere, Institut National de la Statistique, La consummation et les depenses des menages en Tunisie, 1965-68 (Tunis, 1970). a. The surveyed categories "Households," "Population," and "Recipients" are variously defined in the sources. Table B.4 Latin America: Income Distribution Data, Selected Countries, Various Years a (Cumulative percentages) Brazil, 1970 Economically active Chile, 1968 Colombia, 1970 Costa Rica, 1971 population Income Families Income Population Income Families Income 10 1.17 10 2 20 1.5 10 2.1 20 3.49 20 4.5 40 9.4 20 5.4 30 6.91 30 8.5 60 22.0 30 9.6 40 11.56 40 13.0 80 40.5 40 14.7 50 17.71 50 18.9 95 67.5 50 20.9 60 25.37 60 25.7 100 100.0 60 28.4 70 34.78 70 33.8 70 37.7 80 45.63 80 43.2 80 49.4 90 60.32 90 58.0 90 65.6 95 72.29 95 69.6 95 77.2 100 100.00 100 100.0 100 100.0 Table B.4 (continued) El Salvador, 1965-67 Dominican Republic, Economically 1969 active Guatemala, 1966 Honduras, 1967-68 Families Income population Income Households Income Families Income 5.1 0.6 10 2.4 2.8 0.8 33.8 4.8 23.2 5.6 20 5.5 8.8 3.2 52.9 11.3 53.3 20.3 30 8.7 19.7 8.8 62.7 17.1 71.6 35.6 40 12.0 33.4 17.3 69.7 23.0 81.9 47.9 50 16.0 45.7 26.3 76.5 30.5 90.4 62.2 60 20.8 55.8 34.8 80.7 36.0 93.6 70.0 70 28.0 64.0 42.6 84.7 42.2 95.8 76.7 80 38.6 71.1 50.1 86.0 45.8 100.0 100.0 90 54.4 76.6 56.5 89.9 54.4 95 67.0 80.9 62.0 92.4 61.1 99 82.0 84.3 66.8 94.1 66.3 100 100.0 86.8 70.5 95.0 69.4 88.8 73.7 95.6 72.0 90.4 76.5 96.3 75.4 92.1 79.7 97.0 79.3 93.8 82.9 100.0 100.0 94.8 85.0 95.4 86.2 97.1 90.1 98.6 94.2 99.2 96.2 99.8 98.6 100.0 100.0 Peru, 1961 Panama, 1969 Economically Uruguay,1967 Jamaica, 1958 Active active Employed Households Income population Income population Income population Income 20 2.2 23.0 3.7 10 1.0 10 1.0 30 4.7 45.5 12.3 20 2.5 20 2.9 40 8.2 64.8 25.5 30 4.7 30 5.7 50 12.9 77.7 38.0 40 8.0 40 10.0 60 19.0 84.5 46.6 50 12.3 50 16.2 70 27.3 90.9 57.6 60 18.2 60 24.4 80 38.5 96.3 71.9 70 25.8 70 34.5 90 56.5 98.6 81.8 80 35.6 80 47.4 95 69.8 99.5 89.3 90 50.8 90 64.3 100 100.0 99.8 94.7 95 61.0 100 100.0 100.0 100.0 100 100.0 Sources: Brazil, Carlos G. Langoni, "Distribuicao da Renda e Desenvolvimento Economico de Brasil (Rio de Janeiro: Editora Expressao e Cultura, 1973); Chile, Direcci6n General de Estadistica y Censos, Encuesta Nacional sobre Ingresos Familiares, Serie de Investigaciones Muestrales (Santiago, June 1969); Colombia, Departamento Administrativo Nacional de Estadistica, Encuesta de Hogares de Prop6sitos Multiples: Encuesta de Hogares 1970 (Bogota, June, 1971) and Encuesta de Hogares 1970: Andlists de Ingresos (Bogota, February, 1971); Costa Rica, Universidad de Costa Rica, "La Distribuci6n del Ingreso y del Consunio de Algunos Alinentos" (San Jose, September 1972); Dominican Republic, Banco Central, Oficina Nacional de Estadisticas-Agencia Inter- nacional para el Desarrollo, Esiudio sobre Presupuestos Familiares, vol. 1: Ingresos y las familias en la ciudad de Santo Domingo, 1969 (Santo Domingo, 1961); El Salvador, United Nations Economic Commission for Latin America, Economic Survey of Latin America, 1968 part one (E/CN. 12/825 Rev. 1) (New York: United Nations, 1969); Guatemala, R. A. Orellana Gonzales, Encuesta sobre Ingresos y Gastos de la Familia del Cam- pesino Asalarido de Guatemala (Guatemala City, 1966); Honduras, Direcci6n de Estadisticas y Censos, Encuesta de Ingresos y Gastos Familiares 1967-68 (Tegucigalpa, April, 1970); Jamaica, E. Abiram, "Income Distribution in Jamaica, 1958," Social and Economic Studies, vol. 13, no. 3 (September 1964), pp. 333-70 (Jamaica: University of the West Indies, Institute of Social and Economic Research); Panama, Charles C. McLure, Jr., "The Distribution of Income and Tax Incidence in Panama, 1969," Working Paper no. 36 (Houston, Tex.: Rice University, Program of Development Studies); Peru, Richard Webb, "The Distribution of Income in Peru," Discussion Paper no. 26 (Princeton, N.J.: Woodrow Wilson School, Princeton University, 1972); Uruguay, Uni- versidad de la Rep6blica, Instituto de Economia, La Distribucion del Ingreso en Uruguay, Documento de Referencia no. 6 (Santiago: Facultad de Ciencias Econ6micas y de Administraci6n, 1971). a. The surveyed categories "Economically active population," "Families," "Population," and so on are defined variously in the sources. 68 MALNUTRITION AND POVERTY Table B.5. Estimates of the Parameters of the Lorenz Curve Region and country a a P Asia Hong Kong 0.386 0.879 0.786 India 0.351 0.871 0.663 Korea, Republic of 0.366 0.949 0.901 Malaysia 0.485 0.899 0.849 Pakistan 0.294 0.887 0.770 Philippines 0.465 0.875 0.807 Africa Chad 0.315 0.877 0.700 Gabon 0.679 1.002 0.853 Ivory Coast 0.372 0.893 0.614 Malawi 0.411 0.912 0.715 Senegal 0.568 0.969 0.850 Tanzania 0.457 0.945 0.756 Uganda 0.369 0.888 0.829 Middle East Iraq 0.697 1.047 1.042 Libya 0.249 0.919 0.821 Sudan 0.437 0.927 0.913 Tunisia 0.533 1.006 0.999 United Arab Republic 0.413 0.875 0.908 Latin America Brazil 0.576 0.978 0.910 Chile 0.465 0.939 0.780 Colombia 0.557 0.983 0.904 Costa Rica 0.421 0.921 0.850 Dominican Republic 0.472 0.929 0.867 El Salvador 0.536 0.950 0.921 Guatemala 0.270 0.875 0.799 Honduras 0.661 0.993 1.021 Jamaica 0.585 0.949 0.960 Panama 0.533 0.925 0.868 Peru 0.606 0.978 0.886 Uruguay 0.394 0.834 0.883 Source: Calculated from data in Tables B.1.-B.4. Table B.6. Estimates of the Country Distribution of Population by per Capita Income Ranges: Africa, Asia, and Middle East (Percentages) Per capita income in 1965 (1972 U.S. dollars) Country 0-50 50-100 100-150 150-200 200-250 250-300 300-350 350+ Africa Chad 25.0 47.6 16.8 5.9 2.4 1.1 0.6 0.6 Gabon 9.3 15.9 12.2 9.6 7.7 6.2 5.1 34.0 Ivory Coast 1.0 11.8 24.5 20.1 13.6 9.0 6.0 14.0 Malawi 53.6 31.4 9.0 3.3 1.4 0.7 0.3 0.3 Senegal 17.6 25.1 16.2 10.7 7.3 5.2 3.7 14.2 Tanzania 42.2 33.2 12.6 5.6 2.8 1.5 0.9 1.2 Uganda 22.7 38.0 18.9 9.3 4.8 2.5 1.4 2.4 Asia Hong Kong 0.6 1.2 2.9 5.4 7.3 8.0 7.9 66.7 India 23.4 45.2 18.1 7.1 3.1 1.5 0.9 0.7 Korea, Republic of 2.2 30.3 23.0 14.5 9.4 6.3 4.3 10.0 Malaysia 6.1 13.4 14.5 12.3 9.8 7.7 6.1 30.1 Pakistan 10.4 46.0 24.0 10.4 4.6 2.1 1.0 1.5 Philippines 16.3 25.9 18.7 12.1 7.9 5.2 3.6 10.3 Middle East Iraq 20.0 10.0 21.7 8.7 6.4 4.9 3.9 24.4 Libya 0 0 0 0 0.01 0.1 0.2 97.6 Sudan 24.0 31.6 17.2 9.7 5.9 3.7 2.4 5.5 Tunisia 11.0 20.0 16.0 11.0 8.0 6.0 5.0 23.0 United Arab Republic 7.8 19.0 18.1 13.7 10.0 7.3 5.5 18.6 Table B.7. Estimates of the Country Distribution of Population by per Capita Income Ranges: Latin America (Percentages) Per capita income in 1965 (1972 U.S. dollars) 50- 100- 150- 200- 250- 300- 350- 400- 500- 600- 800- Country 0-S0 100 150 200 250 300 350 400 500 600 800 1,000 1,000+ Brazil 7.3 18.5 14.5 10.9 8.4 6.5 5.2 4.1 6.1 4.2 5.2 2.9 6.2 Chile 0.1 1.0 .4.4 8.0 9.0 8.5 7.7 6.8 11.1 8.5 11.7 7.2 16.0 Colombia 5,4 20.5 15.6 11.5 8.6 6.6 5.1 4.1 5.9 4.0 4.8 2.7 5.2 Costa Rica 0.4 3.4 9.5 11.6 10.8 9.3 7.9 6.6 10.2 7.3 9.1 5.1 8.8 Dominican Re- public 2.8 13.2 15.3 12.8 10.1 8.0 6.3 5.0 7.4 5.0 5.9 3.1 5.1 El Salvador 8.9 18.5 15.1 11.3 8.5 6.6 5.1 4.0 5.9 4.0 4.7 2.6 4.8 Guatemala 0.1 0.8 5.1 13.1 15.6 14.0 11.4 9.1 12.6 7.4 6.8 2.3 1.7 Honduras 23.0 18.0 12.4 8.8 6.5 5.0 3.9 3.1 4.5 3.1 3.8 2.2 5.6 Jamaica 5.9 9.0 9.2 8.3 7.2 6.3 5.4 4.7 7.7 5.9 8.4 5.5 16.5 Panama 3.2 6.4 8.6 8.7 8.0 7.1 6.2 5.4 8.9 6.8 9.4 6.0 15.4 Peru 6.1 15.2 12.9 10.3 8.2 6.7 5.4 4.5 6.9 3.6 8.9 7.3 4.0 Uruguay 2.2 1.9 3.0 4.1 5.0 5.5 5.7 5.6 10.4 9.0 13.9 9.7 24.0 Appendix C Estimated Calorie Consumption Functions Based on Cross-country Data IT IS GENERALLY OBSERVED that income elasticities decline as income increases. For instance, in Table C.1 are the calorie elas- ticities implied by FAO projections of food consumption, which in turn are based on an examination of a large number of food demand studies. In theory, many functional forms could be used to specify a calorie-income equation that yields declining calorie elasticities as income rises. In regressing country per capita calorie consump- tion and income data, several such functional fornqs of the equation have been estimated, as well as the constant elasticity equation. The estimated functions and corresponding elasticity equations are as follows: Semilog: C=a+bLnX tt=b Inverse: C=a+b(1/X) /A=-b/C X Double log: LnC=a+bLnX s =b Loginverse: LnC=a+b(1/X) ,a=-b/X The statistical results as well as the implied elasticities and daily calorie consumption at two extreme levels of income are shown in Table C.2. Whereas on statistical goodness-of-fit criteria there is little basis for choosing between the functional forms, the estimates of elastici- ties and consumption levels at extreme levels of income show that some functional forms are more plausible than others. The esti- 71 72 MALNUTRITION AND POVERTY mated parameters in the inverse and log-inverse function, for instance, imply that calorie consumption does not exceed 2,600 calories at any income level in all regions. This does not seem plausible. Furthermore, the inverse functions imply too low con- sumption levels at a low per capita income level in Latin America; in fact, the inverse function implies negative consumption. Given these considerations of plausibility and the a priori preference for a function that specifies declining elasticity as income rises, the semi- log form was chosen as providing the best representation of the calorie consumption function. Table C.1. Average per Capita Private Consumption Expenditure and Income Elasticities of Demand for Food, Selected Regions and Countries, 1964-66 Per capita private consumption expenditure Income elasticities of (U.S. doUars, demand for all food Geographic constant 1972 expressed in areas/countries prices) Calories Total proteins South Asia 75 0.39 0.37 East Africa 78 0.32 0.36 Central Africa 81 0.31 0.21 West Africa 94 0.32 0.41 East and Southeast Asia 119 0.23 0.30 Northwest Africa 158 0.40 0.35 Middle East 183 0.15 0.15 Latin America 342 0.19 0.19 Eastern Europe 553 0.02 0.09 Japan 603 0.13 0.19 Israel 942 0.06 0.13 Western Europe 1,050 0.06 0.11 North America 2,611 -0.01 0.00 Source: Compiled from Table 8 and Table B in Part II, Statistical Appendix of Volume 11, FAO, Agricultural Commodity Projections 1970-1980 (Rome, 1971). Table C.2. Cross-country Data, 1965, Based on Regressions of Daily per Capita Calorie Consumption (C) on per Capita Income (X) Region Elasticity Calorie consumption and form of equation Equation R 2 X=US$25 X=US$2,000 X=US$25 X=US$2,000 Latin America Semilog C=256+417 LnX 0.44 0.260 0.122 1,598 3,425 (4.1) Inverse C=2,594-117,979(1/X) 0.34 2.221 0.023 <0 2,535 (-3.3) Double log LnC=6.58+0.19 LnX 0.47 0.190 0.190 1,335 3,050 (4.3) Log-inverse LnC=7.86-54(1/X) 0.37 2.160 0.027 300 2,520 (-3.5) Asia Semilog C= 1,191 + 188 LnX 0.46 0.105 0.07 1,796 2,620 (3.5) Inverse C=2,390-32,900(1/X) 0.45 1.224 0.01 1,075 2,375 (-3.4) Double log LnC=7.22+.09 LnX 0.45 0.090 0.09 1,825 2,708 (3.4) Log-inverse LnC=7.78-15.4(1/X) 0.45 0.616 0.01 1,292 2,373 (-3.4) Table C.2 (continued) Region Elasticity Calorie consumption and form of equation Equation a R2 X=US$25 X=US$2,000 X=US$25 X=US$2,000 Middle East Semilog C=1,857+71 LnX 0.04 0.034 0.030 2,082 2,394 (0.6) Inverse C=2,368-26,080(1/X) 0.10 0.787 0.006 1,325 2,355 (-1.1) Double log LnC=7.53-.03 LnX 0.04 0.030 0.030 2,050 2,340 (0.7) Log-inverse LnC=7.77-11.6(1/X) 0.11 0.465 0.006 1,490 2,355 (-1.2) Africa Semilog C=1.700+96 LnX 0.15 0.048 0.040 2,009 2,424 (2.4) Inverse C=2,298-14,736(1/X) 0.20 0.345 0.003 1,710 2,290 (-2.9) Double log LnC=7.46+.05 LnX 0.15 0.050 0.050 2,040 2,540 (2.4) Log-inverse LnC=7.74-6.89(1/X) 0.21 0.276 0.003 1,745 2,290 (-1 .9) a. Figures in parentheses are t-statisties. czi C c_ I f3O.'iq + 1 f. Is.hX ' 3 ~SS96002 C 4 2 vs 6,~~~~} V ~~~ { / p*)~ 4a °y f1.- b Appendix D Projected Calorie Deficits THE CALCULATIONS discussed in Chapter 2 are amplified here. Population projections by income groups in each region appear in Table D.1. Estimated calorie deficits with alternative consumption functions are given in Table D.2. Daily per capita calorie deficit estimates in the affected population by region, for 1975 and 1990, are shown in Table D.3. 75 Table D.1. Projected Population by Income Group and Region, 1965, 1975, and 1990 (Millions) Latin America Asia Middle East Africa Total population Income group 1965 1975 1990 1965 1975 1990 1965 1975 1990 1965 1975 1990 1965 1975 1990 I 16 21 30 176 221 311 19 24 35 73 93 135 284 359 511 II 39 50 72 387 486 683 29 37 54 78 100 145 533 673 954 III 32 41 59 173 217 305 26 33 48 39 50 72 270 341 484 IV 26 33 48 74 93 131 16 21 30 21 27 39 137 174 248 V 21 27 39 36 45 64 12 15 22 12 15 22 81 102 147 VI 17 22 32 19 24 34 8 10 15 7 9 13 51 65 94 VII 14 18 26 12 15 21 6 8 11 5 6 9 37 47 67 VIII 79 101 146 19 24 34 27 35 50 11 14 20 136 174 250 Total 244 313 452 896 1,125 1,583 143 183 265 246 314 455 1,529 1,935 2,755 Source: United Nations Population Division, "Trends and Prospects in Urban and Rural Populations" (New 'York, April 1975). Appendix D 0 Projected Calorie Deficits 77 Table D.2. Estimated Daily Calorie Deficit, by Income Group and Region, 1965 Calorie income Calorie income elasticity= 0.15 elasticity=0.30 Total Total Region and (thousand (thousand income group Per capita millions) Per capita millions) Latin America I 643 10.3 1,365 22.0 II 369 14.4 818 31.9 III 186 5.9 452 .14.3 IV 65 1.7 211 5.5 Asia I 539 94.6 850 149.3 II 286 110.6 345 133.6 III 116 20.1 0 0 Middle East I 687 13.2 1,236 23.8 II 406 11.9 676 19.8 III 218 5.6 300 7.6 IV 94 1.6 53 0.9 Africa I 521 38.1 845 61.8 II 252 19.7 308 24.1 III 72 2.8 0 0 78 MALNUTRITION AND POVERTY Table D.3. Estimated Daily per Capita Calorie Deficit in Affected Population, by Income Group and Region, 1975 and 1990 Daily calorie deficit by year and projection alternative Region and 1965 1975 1990 income group A B C A B C Latin America I 643 571 535 489 463 374 245 II 369 297 261 208 189 100 0 III 186 114 78 20 0 0 0 IV 65 0 0 0 0 0 0 Asia I 539 506 473 429 456 373 255 II 286 253 220 170 203 120 0 III 116 83 50 0 33 0 0 Middle East I 687 577 467 417 411 135 0 II 406 296 186 129 130 0 0 III 218 108 0 0 0 0 0 IV 94 0 0 0 0 0 0 Africa I 521 486 451 404 433 355 227 II 252 217 182 128 167 86 0 III 72 37 0 0 0 0 0 Appendix E Cost of Programs Oriented to Target Groups ASSUME THAT Figure E.1 represents the food market rele- vant for the target group; Dp represents the demand by the target (poorer) group as a function of initial income Y; and S, represents the (excess) supply faced by the group. That supply is equal to the total supply, ST, minus the demand by the nontarget (richer) group, DR. Initial consumption and price is FO and PO,, respectively. The object is to induce an increase in food consumption equal to aF: that is, the goal is an increase equal to k=AF/F,. Some basic expressions are first derived that later will be used in computing the cost of these alternative options. -The change (increase) in price necessary to induce an increment in supply equal to AF. Such an increase is equal to A1. Denoting by e, the elasticity of supply faced by the target group, the following expression may be defined: (1) PA fp -The change (decline) in price required to induce the target group to increase consumption by A. That change is equal to A2 Denoting by q, the price elasticity of demand of the target group, the following expression may be defined (in absolute value): (2) A2 A Po 17P -The change (decline) in consumption because of the increase in price defined in equation (1), A,. Such a decline is equal to A3. 79 80 MALNUTRITION AND POVERTY Figure E.I. Analysis of the Food Market Relevant for the Target Group Price SP(P)=Sr(P)-DR(P) F. Food Therefore, the following expression may be written (in absolute value): (3) 3=A COST OF AN INCOME TRANSFER What is the value of AY-the income transfer-required to induce an increase in consumption by a fraction x? Denoting as m the marginal propensity to spend in food: (4) maY=PP( AF + A,4) and (5) AY= m (AF+ A). m Appendix E [1 Cost of Programs Oriented to Target Groups 81 Multiplying and dividing by F1 = Fo + AF: (6) Ay= P (F,+AF)FAF + i1. m LF + aFF 4_A3 For small changes, -F - F so substituting from equation (3) and dividing by AF gives: (7) CT= AY = [l+(l+A) e ]' TAF ml + IepIj CT represents the cost of the income transfer per unit increment in food consumption induced by that transfer. COST OF A PRICE SUBSIDY If it is possible to subsidize the consumption of the target group- without subsidizing the nontarget group-the cost of the subsidy is equal to: (8) (& +AF)(a,+ A). Substituting from equations (1) and (2): (9) (F0+AF) A PF-+- 1 Iep 17p Dividing by AF: (10) Cs=P(1+x) [I +-]. C, represents the fiscal cost of the subsidy per unit increment in food consumption by the target group. COST OF A FooD STAMP PROGRAM The suboptimal food stamp program described in the text is de- signed in such a way that the value of the additional food con- sumption is equal to the fiscal cost (transfer) of the program. The question is: What is the value of that transfer required to induce an increment in physical consumption equal to AF? The value of the transfer must be able to finance the increment AF-valued at the new supply price of the commodity-as well as 82 MALNUTRITION AND POVERTY to finance the increased cost of old consumption. The transfer be- comes equal to: (11) AF(P0+a,) +F A, or (12) po A1 (Fo + AF) + P, AF. Substituting A1 from (1): (13) PO(-)(F0+AF) + PO AF. fp Dividing by AF gives: (14) Ct= Po[ 1+ +1 CUt represents the cost of the program per unit increment in food consumption by the target group. Published by The Johns Hopkins University Press Baltimore and London EXPORT SALES REPRESENTATIVES THE UNITED KINGDOM, CONTINENTAL EUROPE, 1THE NEAR EAST AND MIDDLE EAST, AFRICA The Johns Hopkins University Press, Ltd., 2-4 Brook Street, London WIY-IAA, England CANADA Burns & MacEachern Ltd., 62 Rallside Road, Don Mills, Ontario M3A I A6, Canada CENTRAI. AMERICA, SOUTH AMERICA, CARIBBEAN (EXCLUSIVE OF MEXICO, INCLUDING PUER- TO RICO), INDIA, PAKISTAN Kaiman & Polon, Inc. 456 Sylvan Avenue, Englewood Cliffs, New Jersey 07632, U.S.A. MEXICO Centro Interamericano de Librcs Academicos (CILA), 31 Sullivan-Bis, Mexico 4, D.F. AUSTRALIA, NEW ZEALAND, AND SOUTHEAST ASIA Australia and New Zealand Book Co. Pty. Ltd., P.O. Box 459, Brookvale, N.S.W. 2100, Australia JAPAN United Publishers Services Ltd., Shimura Building, 4-1 Kojimachi, Chiyoda-ku, Tokyo, Japan THROUGHOUT REST OF THE WORLD, orders can be sent directly to The Johns Hopkins Uni- versity Press, Baltimore, Maryland 21218, U.S.A. THE WORLD BANK Headquarters European Office Tokyo Office 1818 H Street, N W 66, avenue d'[ena Kokusai Building Washington, D.C 20433 75116 Paris 1-1 Marunouchi 3-chome U.S.A. France Ch,yoda-ku, Tokyo 100 Japan 1m~ Errua - fn a~Q?a0a.fl8~a~8aflumild t~ ~ 11 1 a- I9 Q - 8~l11 ~I C$3 @oWra 0 03 Na ^ a a 0 J8fX ~~ ~ T%mm wakwow W S3 h0 bti Q1QG@a'ca -eI * @ 7f~dWw Am Ra(g F%%=B 3Qm &IIlDtQD mgUl Wm a °aa dmo9li @g $N qw aw Q liQb aO1 i a a iQD f ft - 31 $m&= IRQDQ&FJ ad NlemB a $ aa3[t 0xa 3m ft Dmbpmma fRaxd 0§fw0aftra mm nMmuNa I na ft ffaw Una MSgQmffA a aQ APN WYN Ng &