wPS 2 +31- POLICY RESEARCH WORKING PAPER 2434 Socioeconomic Inequalities Despite the development community's shift in emphasis in Child Malnutrition in toward the poor, the Developing World malnutrition, like other dimensions of poor health, is concentrated among the Adam Wagstaff worst off. Yet targets are still Naoko Watanabe defined in terms of population averages. Consider, then, this information about malnutrition rates among different economic groups in 20 developing countries. The World Bank Development Research Group Poverty and Human Resources and Human Development Network Health, Nutrition, and Population Team September 2000 H POLICY RESEARCH WORKING PAPER 2434 Summary findings Among the conclusions Wagstaff and Watanabe reach the most equal distributions of malnutrition, and about malnutrition rates among different economic Nicaragua, Peru, and, to a lesser extent, Morocco have groups: highly unequal distributions. * Inequalities in malnutrition almost always disfavor * Some countries (such as Egypt and Romania) do the poor. well in terms of both the average (the prevalence of * It's not just that the poor have higher rates of malnutrition) and the distribution (equality). Others do malnutrition. The rate of malnutrition declines badly on both counts. Peru, for example, has a higher continuously with rising living standards. average level of stunting than Egypt and higher poor- * The tendency of poorer children to have higher nonpoor inequality. But many countries do well on one rates of stunting and underweight is not due to chance or count and badly on the other. Brazil, for example, has a sampling variability. Inequalities in stunting and far lower (less than 20 percent) stunting rate overall than underweight, as measured by the concentration index, Bangladesh (more than 50 percent) but has four times as are statistically significant in almost all countries. much inequality (as measured by the concentration * Inequalities in underweight tend to be larger than index). inequalities in stunting, which tend to be larger than * Use of an achievement index that captures both the inequalities in wasting. average level and the inequality of malnutrition leads to * In most cases, whatever the malnutrition indicator, some interesting rank reversals in the country league differences in inequality between countries are not table. With stunting, for example, focusing on the statistically significant. achievement index moves Egypt (a low-inequality * Even if attention is restricted to the cross-country country) from sixth position to fourth, higher than Brazil differences in inequality that are statistically significant, and Russia (two countries with high inequality). interesting conclusions emerge. Egypt and Vietnam have This paper-a product of Poverty and Human Resources, Development Research Group, and the Health, Nutrition, and Population Team, Human Development Network-is part of a larger effort in the Bank to investigate the links between health and poverty. The study was funded by the Bank's Research Support Budget under the research project "Inequalities of Child Health: Comparing the LSMS and DHS" (RPO 683-47). Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Anna Marafion, room MC3-558, telephone 202-473- 8009, fax 202-522-1153, email address amaranon@worldbank.org. Policy Research Working Papers are also posted on the Web at www.worldbank.org/research/workingpapers. The authors may be contacted at awagstaff@worldbank.org or nwatanabe@worldbank.org. September 2000. (36 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center Socioeconomic Inequalities in Child Malnutrition in the Developing World Adam Wagstaff Naoko Watanabe The findings, interpretations and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. Without wishing to incriminate them in any way, we would like to acknowledge the helpful comments of Peter Lanjouw and Milla McLachlan on an earlier draft of this paper. The work towards the paper was supported by a grant from the World Bank's Research Support Board (grant # 683-47), as well as by the Bank's Visiting Fellow Program and the Japanese Trust Fund. 1 I. Introduction Major progress has been made over the last 30 years in reducing the prevalence of malnutrition amongst children. Between 1970 and 2000 the proportion of malnourished children was reduced by 20 % in developing countries [1, 2]. Despite this, approximately 27% of children under the age of five in developing countries are still malnourished [1]. Malnutrition currently claims about the half of the 10 million deaths each year among under-five children in the developing world, and is the risk factor responsible for the greatest loss of DALYs'globally, accounting for 16% of total DALYs [3]. Furthermore, malnutrition is highly associated with poverty-levels of malnutrition are higher in poor countries than in better-off countries. In low income countries 36 % of children are malnourished compared with 12% and 1% in middle income countries and the United States, respectively [4]. There is also growing evidence (see below) that, within countries, the poor suffer from higher rates of malnutrition than the nonpoor. This has led for calls for the focus to be on reducing levels of malnutrition amongst the poor [5- 7]. And yet the goals and targets of international development and bilateral aid agencies continue to be couched in terms of improving population averages [8]. For example, the only nutrition- focused target of the OECD's Development Assistance Committee (DAC) is couched in terms of population averages-to reduce the proportion of children under-five who are underweight by one half between 1990 and 2015 [9]. Part of the reason for this gap between rhetoric and reality is lack of data-until recently, only patchy data existed on poor-nonpoor differences in malnutrition rates. This is now changing. Two recent studies of malnutrition in Africa [10, 11] documented the gaps in malnutrition across consumption groups in that continent, while a new dataset [121 allows comparisons to be made across 48 countries of malnutrition differentials across quintiles of "assets". This paper contributes to this growing empirical literature on poor-nonpoor inequalities in child malnutrition. It differs from the aforementioned studies in a number of respects. First, by using concentration indices and corresponding standard error estimates, it attempts to draw systematic conclusions about cross-country differences in malnutrition inequalities. In this respect, the paper is the analog for child malnutrition of the analysis of inequalities in infant and under-five mortality reported in Wagstaff [13]. The results in the present paper suggest that several countries have high inequalities on more than one indicator of malnutrition, while several have low levels of inequality. We also find, however, that in many of the pairwise comparisons between countries, the differences in inequality are not statistically significant. Second, we present some evidence-albeit very limited-on the issue of why some countries have higher levels of inequality in malnutrition than others. Specifically, we explore the possibility that inter-country differences in inequalities in malnutrition reflect inter-country differences in inequalities in consumption. Third, we emphasize the evidence on both inequalities and population averages. We find some countries that do well on both dimensions, but find many that perform well on one dimension but badly on the other. We therefore go on to compute values of a summary index that captures how well a country does both on its average rate of malnutrition and on its inequality in malnutrition. We find some swapping of positions- 2 countries that do well in terms of average malnutrition slip down the "league table" when account is taken of the degree of inequality between poor and nonpoor children. The plan of the paper is as follows. Section II outlines the methods we use for measuring and testing for inequalities in malnutrition. Section III sets out the data and variable definitions we employ. In doing so, we emphasize the interrelationships between the three measures of malnutrition we use (underweight, stunting and wasting) and compare our overall average rates of malnutrition with those reported elsewhere. In Section IV we report malnutrition rates by consumption quintile. We also report the values of the inequality index we employ (the concentration index) and go on to test for significant differences across countries. In Section III, we also explore the sensitivity of our results to changes in definitions. Section V presents a brief analysis of the extent to which inequality in malnutrition at the country level are associated with inequalities in consumption. In short, are countries that have high inequalities in household consumption also the countries that have high inequalities in malnutrition between poor and less poor children? Finally, in Section VI, we compare average malnutrition rates with inequalities. We show scatter plots to allow easy identification of countries that do well on both counts, as well as an index that allows one to compare the "achievement level" of countries that do well on, say, average malnutrition but badly on inequality in malnutrition with the achievement level of countries that do well on inequality but badly in terms of average malnutrition. Section VII contains our conclusions. II. Measuring and Testing for Inequalities in Malnutrition In this section, we outline the index we use for measuring inequalities in malnutrition amongst children, and the methods used to estimate standard errors for the index. THE CONCENTRATION INDEX We measure a household's living standards by equivalent consumption, but the method outlined below can be used for any socioeconomic ranking variable. We rank children by their household's equivalent consumption. We also have a variable indicating whether or not a given child is malnourished. The curve labeled L(p) in Figure 1 is a malnutrition concentration curve. It plots the cumulative proportion of malnourished children (on the y-axis) against the cumulative proportion of children (on the x-axis), ranked by equivalent household consumption, beginning with the most disadvantaged child. The similarity with the Lorenz curve is obvious. However, one should bear in mind that here we are not ranking by the variable whose distribution we are investigating-we are looking here at the distribution of malnutrition, not by levels of malnutrition, but rather by equivalent consumption. If L(p) coincides with the diagonal, all children, irrespective of their household consumption, suffer from the same malnutrition rates. If, as is more likely, L(p) lies above the diagonal, inequalities in malnutrition favor the better-off children; we will call such inequalities prorich. If L(p) lies below the diagonal, we have propoor inequalities in malnutrition (inequalities to the disadvantage of the better-off). The further L(p) lies from the diagonal, the greater the degree of inequality in malnutrition across quintiles of living standards. If L(p) of country X is everywhere closer to the diagonal than that of country Y, then country Xs 3 concentration curve is said to dominate that of country Y. It seems reasonable in such cases to conclude that there is unambiguously less inequality in malnutrition in country X than in country Y. Fig 1: Malnutrition concentration curve 100 cumul. % of children, ranked by economic status Where concentration curves cross, or where, in any case, one wants a numerical measure of inequality in malnutrition, one can use the concentration index, denoted below by C and defined as twice the area between L(p) and the diagonal. This index, as has been shown elsewhere [14, 15], is related to the relative index of inequality (RII), used extensively by epidemiologists and others in analyses of socioeconomic inequalities in health and mortality [16- 20].' C takes a value of zero when L(p) coincides with the diagonal and is negative (positive) when L(p) lies above (below) the diagonal. It can be calculated as (1) C =2E x - l , where n is the sample size, xi is the malnutrition indicator for child i, ju = (1 / n)n x1 is the mean level of malnutrition and R, is the relative rank in the consumption distribution of the ith child (the best-off child having a value of R of 1). Alternatively, and more simply [15], the value of C can be obtained from the following "convenient regression" (2) 2a 2[xi / ,u] = Y, +51 .R, +ui This measure, like the Gini coefficient (the analog in the case where individuals are ranked by the variable whose inequality is being measured), implicitly assumes a particular set of value judgements about where inequality matters most. This issue is explored in Wagstaff [ 21]. 4 where CR is the variance of R. The estimator of 3, is equal to 2 which, from eqn (1), shows that 61 is equal to C. Readers familiar with the RII will note that eqn (2) is essentially the same as the regression equation used to compute the RII, the difference being that the RII is typically computed using grouped rather than individual-level data. The division of the LHS through by p simply means that the coefficient 81 is the RII rather than the Slope Index of Inequality (SII) [18]. The only difference, then, between eqn (2) and the-equation used to compute the RuI is that the LHS contains the variance of the rank variable. This, however, approaches 1/12 as the sample size grows, and can therefore be treated approximately as a constant across samples. Thus the RII and C ought to rank distributions the same-there is little to choose between the two measurement approaches, though the concentration curve has the attraction of facilitating graphical comparisons of malnutrition inequalities. STATISTICAL INFERENCE When undertaking cross-country (or temporal) comparisons, one needs to bear in mind that the malnutrition data are derived from survey data and are hence subject to sampling variation. It is useful, therefore, to couple numerical comparisons of the index C with statistical tests to assess the statistical significance of any inter-country (or temporal) differences. An attraction of the convenient regression-eqn (2) above-is that it provides a standard error for the concentration index C. This standard error is not, however, wholly accurate, since the observations in each regression equation are not independent of one another due to the nature of the R, variable. The following standard error estimator, derived by Kakwani et al. [15], takes into account the serial correlation in the data: (5) var(C) =-|, (1I /n)ai_(l+)2] where (6) ai, p (2Ri - 1 - C) + 2 - qi-, - q, , and (7) q, = N = M< being the ordinate of L(s), with qo = 0. It is this estimator that is used, rather than that in eqn (2), which is used in section IV. 5 III. Data and Variable Definitions The surveys used are listed in Table 1. Surveys are nationally representative except for Bangladesh, Brazil, China, Guatemala, Indonesia and the Philippines whose coverage is regional. Survey years range from 1987 to 1997. In selecting countries and surveys, we wanted to achieve a degree of geographic heterogeneity. In addition, however, the surveys included also needed to include (i) data on consumption or income, (ii) anthropometric data to measure malnutrition (height, weight, age and gender of children under 5), and (iii) an acceptably large sample size of children. There are 20 countries and surveys in total, of which 11 are Living Standards Measurement Study (LSMS) surveys, 9 are similar multi-topic surveys, which satisfy the above criteria. However, we need to keep in mind that quality standard, sampling method and variable definitions may not be uniform across surveys, especially when conducting agencies are different. MEASURING LIVING STANDARDS As the main focus of this survey is to see the poor-nonpoor inequalities in malnutrition, choosing a variable according to which households are ranked is of paramount importance. Consumption is usually considered a better indicator of living standards than income, as ranking by the latter is generally more unstable than the former over the years. This reflects the fact that households can smooth out their consumption by saving and dissaving, while income fluctuates yearly depending on factors such as the household's employment situation and agriculture yield [22]. Therefore, whenever possible, consumption was used to rank households. Consumption variables were available for all countries except for China and the Philippines. In addition to the choice between consumption and income, different methods by which consumption/income is aggregated reduce data transparency and comparability. It is therefore ideal to use consumption/income aggregates constructed using the same method. Due to lack of such data in some cases, however, we aggregated various consumption components, following the methodology proposed by Hentschel and Lanjouw [23] and Deaton and Zaidi [22].2 Even in cases where consumption data exist, their comprehensiveness varies across surveys. For example, the LSMS is the most comprehensive of all, including consumption items such as home-produced food, the imputed rental value of the household's dwelling, and the annual service value of the durable goods, as well as spending on food, non-food items, health services and education. By contrast, some surveys-such as that for Guatemala-offer only a very limited range of consumption categories with only food and certain non-food items. Readers are advised to bear the above caveats in mind when interpreting the results presented in this paper. We equivalize household consumption to take into account inter-household differences in household size. The two extreme positions on equivalization are (a) to assume that there are no economies of scale in household consumption (it costs two people twice as much to live as one) and (b) to assume that there are maximum economies of scale (two can live as cheaply as one). These two extremes, and the various possible intermediate positions, can be represented by the following relationship between equivalent consumption and actual consumption: 2 This was the case for the following countries: Guatemala, Philippines, Russia and Zambia. 6 (8) E = Aff where E is equivalent consumption, A is actual consumption, H is household size, and e an equivalence scale elasticity [24]. Under the assumption that there no economies of scale, e is set equal to 1, and equivalent consumption is simply per capita consumption. Under the assumption that two (or three, or four, of five,...) can live as cheaply as one, e is set equal to 0, and equivalent consumption is simply aggregate household consumption. Although it is not uncommon to find to find e set equal to one (the per capita adjustment), a more plausible position, at least in countries where a sizeable proportion of consumption is on non-food items, is that there are some economies of scale, but that the elasticity e is greater than zero. In their survey of equivalence scales in OECD countries, Buhmann et al. [24] found that most equivalence scales could be approximated quite closely by eqn (8) and that, on average, the implied value of the elasticity e was around 0.4. Hentschel and Lanjouw (op. cit.), in their work on Ecuador, experiment with three values of e: 0.4, 0.6, and 1.0. In what follows, we set e equal to 0.5, which seems a reasonable intermediate position. MEASURING MALNUTRITION Growth assessment is the single measurement that "best defines the health and nutritional status of children as disturbances in health and nutrition, regardless of their etiology, invariably affect child growth" [25]. Among various growth-monitoring indices, there are three commonly- used anthropometric measures that offer a comprehensive profile of malnutrition: stunting, underweight and wasting. The term "stunting" is used to describe a condition in which children fail to gain sufficient height, given their age. Stunting is therefore an extremely low "height-for-age" (H/A) score. Stunting is often associated with long-term factors such as chronic malnutrition, especially protein-energy malnutrition, and sustained and frequent illness. It is therefore an indicator of past growth failure and is often used for long-term planing of policies and intervention programs in non-emergency situations. The term "underweight" is used to describe a situation where a child weighs less than expected, given his or her age. Underweight is thus an extremely low "weight-for-age" (W/A) score. Unlike height, weight fluctuates over time and therefore reflects current and acute as well as chronic malnutrition. The term "wasting" refers to a situation where a child has failed to achieve sufficient weight for height (W/H). Wasting often results from recent and continuing severe weight loss due to inadequate energy intake, recent and continuing poor health, or a combination of both. The preferred reporting system of H/A, W/A and W/H is in terms of Z-scores3-_ a statistical measure of the distance from the median (mean) expressed as a proportion of the standard deviation. The most common cutoff point is -2Z-score, i.e., two standard deviations below the median values of the international reference. This is the cutoff risk level used to differentiate malnourished children from those adequately nourished. Children whose WA, W/A and W/1l scores fall below this point are therefore considered, stunted, underweight and wasted, 3Z - score =(Observed value) - (Median value of the referencevalue) Standard deviation of the reference population 7 respectively. The World Health Organization adopts US National Child Health Survey (NCHS) anthropometric data as the international reference to estimate its malnutrition indicators. To be comparable with WHO global estimates and other similar studies this paper follows the same methodology. The three dimensions of malnutrition are interrelated. This is shown in Figures 2 and 3 which map the incidence of wasting against I/A and W/A Z-scores for Bangladesh and Brazil, respectively. Children located towards the right end along the x-axis have a high H/A score (i.e. are tall), while those towards the left end have a low H/A score (i.e. are short). Those to the left of -2Z line are classified as stunted. Children towards the top end of the y-axis have a high W/A score (i.e. are heavy), while those towards the bottom end have a low W/A score (i.e. are light). Those below -2Z line are classified as underweight. Children at the bottom-left corner (the framed area) of the figures are both stunted and underweight. Since wasting is reflected in a low W/H score, and since the axes already capture weight and height (albeit adjusted for age), we can also speculate where wasted children lie in the figures. Children who are both light and short (i.e. children in the bottom-left quadrant) will tend to have W/H scores in the normal range. These children, some of whom will be both stunted and underweight, will be most unlikely to be wasted (i.e. to have a very low W/H score). Likewise, children who are both heavy and tall (i.e. children in the top right quadrant) will tend to have W/H scores in the normal range. For the most part, wasted children will be those in the bottom right quadrant-children who are fairly tall but also fairly light in weight. In Figure 2 the oval circle diagonal to the x- and y-axes capturing most wasted children lies outside the framed area. Thus, it makes it clear most children who are both stunted and underweight are not wasted, i.e., children can be stunted and underweight, and be underweight and wasted, but are unlikely to be stunted and wasted. In Figure 3 the principal cluster lies more to the direction of south-west than the one in Figure 2, indicating higher incidence of stunting and underweight in Bangladesh than in Brazil. Although there are more cases of simultaneous stunting, underweight and wasting in Bangladesh, only 9% of children fall in this category. We would therefore not expect to see a country with high inequalities in all three dimensions of malnutrition. COMPARISONS OF MALNUTRITION LEVELS WITH OTHER SURVEYS Malnutrition indicators computed along the lines indicated in section III were checked against WHO Global Database on Child Growth and Malnutrition to ensure their comparability [26]. Table 2 shows that the discrepancies between WHO reference values and our estimates are within +/-10% for most cases. Exceptions are: Egypt (stunting) and Romania (stunting) probably due to the different survey years from reference survey years; Guatemala (stunting) and the Philippines (all) most likely attributable to the regionally representative surveys; and Nicaragua (stunting) and Pakistan (wasting) possibly because of dropped observations due to insufficient consumption information. 8 IV. Inequalities in Malnutrition QUINTILE-SPECIFIC MALNUTRITION RATES Table 3 shows rates of stunting, underweight and wasting by quintile of equivalent consumption. A glance at the table reveals the first finding of the paper: in almost all countries, the poorest quintile has the highest rate of malnutrition-however malnutrition is measured. This is less clear in the case of wasting than in the cases of stunting and underweight but is evident there too. Another finding also emerges from Table 3: it is not simply a question of the poor having elevated rates of malnutrition; rather, the rate of malnutrition declines with living standards, although not always monotonically so. The extent to which the rates decrease indicates how much more the poor suffer from higher rates of malnutrition than the better off. For example, in Peru the rate of stunting among the lowest quintile is about 50%, whilst in the second quintile it is 44%. Then it decreases continuously until it reaches 10%. The prevalence of stunting among the poorest segment of the Peruvian population is relatively high compared with other countries, while that of the richest quintile is arnong the lowest. This points to a third finding-inequalities in malnutrition appear to vary across countries. Inequalities seem to be more pronounced in Peru than other countries. The opposite applies to Egypt where the poor- nonpoor gap in underweight is very small. We examine this issue in more detail in the next two subsections. CONCENTRATION INDICES Quintile comparisons do not lend themselves easily to inter-country comparisons of inequalities. The concentration index, introduced in section II, provides a straightforward way of capturing these inequalities. It also provides a means of testing the significance of inequalities in malnutrition. Table 4 shows concentration indices for stunting, underweight and wasting amongst under-five children ranked by equivalent consumption. All concentration indices for stunting and underweight are negative, reflecting the higher rates of malnutrition ainongst the poor. The values of the t-statistics bring us to another finding: inequalities in stunting and underweight are statistically significant in all countries, except in the cases of Egypt (both) and Russia (underweight). In other words, the tendency of poorer children to have higher rates of stunting and underweight is not due to chance or sampling variability. In the case of wasting, the picture is rather different-only eight countries have statistically significant concentration indices. Looking at the average concentration indices in the bottom row reveals another interesting result: inequalities in underweight tend to be larger than inequalities in stunting, which in turn tend to be larger than inequalities in wasting. There are, of course, exceptions to this pattern. In China, Ghana and Nicaragua, inequalities in wasting are more pronounced than inequalities in either stunting or underweight. The indices also reveal some interesting cross-country differences. Peru has the most negative concentration indices for stunting and underweight, and Nicaragua for wasting. Egypt exhibits the most pro-poor distribution of stunting and underweight, while Vietnarn leads the 9 ranking of wasting with a positive concentration index. It is not surprising that the top and bottom of the ranking for wasting did not coincide with those for stunting and underweight, as it is rare to observe children who are stunted, underweight and wasted at the same time (cf. Figures 2 and 3) The overall concentration index rankings of all three categories are similar to some extent. Peru, Morocco, Nicaragua are found on the lower side of the spectrum and Egypt, Vietnam, Romania and Pakistan on the opposite side and the other countries somewhere in the middle. TESTS OF SIGNIFICANT DIFFERENCES BETWEEN CONCENTRATION INDICES We now rank countries by inequality in a statistically more rigorous way. The standard errors of the concentration indices shown in Table 4 enable us to rank countries according to whether they have significantly more inequality than others. Tables 5-7 report the results of t- tests indicating the significance of the difference between the concentration indices of the column and row countries. Thus, for example, in Table 5, Bangladesh has a significantly less inequality in stunting than Brazil (hence the plus sign in front of the 4.05). Bangladesh also has less inequality in stunting than Cote d'Ivoire, but the difference in this case is not statistically significant. Bangladesh has more inequality in stunting than Egypt, but the difference is again not statistically significant. A glance across Tables 5-7 reveals one important point-in the majority of cases, whatever the indicator, the differences in inequality between countries are not statistically significant. Thus in only 44% of the 190 pairwise comparisons for stunting inequalities is the t- ratio larger than 1.96 in absolute size, while the equivalent percentages for underweight and wasting are 42% and 23% respectively. This warns against reading too much into concentration index differences unless accompanied by statistical tests. A Hasse diagram indicating the hierarchical order of countries makes it easier to grasp which inter-country differences are significant. The principle of the chart is that the concentration indices of all countries on the same level are not significantly different from one another, but are significantly larger (or smaller) than those of all countries on a different level. However, with a large number of countries, a perfectly accurate chart would become extremely cumbersome and would make it harder-rather than easier-to grasp the essential results. The Hasse charts presented here are therefore simplified Hasse charts, intended to convey the broad results of the various pairwise comparisons. Figure 4. Simplified Hasse diagram for under-five stunting by equivalent consumption Egypt Bangladesh Ghana Guatemala Indonesia Nepal Pakistan Romania Vietnam Zambia L A L L~~~~~~~~~~~~~~~~~~~~------------I- r-------------r--------------T----------------------- -------- - ----- ---------------T ---- 1-- ----- T---~-~------1 Brazil China Cote d'lvoire Guyana Morocco Nicaragua Philippines S Africa Per __- ---- ----------- Peru Rssia 10 The simplified Hasse diagram for under-five stunting (Figure 4) exhibits a four-level structure, characterized by a few countries at the top and the bottom, and a bulge in the middle levels. Egypt heads the hierarchy, although its concentration index is not in actuality significantly larger than all the countries in the lower levels. It is followed by a set of a large number of countries that more or less cluster together with those in the same concentration-index level. Peru and Russia have the most unequal distributions of stunting among children, with only Peru's concentration index being significantly the lowest of all. Figure 5. Simplified Hasse diagram for under-five underweight by equivalent consumption Egypt r---1 ---"7 -- --------- -------L---------r--- - l Bangladesh Indonesia Pakistan Romania Russia Vietnam I I - - -- r ----- ------------------------T-----------r---I------------ -- China Cote d'lvoire Ghana Guatemala Nepal Philippines S Africa Zambia Brazil 'Guyana Morocco Nicaragua Peru In the case of underweight (Figure 5), Egypt again appears at the top of the diagran and Peru at the bottom along with four other countries. Unlike the case of inequalities in stunting, there is no country that is significantly different from all the other countries. In other words, it is not clear whether inter-country differences in concentration indices are due to actual differences in inequality or to sampling variations. Figure 6. Simplified Hasse diagram for under-five wasting by equivalent consumption Cote d'lvoire Russia Vietnam I I i Bangladesh Brazil Egypt Guatemala Guyana Indonesia Pakistan Prilippines Romania S Africa Zambia I I I I I- IIi r-------- ---- ----- T-- ----------- --t---------- - ------T--------------- --1 China Ghana Morocco Nepal Peru Nicaragua As for wasting, the concentration indices of C6te d'Ivoire, Russia and Vietnam are the largest and are not significantly different from one another or from those of the countries one level down, except in the case of Vietnam (Figure 6). The concentration indices of all the countries in the same level are not significantly different from one another, with the exception of Philippines whose concentration index is significantly lower than Zambia. The Nicaraguan concentration index is the lowest of all and is statistically significantly lower than that of any other country. 11 Although the ranking of countries differs across the indicators, some broad conclusions can be drawn from the observation of the Hasse diagrams. Egypt and Vietnam recurrently emerge as countries with the least prorich distributions of malnutrition. By contrast, the most unequal distributions are to be found in Peru and Nicaragua, followed by Morocco. SENSITIVITY ANALYSES When interpreting the above results, two cautionary points merit attention. First, there are certain variations in the sample of children in terms of age interval and age range across countries. As Table 1 shows, all countries adopt age in months except for the Philippines (age in years) and the upper age limit of 4.99 years except for Nepal (3.99 years). Table 8 shows the effects on the concentration index of changing the age interval from monthly to yearly. Although the results are not identical, they are very comparable with only small gaps occurring in 5 of 12 comparisons. By contrast Table 9 suggests that progressively narrowing down the sample, first to children below the age of four and then to children below the age of three, causes significant changes in the concentration indices. On the whole, reducing the upper age limit may increase or reduce concentration indices, but larger gaps emerge when the limit is reduced from four to three than when it is reduced from five to four. Despite such discrepancies it is important to note that changes in the age limit do not systematically produce an upward or downward bias in concentration indices. Second, throughout we have adopted -2 standard deviations below the median (-2Z- score) as the cut-off point below which children are classified as malnourished. The term "severe malnutrition" is applied to children who fall below the more demanding threshold of -3Z-score. Table 10 shows the effect of reducing the threshold for four Asian countries- Bangladesh, Nepal, Pakistan and Vietnam. The comparisons for stunting and underweight show that there is more pro-rich inequality when the more demanding cut-off point is used. One exception is for underweight in the case of Vietnam. In the case of Nepal, the value of the concentration index is especially sensitive to the change of cut-off point. V. Inequalities in Malnutrition vs. Inequalities in Consumption It is beyond the scope of this paper to answer the question of why countries vary so much in their inequalities in child malnutrition. But there is one interesting question that can be answered readily with the data to hand: Are the countries with the most unequal distributions of malnutrition the countries with the most unequal distributions of consumption? We report here scatter plots and bivariate regressions showing the relationship between inequality in malnutrition, measured using the concentration index, and the Gini coefficient for consumption inequality. Although the Gini coefficient for consumption is available from various published sources (e.g., World Development Indicators), we chose, for various reasons, to compute Ginis for our samples, not least because several of our surveys are not nationally representative. The results are shown in Figure 7-9. The following is evident: especially in the case of stunting, and to a lesser extent in the case of underweight as well, it is indeed the case that countries with unequal income distributions also tend to have unequal distributions of malnutrition. This is not altogether surprising. Unequal distribution of purchasing power, prima 12 facie, leads to an unequal distribution of food spending (intake), health spending and utilization of health services, and consequently unequal health outcomes. It is also in line with the theoretical results of Contoyannis and Forster [27]. They showed that if the relationship between health and income is concave, a mean-preserving reduction in income inequality, with the new Lorenz curve for income strictly dominating the old, will result in a reduction in the concentration index for health inequality. What is more interesting, perhaps, is the fact that the fit of the bivariate regressions is fairly bad-there are, in other words, many countries that buck the trend Nepal and Peru, for example, have roughly the same level of income inequality, and yet Nepal has far lower levels of inequality in stunting and underweight than Peru. This implies that there must be some form of mechanism in these countries that breaks the link between poverty and malnutrition. For example, in the case of Egypt, which tends to positively deviate from the mainstream trend, it would be of interest to explore what factors, given the level of consumption inequality, contribute to relatively low inequalities in malnutrition. Similarly, it would be of interest to investigate why, in Peru, the level of inequality in malnutrition is higher than one would expect, given what other countries appear to achieve at the same level of consumption inequality. These questions are left for future research. VI. Inequalities in Malnutrition vs. Average Rates of Malnutrition Given the focus in international development targets on average rates of malnutrition, it is of some interest to establish how countries compare on average rates of malnutrition and inequalities in malnutrition. Ideally, one would like policymakers and target-setters to concern themselves with both dimensions. SCATTER PLOTS: AVERAGES VS. INEQUALITIES Figures 10-12 show scatter plots with the prevalence of malnutrition (i.e. the average rate) on the x-axis and the concentration index (i.e. the degree of inequality) on the y-axis. As far as stunting is concerned, countries can be roughly classified into four groups based on their stunting rate and their concentration index. The first group can be characterized by a "win-win" situation with a relatively low prevalence of stunting along with a small rich-poor gap (Egypt and Romania). The second group combines a low stunting rate with a relatively high concentration index (Guyana, the Philippines, Nicaragua, Brazil and China). The third group combines a relatively high stunting rate with a low concentration index (Russia, South Africa, Morocco and Peru). The last group consists of all the other countries with a relatively high stunting rate and a low concentration index (Figure 10). For underweight, the overall picture is similar to that of stunting, but a clearer trend emerges in the scatter plot. Starting from Brazil, Nicaragua and Peru in the bottom-left corner, countries move up both along the x-axis and the y-axis toward the top-right corner. Generally speaking, a country that enjoys a low malnutrition rate at the national (or regional) level is likely to suffer from a relatively wide poor-nonpoor gap in prevalence. By contrast, in a more egalitarian society in terms of health outcomes, the prevalence of adverse outcomes is higher across the socioeconomic distribution (Figure 11). 13 The story is slightly different for wasting. The countries are distributed in a bipolar fashion with the Philippines, Bangladesh and Pakistan clustering between prevalence rates of 20- 25%, and others having rates below 15%. In the latter cluster, Vietnam, Cote d'Ivoire, Russia, Zambia, Egypt and South Africa demonstrate a pro-poor distribution of wasting. The opposite is observed for Nicaragua, China, Morocco and Peru, although Nicaragua lies far from the other countries with an extremely unequal distribution (Figure 12). Some countries clearly do better in term of both the average (the prevalence of malnutrition) and the distribution (equality), e.g., Egypt vs. Peru for stunting. However, the scatter plots show the danger of setting targets and comparing countries in terms solely of average malnutrition rates. Brazil, for example, has a far lower overall stunting rate than Bangladesh (below 20% in Brazil, compared to in excess of 50% in Bangladesh). Without knowledge of the inequality, one would conclude that malnutrition is worse in Bangladesh than in Brazil. But knowing that there is a much larger inequality in stunting rates between poor and nonpoor children in Brazil than there is between poor and nonpoor children in Bangladesh makes it much harder to jump to this conclusion. The two countries simply perform differently in the two dimensions. AN ACHIEVEMENT INDEX CAPTURING INEQUALITY AND AVERAGE MALNUTRITION In an earlier paper [28], one of us proposed an index that captures the goals of policymakers-a low average level of malnutrition, and a small gap in malnutrition rates between poor and better-off children. This index is a weighted average of the nutrition rates of the various consumption groups, where poorer groups are assigned higher weights than richer groups. The general form for the achievement index is: (9) 1= NE=h where N is the number of people in the sample, hi is the ill-health of person i, and wi is a weight attached to person i's ill-health when computing the index I. The weights used are simply the person's absolute rank in the distribution of living standards, denoted by ri. This is equal to 1 for person 1, 2 for person 2, and N for person N. Then the weights are defined as (10) w1 =2 N ' - Thus we is equal to 2 for the most disadvantaged person, declines by 2/N for each one-person step up through the living-standard distribution, and reaches 2/N for the least disadvantaged person. Thus the difference in wi between the most disadvantaged person and the second most disadvantaged person is the same as the difference between the second most advantaged person and the most advantaged person. When the wi are so defined, the index I is equal to (11) IR = p(l - C), 14 where IR denotes the value of I when the weights are based on the person's rank in the socioeconomic distribution, and C is the concentration index for ill health, defined along the lines of eqn (1). The implications of the index in eqn (11) are straightforward. When everyone- irrespective of their living standard-has the same level of malnutrition, C is zero, and IR equals ,u. When poor individuals have higher levels of malnutrition than better-off individuals, C will be negative (but larger in numerical size than minus one). In this case, IR will be larger than p- the inequality in the distribution of malnutrition forces the index IR above the mean. For example, in the case where C=-0.25, IR will be 25% higher than M. The opposite will happen when inequality in malnutrition favors the disadvantaged. In this case, C will be positive (and less than one), and IR will fall below ,u. Evidently, the index IR allows some trade-off to be made between the average nutritional status of the population and socioeconomic inequality in the distribution of malnutrition. Suppose the average level of malnutrition, p, is lower in country X than in country Y. Then the index IR could still be higher in country X than in country Y if the distribution of malnutrition in country X is that much more pro-rich than in country Y. It is worth noting that the particular weighting scheme for the w, in eqn (10) is precisely the same scheme that underlies-albeit implicitly-the concentration index, C, as well as the aforementioned RII. Insofar as these indices are considered acceptable health inequality measures, the index IR ought also to be considered an acceptable index that combines information on inequality with information on the average level of malnutrition. ACHIEVEMENT INDICES FOR TWENTY COUNTRIES Table 11 shows the values of the index IR for the 20 countries for the three measures of malnutrition. In the low-inequality countries, IR is, inevitably, close to the sample average, whilst in the high-inequality countries, it exceeds the average. In the case of Peru, for example, although the sample average rate of stunting is only 0.31, IR is over 0.40, reflecting the high inequality in that country. Despite the relatively few countries included in the analysis, moving from the sample mean, M, to the achievement index, IR, produces several rank reversals. In the case of stunting, there are two sets of rank reversals. In one of the two cases, these simply involve two adjacent countries swapping places-Zambia slips behind Nepal. In the other, there is more movement-Egypt (a low inequality country) moves from sixth position to fourth, overtaking Russia and Brazil (two high inequality countries). In the cases of underweight and wasting, there are three and five rank reversals respectively, all involving two adjacent countries swapping places. Evidently as the number of countries in the sample increases, the chances increase of high-inequality countries falling behind low-inequality countries as one moves from a focus on the mean of the distribution of malnutrition to a focus on the poverty-sensitive achievement index IR. 15 VII. Conclusions Our aim in this paper has been to shed light on the extent of inequalities in malnutrition between poor and nonpoor children in 20 countries in the developing world. We can summarize our main conclusions as follows: 1) Inequalities in malnutrition almost always disfavor the poor. In almost all countries, the poorest quintile has the highest rate of malnutrition, however malnutrition is measured. This is less clear in the case of wasting than in the cases of stunting and underweight but is evident there too. 2) It is not simply a question of the poor having elevated rates of malnutrition. Rather the rate of malnutrition declines continuously with rising living standards, although not always monotonically. 3) Inequalities in stunting and underweight, as measured by the concentration index, are statistically significant in all countries, with the exceptions of Egypt (both indicators) and Russia (underweight). In other words, the tendency of poorer children to have higher rates of stunting and underweight is not due to chance or sampling variability. In the case of wasting, the picture is rather different-only eight countries have statistically significant concentration indices. 4) Inequalities in underweight tend to be larger than inequalities in stunting, which in turn tend to be larger than inequalities in wasting. There are exceptions to this pattern-in China, Ghana and Nicaragua, inequalities in wasting are more pronounced than inequalities in either stunting or underweight. 5) Although there are large cross-country variations in inequality, as measured by the concentration index, in the majority of cases, whatever the malnutrition indicator, the differences in inequality between countries are not statistically significant. Thus in only 44% of the 190 pairwise comparisons for stunting inequalities is the t-ratio larger than 1.96 in absolute size. The equivalent percentages for underweight and wasting are even smaller- 42% and 23% respectively. This warns against reading too much into concentration index differences unless accompanied by statistical tests. 6) Even if attention is restricted only to those cross-country differences in inequality that are statistically significant, interesting conclusions still emerge. Egypt and Vietnam emerge as countries with the least pro-rich distributions of malnutrition, while highly unequal distributions are consistently found in Peru and Nicaragua, and to a lesser extent in Morocco. 7) Sensitivity analysis reveals that changing age range and interval in the sample causes certain variations in concentration indices although not a systematic increase or decrease. By contrast, selected Asian countries' data suggest that lowering the cutoff point from conventional -2Z-score to more demanding -3Z-score almost always leads to more pro-rich inequality in stunting and underweight. 8) Especially in the case of stunting, and to a lesser extent in the case of underweight as well, there is an association at country level between an unequal income distribution and an unequal distribution of malnutrition. However, the fit of the bivariate regressions on these 16 data is fairly bad-there are, in other words, many countries that buck the trend. Nepal and Peru, for example, have roughly the same level of income inequality, and yet Nepal has far lower levels of inequality in stunting and underweight than Peru. 9) Some countries do well in terms of both the average (the prevalence of malnutrition) and the distribution (equality). Exarnples for underweight include Egypt and Russia. Others do relatively badly on both counts. For example, Peru has a higher average level of stunting than Egypt and a larger level of poor-nonpoor inequality. In many cases, however, countries do well on one count (the average, say) while doing badly on the other (the level of inequality, say). Brazil, for example, has a far lower overall stunting rate than Bangladesh (below 20% in Brazil, compared to in excess of 50% in Bangladesh), but has four times as much inequality (as measured by the concentration index). 10) Use of an achievement index that captures both the average level of malnutrition and the inequality in malnutrition leads to some interesting rank reversals in the country league table. In the case of stunting, for example, moving from a focus on the average to a focus on the achievement index results in Egypt (a low inequality country) moving up from sixth position to fourth, overtaking Russia and Brazil (two high inequality countries). 17 References 1. World_Health_Organization, WHO Global Database on Child Growth and Malnutrition. 1999, Geneva: World Health Organization. 2. Smith, L. and L. Haddad, Overcoming child malnutrition in developing countries: Past achievement andfuture choices. 2000, International Food Policy Research Institute: Washington, DC. 3. Murray, C. and A. Lopez, Quantifying the Burden of Disease and Injury Attributable to Ten Major Risk Factors, in The Global Burden of Disease, C. Murray and A. Lopez, Editors. 1996, The Harvard School of Public Health on behalf of The World Health Organization and The World Bank.: Cambridge, MA. 4. World_Bank, World Development Indicators 2000. 2000, Washington DC: The World Bank. 5. UK_Department_for_International_Development, International Development Target Strategy Paper: Better Health for Poor People. 1999, London: UK Department for International Development. 6. World_Bank, Health, Nutrition and Population Sector Strategy. 1997, Washington DC: The World Bank. 7. World_Health_Organization, The World Health Report 1999: Making a Difference. 1999, Geneva: The World Health Organization. 8. Gwatkin, D., Health inequalities and the health of the poor: What do we know? What can we do? Bulletin of the World Health Organization, 2000. 78(1): p. 3-17. 9. OECD, Measuring development progress: A working set of core indicators. 2000, OECD: Paris. 10. World_Bank_Africa_Region, Nutritional Status and Poverty in Sub-Saharan Africa, in Findings. 1998, World Bank: Washington, DC. 11. Demery, L.. 12. Gwatkin, D., S. Rutstein, K. Johnson, R. Pande, and A. Wagstaff, Socioeconomic Differences in Health, Nutrition and Population. 2000, The World Bank: Washington DC. 13. Wagstaff, A., Socioeconomic inequalities in child mortality: comparisons across nine developing countries. Bulletin of the World Health Organization, 2000. 78(1): p. 19-29. 14. Wagstaff, A., P. Paci, and E. van Doorslaer, On the measurement of inequalities in health. Social Science and Medicine, 1991. 33: p. 545-557. 18 15. Kakwani, N., A. Wagstaff, and E. Van Doorlsaer, Socioeconomic inequalities in health: Measurement, computation and statistical inference. Journal of Econometrics, 1997. 77(1): p. 87-104. 16. Kunst, A.E., J.J. Geurts, and J. van den Berg, International variation in socioeconomic inequalities in self reported health J Epidemiol Community Health, 1995. 49(2): p. 117-23. 17. Mackenbach, J.P., A.E. Kunst, A.E. Cavelaars, F. Groenhof, and J.J. Geurts, Socioeconomic inequalities in morbidity and mortality in western Europe. The EU Working Group on Socioeconomic Inequalities in Health Lancet, 1997. 349(9066): p. 1655-9. 18. Pamuk, E., Social class inequality in mortality from 1921-1972 in England and Wales. Population Studies, 1985. 39: p. 17-31. 19. Pamuk, E., Social class inequality in infant mortality in England and Wales from 1921 to 1980. European Journal of Population, 1988. 4: p. 1-2 1. 20. Schalick, L.M., W.C. Hadden, E. Parnuk, V. Navarro, and G. Pappas, The widening gap in death rates arnong income groups in the United States from 1967 to 1986. International Journal of Health Services, 2000. 30(1): p. 13-26. 21. Wagstaff, A., Caring more about the poor: Inequality aversion, health inequalities and health achievement. 2000, World Bank: Washington DC. 22. Deaton, A. and S. Zaidi, Guidelines for constructing consumption aggregates for welfare analysis. 2000, World Bank: Washington DC. 23. Hentschel, J. and P. Lanjouw, Constructing an indicator of consumption for the analysis of poverty. Principles and illustrations with reference to Ecuador. 1996, World Bank: Washington DC. 24. Buhmann, B., L. Rainwater, G. Schmaus, and T. Smeeding, Equivalence scales, well- being, inequality and poverty. Review of Income and Wealth, 1988. 34: p. 115-142. 25. de Onis, M., C. Monteiro, J. Akre, and G. Clugston, The worldwide magnitude of protein- energy malnutrition: An overview from the WHO global database on child growth Bulletin of the World Health Organization, 1993. 71(6): p. 703-712. 26. World_Health_Organization, WHO Global Database on Child Growth and Malnutrition. 1997, World Health Organization: Geneva. 27. Contoyannis, P. and P. Forster, The distribution of health: a theoretical framework. Hournal of Health Economics, 1999. 18: p. 605-622. 28. Wagstaff, A., If the health of the poor matters more: Child survival inequalities in nine developing countries. 2000. 19 Table 1. Surveys used in under-5 malnutrition inequalities analysis Survey Represen- No. No. No. No. Age Age Country Survey name year tation Households Children children children interval range Comments on data (Stunting) (Underweight) (Wasting) Bangladesh Matlab Health and Socieconomic 1996 Regional 4364 1512 1543 1504 Month 0-4.99 The survey only covers a rural region of Matlab, which is Survey located just in the south of Dakha Brazil Presquisa sobre Padr6es de Vida 1995-96 Regional 4940 1697 1791 1678 Month 0-4.99 The areas are south-east and north-east regions only. China China Health and Nutrition Survey 1991 Regional 3616 865 883 850 Month 0-4.99 8 provinces are covered by the survey covering both urban and rural areas within them. This sample is diverse in termns of socioeconomic factors (income, employment, education and modernization) and other related health, nutritional and demographic measures. Cote LSMS 1988 National 1600 2121 2121 2120 Month 0-4.99 d'Ivolre Egypt Egypt Integrated Household 1997 National 2500 1427 1434 1430 Month 0-4.99 Survey Ghana LSMS 1987-88 National 3200 2349 2350 2341 Month 0-4.99 Guatemala Guatemalan Survey of Family 1995 Regional 4792 2817 2854 2814 Month 0-4.99 The survey covers 4 departments (out of 22 in total). Health Guyana LSMS 1992-93 National 5340 590 589 587 Month 0-4.99 Indonesia Indonesian Family Life Survey 1993 RegIonal 7224 1250 1371 1233 Month 0-4.99 The survey covers 13 provinces that represent 83 % of the population. Morocco LSMS 1990-91 National 3323 2121 2121 2120 Month 0-4.99 Nepal LSMS 1996 National 3373 1597 1603 1586 Month 0-3.99 Nicaragua LSMS 1993 National 4200 514 520 511 Month 0-4.99 Pakistan LSMS 1991 National 4800 3773 4051 4127 Month 0-4.99 Peru LSMS 1994 National 3623 2093 2110 2075 Month 0-4.99 Philippines Cebu Longitudinal Health and 1991 Regional 2264 2033 2036 2139 Year 0-4.99 The survey area Is the city of Cebu, the region center of Nutriton Survey Central Vlsayas region. Romania LSMS 1996 National 36000 3740 3755 3737 Month 0-4.99 Russia Russia Longitudinal Monitoring 1997 National 3750 386 417 377 Month 0-4.99 Survey South Africa LSMS 1993 National 9000 3971 3998 3947 Month 0-4.99 Vietnam LSMS 1992-93 National 4800 2623 2773 2609 Month 0-4.99 Zambia Uving Condions Monitoring 1996 Natlonal 11770 4500 8154 4545 Month 0-4.99 Survey I 20 Figure 2. Mapping Incidence of Wasting against H/A & W/A Z-scores (Bangladesh) 0~~~~~~~~~~~~~~~~~~~~~~~~~~~ N~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ o 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 0% S 0 ~~0 -a ! 0 1 10 -8 * --, * ~4 *{li_l * t++.2 4 4 . 6 8 ....... ............. Stunting H/A Z-scores Wasted Notwasted 21 Figure 3. Mapping Incidence of Wasting against H/A & W/A Z-scores (Brazil) ¢ I .* , , 22s Xs *-.+ . . 0~~~~~~~~~~~~~~ 0 ___ _ ..__ ._.- _ o - _ -8-6 _ _ _ 6 _ _ ... . ..... .. .. Stunting H/A Z-scores I_____ __Nwa t I~ ~ Waste * No wased 22 Table 2. Sample under-5 malnutrition rate estimates Country Survey Reference Reference Stunting Underweight Wasting year survey survey yr Sample Reference % Sample Reference % Sample Reference % mean Value discre- mean value discre- mean value disare- Pancy pancy Pancy Bangladesh 1996 DHS 1996-97 51% 55% -3% 54% 56% -2% 22% 18% 5% Brazil 1995-96 DHS 1996 15% 11% 5% 6% 6% 1% 6% 2% 4% China 1991 The dietary and nutritional 1992 31% 31% 0% 15% 17% -2% 4% 4% -1% status of Chinese population CMte d'lvoire 1987-88 Malnutrition in CMte d'lvoire 1986 20% 17% 2% 17% 12% 5% 10% 9% 2% (WB wp) Egypt 1997 DHS 1995-96 17% 30% -13% 11% 12% -2% 5% 5% 0% Ghana 1987-88 LSMS 1987-88 31% 31% 1% 26% 27% -1% 5% 7% -3% Guatemala 1995 DHS 1995 61% 50% 12% 33% 27% 6% 1% 4% -3% Guyana 1992-93 Stunting and wasting: Guyana 1993 12% 21% -9% 19% 18% 1% 7% 9% -1% Nutritional Status Survey 1985; Underweight: HIES/LSMS Indonesia 1993 Indonesia multiple indicator 1995 40% 42% -2% 37% 34% 3% 11% 13% -2% duster survey Morocco 1990-91 DHS 1992 28% 24% 4% 15% 10% 5% 8% 2% 6% Nepal 1996 DHS * 1996 49% 49% 0% 47% 47% 0% 13% 11% 2% Nicaragua 1993 LSMS 1993 15% 24% -9% 8% 12% -4% 3% 2% 1% Pakistan 1991 DHS 1990-91 42% 50% -8% 43% 40% 3% 25% 9% 16% Peru 1994 DHS 1996 31% 26% 6% 12% 8% 4% 2% 1% 1% Philippines 1991 Regional nutrition survey 1992 14% 35% -21% 21% 33% -12% 22% 7% 15% Romania 1994 National nutrition survey 1991 24% 8% 16% 7% 6% 1% 5% 3% 3% Russia 1996 RLMS (R6) 1995 17% 13% 4% 6% 3% 3% 6% 4% 3% South Africa 1993 Anthropometric study ** 1994-95 26% 23% 3% 18% 9% 9% 10% 3% 7% Vietnam 1992-93 Malnutrition prevalence survey 1994 53% 47% 6% 41% 45% -4% 6% 12% -6% Zambia 1996 DHS 1996-97 48% 42% 6% 22% 24% -1% 6% 4% 1% Notes: * Age group 0.50-2.99. ** Age group 0.504.99. 23 Table 3. Rates of under-5 stunting, underweight and wasting, by quintile of equivalent consumption Stunting Underweight Wasting Quintiles Quintiles Quintiles Country 1 2 3 4 5 Overall 1 2 3 4 5 Overall 1 2 3 4 5 Overall average average average Bangladesh 0.56 0.55 0.52 0.50 0.43 0.51 0.59 0.62 0.55 0.50 0.44 0.54 0.28 0.21 0.24 0.19 0.21 0.22 Brazil 0.23 0.17 0.16 0.11 0.09 0.15 0.09 0.09 0.06 0.04 0.03 0.06 0.09 0.05 0.04 0.07 0.06 0.06 China 0.38 0.32 0.29 0.25 0.14 0.28 0.21 0.15 0.11 0.12 0.06 0.13 0.06 0.03 0.04 0.04 0.02 0.04 Cote d'Ivoire 0.26 0.20 0.21 0.13 0.18 0.20 0.21 0.22 0.14 0.13 0.15 0.17 0.10 0.10 0.07 0.12 0.11 0.10 Egypt 0.20 0.17 0.14 0.18 0.16 0.17 0.10 0.14 0.10 0.09 0.10 0.11 0.04 0.07 0.05 0.02 0.06 0.05 Ghana 0.38 0.36 0.32 0.27 0.25 0.31 0.32 0.29 0.26 0.24 0.19 0.26 0.06 0.05 0.05 0.04 0.03 0.05 Guatemala 0.70 0.69 0.66 0.57 0.47 0.62 0.41 0.39 0.34 0.27 0.25 0.33 0.01 0.01 0.00 0.00 0.01 0.01 Guyana 0.15 0.14 0.12 0.13 0.06 0.12 0.25 0.25 0.21 0.15 0.08 0.19 0.07 0.09 0.08 0.09 0.03 0.08 Indonesia 0.54 0.50 0.48 0.44 0.35 0.46 0.46 0.42 0.43 0.35 0.33 0.40 0.14 0.09 0.10 0.09 0.11 0.10 Morocco 0.39 0.36 0.31 0.20 0.15 0.28 0.23 0.20 0.16 0.10 0.06 0.15 0.12 0.09 0.06 0.08 0.05 0.08 Nepal 0.55 0.50 0.53 0.45 0.39 0.49 0.60 0.47 0.51 0.48 0.27 0.47 0.16 0.14 0.15 0.14 0.08 0.13 Nicaragua 0.24 0.15 0.14 0.14 0.09 0.15 0.15 0.10 0.06 0.05 0.04 0.08 0.07 0.01 0.03 0.02 0.00 0.03 Pakistan 0.46 0.48 0.47 0.37 0.31 0.42 0.48 0.48 0.43 0.40 0.35 0.43 0.28 0.29 0.24 0.21 0.22 0.25 Peru 0.51 0.44 0.32 0.20 0.10 0.31 0.22 0.14 0.13 0.06 0.05 0.12 0.04 0.02 0.02 0.02 0.02 0.02 Philippines 0.21 0.17 0.16 0.10 0.08 0.14 0.26 0.24 0.20 0.24 0.13 0.21 0.29 0.23 0.22 0.22 0.16 0.22 Romania 0.25 0.28 0.23 0.24 0.20 0.24 0.09 0.07 0.06 0.07 0.06 0.07 0.06 0.05 0.04 0.06 0.04 0.05 Russia 0.22 0.22 0.22 0.12 0.06 0.17 0.07 0.06 0.08 0.05 0.05 0.06 0.07 0.07 0.07 0.08 0.04 0.06 South Afica 0.39 0.30 0.23 0.23 0.12 0.26 0.24 0.22 0.17 0.17 0.11 0.18 0.10 0.11 0.11 0.10 0.08 0.10 Vietnam 0.60 0.61 0.56 0.50 0.38 0.53 0.48 0.46 0.42 0.42 0.29 0.41 0.04 0.05 0.07 0.07 0.07 0.06 Zambia 0.60 0.52 0.52 0.40 0.37 0.48 0.29 0.26 0.25 0.18 0.14 0.22 0.06 0.05 0.05 0.06 0.06 0.06 24 Table 4. Concentration indices, standard errors, t-values, and 95% confidence intervals for under-5 stunting, underweight and wasting County Stunting Underweight Wasting CI se(C) t(C) Low High CI se(C) t(C) Low High CI se(C) t(C) Low High Bangladesh -0.051 0.015 -3.491 -0.080 -0.022 -0.067 0.014 -4.939 -0.094 -0.040 -0.067 0.028 -2.391 -0.123 -0.011 Brazil -0.194 0.032 -5.996 -0.259 -0.129 -0.241 0.047 -5.168 -0.334 -0.147 -0.067 0.057 -1.176 -0.182 0.047 China -0.142 0.031 -4.570 -0.205 -0.080 -0.167 0.044 -3.746 -0.255 -0.078 -0.201 0.099 -2.041 -0.398 -0.004 Cote d'Ivoire -0.109 0.035 -3.063 -0.179 -0.038 -0.101 0.039 -2.589 -0.178 -0.023 0.020 0.052 0.384 -0.084 0.124 Egypt -0.039 0.034 -1.125 -0.107 0.030 -0.034 0.043 -0.784 -0.121 0.053 -0.036 0.067 -0.539 -0.170 0.098 Ghana -0.094 0.017 -5.427 -0.129 -0.059 -0.105 0.020 -5.305 -0.144 -0.065 -0.138 0.053 -2.584 -0.245 -0.031 Guatemala -0.078 0.008 -9.240 -0.095 -0.061 -0.109 0.015 -7.234 -0.139 -0.079 -0.086 0.134 -0.641 -0.355 0.183 Guyana -0.146 0.063 -2.315 -0.272 -0.020 -0.201 0.046 -4.413 -0.293 -0.110 -0.087 0.075 -1.170 -0.236 0.062 Indonesia -0.091 0.015 -6.033 -0.121 -0.061 -0.064 0.018 -3.513 -0.100 -0.027 -0.064 0.048 -1.337 -0.161 0.032 Morocco -0.185 0.019 -9.808 -0.222 -0.147 -0.251 0.027 -9.251 -0.305 -0.197 -0.169 0.042 -3.994 -0.253 -0.084 Nepal -0.065 0.015 -4.414 -0.095 -0.036 -0.120 0.034 -3.503 -0.189 -0.052 -0.108 0.046 -2.340 -0.201 -0.016 Nicaragua -0.158 0.015 -2.695 -0.188 -0.129 -0.276 0.034 -3.246 -0.345 -0.208 -0.496 0.046 -3.680 -0.588 -0.403 Pakistan -0.074 0.011 -6.771 -0.134 -0.015 -0.064 0.010 -6.174 -0.112 -0.017 -0.063 0.016 -4.061 -0.109 -0.017 Peru -0.280 0.017 -16.791 -0.314 -0.247 -0.307 0.030 -10.209 -0.368 -0.247 -0.155 0.088 -1.770 -0.330 0.020 Philippines -0.188 0.030 -6.344 -0.218 -0.159 -0.105 0.024 -4.444 -0.143 -0.067 -0.098 0.023 -4.265 -0.152 -0.044 Romania -0.051 0.016 -3.109 -0.084 -0.018 -0.087 0.035 -2.477 -0.157 -0.017 -0.053 0.042 -1.252 -0.138 0.032 Russia -0.221 0.058 -3.795 -0.338 -0.105 -0.077 0.105 -0.733 -0.288 0.134 0.006 0.113 0.051 -0.220 0.232 South Africa -0.199 0.015 -13.409 -0.228 -0.169 -0.142 0.019 -7.555 -0.179 -0.104 -0.033 0.027 -1.218 -0.087 0.021 Vietnam -0.088 0.011 -8.306 -0.109 -0.067 -0.089 0.013 -6.910 -0.115 -0.063 0.076 0.043 1.766 -0.010 0.161 Zambia -0.099 0.027 -3.653 -0.153 -0.045 -0.144 0.011 -12.603 -0.167 -0.121 -0.016 0.023 -0.676 -0.062 0.031 25 Table 5. Test of significance between concentration indices for under-5 stunting Bangladesh Brazil China C dIvoire Egypt Ghana Guatemala Guyana Indonesia Morocco Nepal Nicaragua Pakistan Peru Philippines Romanla Russia S Africa Vietnam Brazil 4.05 China 2.67 -1.15 Cote d'Ivoire 1.51 -1.79 -0.72 Egypt -0.32 -3.30 -2.24 -1.42 Ghana 1.92 -2.73 -1.36 -0.37 1.44 Guatemala 1.65 -3.46 -1.99 -0.83 1.13 -0.81 Guyana 1.47 -0.68 0.05 0.52 1.50 0.80 1.06 Indonesia 1.92 -2.90 -1.49 -0.46 1.39 -0.14 0.72 -0.85 Morocco 5.64 -0.26 1.16 1.89 3.73 3.54 5.15 0.58 3.89 Nepal 0.70 -3.63 -2.24 -1.13 0.71 -1.27 -0.77 -1.25 -1.21 -4.99 Nicaragua 5.21 -1.01 0.46 1.30 3.21 2.83 4.71 0.19 3.21 -1.09 4.47 Pakistan 1.30 -3.51 -2.06 -0.92 0.99 -0.96 -0.29 -1.12 -0.89 -5.06 0.49 -4.58 Peru 10.38 2.36 3.90 4.38 6.34 7.74 10.79 2.05 8.43 3.80 9.65 5.46 10.31 Philippines 4.17 -0.13 1.07 1.73 3.30 2.75 3.57 0.61 2.93 0.11 3.72 0.90 3.61 -2.69 Romania 0.03 -3.94 -2.59 -1.47 0.33 -1.79 -1.46 -1.46 -1.77 -5.33 -0.63 -4.85 -1.17 -9.77 -4.04 Russia 2.84 0.41 1.19 1.65 2.70 2.09 2.43 0.88 2.17 0.60 2.60 1.05 2.48 -0.97 0.50 2.81 S Africa 7.14 0.12 1.63 2.34 4.28 4.59 7.05 0.81 5.11 0.59 6.38 1.92 6.75 -3.66 0.30 6.65 -0.38 Vietnam 2.08 -3.12 -1.65 -0.55 1.38 -0.29 0.72 -0.91 -0.15 -4.47 1.26 -3.87 0.90 -9.72 -3.18 1.88 -2.25 -6.07 Zambia 1.57 -2.27 -1.06 -0.22 1.38 0.14 0.72 -0.69 0.25 -2.61 1.09 -1.95 0.84 -5.72 -2.24 1.50 -1.91 -3.25 0.36 26 Table 6. Test of significance between concentration indices for under-5 underweight Bangladesh Brazil China C dIvoire Egypt Ghana Guatemala Guyana Indonesia Morocco Nepal Nicaragua Pakistan Peru Philippines Romania Russia S Africa Vietnam Brazil 3.58 China 2.14 -1.15 Cote dIvoire 0.82 -2.31 -1.11 Egypt -0.73 -3.25 -2.14 -1.15 Ghana 1.58 -2.68 -1.27 0.10 1.49 Guatemala 2.07 -2.69 -1.22 0.20 1.64 0.17 Guyana 2.82 -0.60 0.55 1.68 2.66 1.94 1.92 Indonesia -0.15 -3.54 -2.14 -0.86 0.63 -1.54 -1.92 -2.80 Morocco 6.06 0.20 1.62 3.17 4.25 4.35 4.58 0.94 5.74 Nepal 2.62 -2.45 -0.98 0.47 1.89 0.62 0.53 -1.68 2.40 -4.20 Nicaragua 10.28 0.73 2.34 4.21 5.28 6.88 7.82 1.56 8.99 0.81 7.27 Pakistan -0.16 -3.70 -2.24 -0.90 0.68 -1.82 -2.44 -2.93 0.03 -6.43 -3.05 -11.53 Peru 7.27 1.21 2.62 4.20 5.19 5.62 5.89 1.94 6.93 1.39 5.55 0.92 7.63 Philippines 1.39 -2.60 -1.22 0.09 1.44 0.00 -0.15 -1.88 1.39 -4.06 -0.55 -6.11 1.58 -5.29 Romania 0.52 -2.64 -1.41 -0.27 0.95 -0.45 -0.59 -2.00 0.58 -3.71 -0.88 -4.97 0.61 -4.78 -0.43 Russia 0.10 -1.42 -0.78 -0.21 0.38 -0.26 -0.30 -1.08 0.13 -1.60 -0.41 -1.87 0.12 -2.10 -0.26 -0.08 S Africa 3.22 -1.97 -0.51 0.95 2.28 1.35 1.36 -1.21 2.99 -3.32 0.88 -5.58 3.61 -4.67 1.22 1.39 0.60 Vietnam 1.16 -3.14 -1.68 -0.29 1.22 -0.68 -1.02 -2.37 1.13 -5.40 -1.59 -9.43 1.48 -6.68 -0.60 0.06 0.11 -2.33 27 Table 7. Test of significance between concentration indices for under-5 wasting Bangladesh Brazil China C d'lvoire Egypt Ghana Guatemala Guyana Indonesia Morocco Nepal Nicaragua Pakistan Peru Philippines Romania Russia S Africa Vietnam Brazil 0.01 China 1.31 1.17 C6te dIvoire -1.47 -1.13 -1.98 Egypt -0.43 -0.36 -1.39 0.66 Ghana 1.18 0.90 -0.56 2.12 1.19 Guatemala 0.14 0.13 -0.69 0.74 0.33 -0.36 Guyana 0.25 0.21 -0.92 1.18 0.51 -0.56 0.01 Indonesia -0.05 -0.04 -1.25 1.19 0.34 -1.02 -0.15 -0.26 Morocco 2.00 1.42 -0.30 2.81 1.67 0.45 0.58 0.95 1.63 Nepal 0.91 0.60 -0.89 2.04 0.95 -0.47 0.16 0.25 0.73 -1.10 Nicaragua 9.52 6.36 2.81 8.21 6.07 5.58 2.95 4.96 7.23 5.95 7.78 Pakistan -0.13 -0.08 -1.39 1.53 0.39 -1.35 -0.17 -0.32 -0.03 -2.35 -1.17 -11.24 Peru 0.96 0.84 -0.35 1.72 1.08 0.16 0.43 0.59 0.91 -0.14 0.50 -3.61 1.03 Philippines 0.86 0.50 -1.02 2.08 0.88 -0.69 0.09 0.14 0.63 -1.46 -0.24 -9.44 1.27 -0.63 Romania -0.27 -0.20 -1.38 1.09 0.21 -1.25 -0.23 -0.40 -0.18 -1.93 -1.00 -8.03 -0.22 -1.05 -0.94 Russia -0.63 -0.58 -1.38 0.11 -0.32 -1.15 -0.52 -0.69 -0.57 -1.45 -0.96 -4.24 -0.60 -1.12 -0.90 -0.49 S Africa -0.87 -0.54 -1.65 0.90 -0.04 -1.75 -0.39 -0.68 -0.57 -2.70 -1.69 -10.40 -0.96 -1.33 -1.83 -0.40 0.33 Vietnam -2.79 -2.00 -2.58 -0.83 -1.40 -3.12 -1.15 -1.89 -2.17 -4.06 -3.31 -10.31 -3.04 -2.37 -3.58 -2.14 -0.58 -2.14 Zambia -1.41 -0.84 -1.83 0.63 -0.29 -2.10 -0.52 -0.92 -0.91 -3.17 -2.19 -11.37 -1.69 -1.54 -2.52 -0.77 0.19 -0.49 1.88 28 Table 8. Concentration indices with different age intervals Country Stunting Underweight Wasting Brazil Age in months -0.19 -0.24 -0.07 Age in years -0.19 -0.22 -0.07 Ghana Age in months -0.09 -0.10 -0.13 Age in years -0.11 -0.13 -0.13 Pakistan Age in months -0.07 -0.06 -0.05 Age in years -0.06 -0.06 -0.05 Vietnam Age in months -0.08 -0.08 0.08 Age in years -0.08 -0.07 0.08 Note: Concentration indices are at quintile (not individual) level. 29 Table 9. Concentration indices with different age range s Table 10. Concentration indices with different cut-off points Country Stunting Underweight Wasting Country Stunting Underweight Wasting Brazil Bangladesh Under 5 -0.19 -0.24 -0.07 -2Z -0.05 -0.06 -0.06 Under 4 -0.17 -0.25 -0.09 -3Z -0.08 -0.10 -0.16 Under 3 -0.13 -0.27 -0.18 Nepal -2Z -0.06 -0.11 -0.10 Cote d'lvoire -3Z -0.15 -0.21 0.04 Under 5 -0.10 -0.10 0.03 Under 4 -0.09 -0.10 0.01 Pakistan Under 3 -0.13 -0.09 0.03 -2Z -0.07 -0.06 -0.05 -3Z -0.10 -0.10 -0.04 South Africa Under 5 -0.19 -0.14 -0.04 Vietnam Under 4 -0.18 -0.13 -0.03 -2Z -0.08 -0.08 0.08 Under 3 -0.18 -0.24 -0.13 -3Z -0.12 -0.06 0.35 Note: Concentration indices are at quintile (not individual) level. Vietnam Under 5 -0.08 -0.08 0.08 Under 4 -0.10 -0.09 0.07 Under 3 -0.18 -0.05 0.37 Note: Concentration indices are at quintile (not individual) level. 30 Figure 7. Gini vs CI (stunting) 0.00 3 .0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 7 .O * Egypt -0.05 - * Romania * Bangladesh * Nepal * Guatemala * Pakistan ~~< +~~ Ghana *Indonesia -0.10 *Cote d'lvoir i -0.15 - *G n sn - * ~~~~~~~Morocco * rzl*Philippines -0.20 * Russia -0.25 -__ * Peru -0.30 - Gini 31 Figure 8. Gini vs CI (underweight) 0.00 - I I l 3 .0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 7, .0 * Egypt -0.05 - * Komania * Bangladesh * Nepal * Guatemala * Pakistan -0.10 5hana * Zambia, - *Cote d'lvoire o -0.15 - ~* Guyana Cia *Morocco *Philippines -0.20 - * Russia -0.25 - P Peru -0.30 - Gini 32 Figure 9. Gini vs CI (wasting) 0.10 * Vietnam * Cote d'lvoire 0.00 * Zambia nia * Egypt Indonesa* South Africa lia + ~~~Indonesia * *~~~~~~~ Pakistan * *Brazil -0.10 - Guatemal Bangladpiesh *Ghana *Peru *Morocco O -020 - *China -0.30- -0.40 -0.50 *Nicaragua 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0 Gini 33 Figure 10. Prevalence of under-5 stunting vs CI 0.00 10. % 20.0% 30.0% 40.0% 50.0% 60.0% 70. % * Egypt -0.05 - *-R-mania- -a- Bangiadesh - -- * Nepal * Pakistan * Guatemala * Ghana * Indonesia * Vietnam -0.10 - _*-Zambia * Cote d'lvoire o-0. 15 - * Guyana * China * Nicaragua * Philippines * Morocco -0.20 Brazil_ _ *Southfricaf - * Russia -0.25 - _ * Peru -0.30 Stunting (%) 34 Figure 11. Prevalence of under-5 underweight vs CI 0.00 - I I I I 0. % 10.0% 20.0% 30.0% 40.0% 50.0% 60. D% * Egypt -0.05 - - * Indonesia * Pakistan * Bangladesl * Russia * Romania Cte d'lvoire * Vietnam -0.10 - - _ _ _ _ _ _ _ _ -010- +* * Ghana * Guatemala Philippines * Nepal South Africa * Zambia -0.15 -__ _ _ _ ___ * China -0.20 -*-Guyana * Brazil -0.25 *-M-orocco * Nicaragua -0.30 __ ___ * Peru -0.35 Underweight (%) 35 Figure 12. Prevalence of under-5 wasting vs CI 0.10 * Vietnam * Cote d'lvoire 0.00 - Russia * Zambia * Egypt * South Africa * Romania P nakistan * Brazil Indonesa Bangladesh *Guatemala *Guyana * Np Bangladesh -0.10 - -- * Nepal - Philippines * Ghana * Peru *Morocco -0.20 -0.30 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ , ___ _ __ _ - - - -* - _ _ __ _ ___ _ _- __ _ _ _ -0.40 - Nicaragua -0.50 . . 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% Wasting (%) 36 Table 11. Achievement indices Stunting Underweight Wasting Country R R R Bangladesh 0.5126 0.5385 0.5405 0.5767 0.2241 0.2391 Brazil 0.1503 0.1795 0.0625 0.0776 0.0620 0.0662 China 0.2775 0.3170 0.1303 0.1519 0.0365 0.0438 Cote d'lvoire 0.1953 0.2165 0.1697 0.1868 0.1014 0.0994 Egypt 0.1710 0.1776 0.1073 0.1110 0.0483 0.0500 Ghana 0.3146 0.3442 0.2592 0.2863 0.0457 0.0520 Guatemala 0.6173 0.6656 0.3332 0.3696 0.0071 0.0077 Guyana 0.1186 0.1359 0.1867 0.2243 0.0750 0.0816 Indonesia 0.4616 0.5035 0.3982 0.4236 0.1031 0.1097 Morocco 0.2829 0.3351 0.1499 0.1876 0.0807 0.0943 Nepal 0.4865 0.5182 0.4654 0.5214 0.1324 0.1467 Nicaragua 0.1501 0.1739 0.0771 0.0984 0.0253 0.0379 Pakistan 0.4187 0.4498 0.4273 0.4548 0.2474 0.2630 Peru 0.3149 0.4031 0.1204 0.1574 0.0231 0.0267 Philippines 0.1417 0.1684 0.2142 0.2366 0.2230 0.2449 Romania 0.2406 0.2530 0.0692 0.0752 0.0498 0.0524 Russia 0.1683 0.2055 0.0624 0.0672 0.0636 0.0632 South Africa 0.2564 0.3073 0.1808 0.2065 0.0998 0.1031 Vietnam 0.5299 0.5766 0.4129 0.4496 0.0590 0.0546 Zambia 0.4822 0.5298 0.2241 0.2563 0.0559 0.0568 Policy Research Working Paper Series Contact Title Author Date for paper WPS2416 The Swiss Multi-Piilar Pension Monika Queisser August 2000 A. Yaptenco System: Triumph of Common Sense? Dimitri Vittas 31823 WPS2417 The Indirect Approach David Ellerman August 2000 B. Mekuria 82756 WPS2418 Polarization, Politics, and Property Philip Keefer August 2000 P. Sintim-Aboagye Rights: Links between Inequality Stephen Knack 37644 and Growth WPS2419 The Savings Collapse during the Cevdet Denizer August 2000 I. Partola Transition in Eastern Europe Holger C. Wolf 35759 WPS2420 Public versus Private Ownership: Mary Shirley August 2000 Z. Kranzer The Current State of the Debate Patrick Walsh 38526 WPS2421 Contractual Savings or Stock Market Mario Catalan August 2000 P. Braxton Development-Which Leads? Gregorio Impavido 32720 Alberto R. Musalem WPS2422 Private Provision of a Public Good: Sheoli Pargal August 2000 S. Pargal Social Capital and Solid Waste Daniel Gilligan 81951 Management in Dhaka, Bangladesh Mainul Huq WPS2423 Financial Structure and Economic Thorsten Beck August 2000 K. Labrie Development: Firm, Industry, and Asll Demirgug-Kunt 31001 Country Evidence Ross Levine Vojislav Maksimovic WPS2424 Global Transmission of Interest Jeffrey Frankel August 2000 E. Khine Rates: Monetary Independence and Sergio Schmukler 37471 the Currency Regime Luis Serven WPS2425 Are Returns to Investment Lower for Dominique van de Walle August 2000 H. Sladovich the Poor? Human and Physical 37698 Capital Interactions in Rural Vietnam WPS2426 Commodity Price Uncertainty in Jan Dehn August 2000 P. Varangis Developing Countries 33852 WPS2427 Public Officials and Their Institutional Nick Manning August 2000 C. Nolan Environment: An Analytical Model for Ranjana Mukherjee 30030 Assessing the Impact of Institutional Omer Gokcekus Change on Puolic Sector Performance WPS2428 The Role of Foreign Investors in Debt Jeong Yeon Lee August 2000 A. Yaptenco Market Development: Conceptual 38526 Frameworks and Policy Issues Policy Research Working Paper Series Contact Title Author Date for paper WPS2429 Corruption, Composition of Capital Shang-Jin Wei August 2000 H. Sladovich Flows, and Currency Crises 37698 WPS2430 Financial Structure and Bank Asli Demirg0c-Kunt August 2000 K. Labrie Profitability Harry Huizinga 31001 WPS2431 Inside the Crisis: An Empirical Asli Demirguc-Kunt August 2000 K. Labrie Analysis of Banking Systems in Enrica Detragiache 31001 Distress Poonam Gupta WPS2432 Funding Growth in Bank-Based and Asli Demirguc-Kunt August 2000 K. Labrie Market-Based Financial Systems: Vojislav Maksimovic 31001 Evidence from Firm-Level Data WPS2433 External Interventions and the Ibrahinm A. Elbadawi September 2000 H. Sladovich Duration of Civil Wars Nicholas Sambanis 37698