Why is inequality so high, but also so variable, in Sub Saharan Africa?

Transcript

1 Preliminary Draft August 2004 Why is inequality so high, but also so variable, in Sub-Saharan Africa? Edward Anderson and Andy McKay Poverty and Public Policy Group Overseas Development Institute London [email protected] [email protected] Abstract Levels of income and/or consumption inequa lity are on average at least as high in sub-Saharan Africa as they are in Latin America. They are also highly variable. The ain these findings us ing cross-country paper makes an initial attempt to expl econometrics. We find that around 50% of th e region’s high level of inequality can be attributed to its underlying factor endow ments, notably its high ratios of land to labour ratios, its sometimes high ratios of know n minerals to land ratios, and the large proportion of its land area which is located in the tropics. However, the same factors account for only a small proportion of the variation in levels of inequality within the region. A fuller explanation will require fu rther work, with the most promising route being case-study analysis.

2 2 1 Introduction When we think of the most unequal region in the world, we tend to think of Latin 1 shows that levels of inequality are as America. However, a quick glance at Figure high, on average, in sub-Saharan Africa. The measure of inequality shown is the Gini coefficient, but a similar result holds if we e of inequality, the look at another measur share of the poorest qu intile in national income (Figure 2). spent seeking to understand the various A considerable amount of effort has been reasons for the high levels of inequality witnessed in Latin American societies. The most recent example is the 400-page Worl d Bank (2003) study ‘Inequality in Latin story?’ There is also a widespread America and the Caribbean: breaking with hi essential pre-requisite for any sensible consensus that addressing inequality is an the region. The same cannot be said for long-term strategy for reducing poverty in sub-Saharan Africa. On the one hand, there have been fewer attempts to understand the causes of high levels of measured inequa lity in many African societies. On the other hand, there is much less agreement about the extent to which explicit efforts to tackle inequality need or should be part of strategies for reducing poverty in the region. This paper begins to fill these gaps in our knowledge, by asking why levels of inequality are so high in sub-Saharan Africa. portant task is to In doing so, an im average level of inequality but also the large amount of explain not only the high . Even among the low-income countries of the region, recent gini variability coefficients of income or consumption in equality vary from 0.33 in Ghana, which is low by international standards, to around 0.50 in Nigeria and Zambia, which is high by international standards, to above 0.60 in Malawi, which is exceptionally high by any standard (see next section). A large body of recent research has highlight ed the potentially adverse consequences of high inequalities for economic growth and poverty reduction. High inequalities reduces the impact of a given rate of economic growth on poverty, as has been shown by Ravallion (2001) and Hanm er and Naschold (2000) (a lthough these studies differ in their estimates of by exactly how much inequality affects the impact of growth).

3 3 They are also likely to lower the rate of growth itself. The basic motivation of this paper is that an improved understanding of gh average level of the reasons for the hi inequality in sub-Saha ran Africa, as well as its variabil ity, is essential if inequalities are to be tackled effectively. The evidence 2 In this section we review what we know about levels of inequality in individuals’ or societies relative to othe r regions. The difficulties households’ well-being, in African of compiling such estimates for a large number of countries means that the evidence is t some evidence on quality, but we presen limited mainly to income or consumption ine inequalities in education attainments and health status. Evidence on inequalities in bute to inequalities in well-b assets (e.g. land), which contri eing but are not aspects of well-being in themselves, are considered later (Section 4). Income/consumption inequality Table 1 shows the list of countries in sub-Saharan Africa contained in the Dollar and Kraay (2002) dataset, together with the most recent estimates for each one of the Gini coefficient and the share of the poorest qui ntile in national income, and information regarding the type of survey from which these figures are calculated. The majority vels across individuals. However, they all (80%) refer to inequality in consumption le ral are by now quite out of date. relate to different years, and seve Some of the difference between the average level of inequality in sub-Saharan Africa may reflect differences in the year in and other regions shown in Figures 1 and 2 which inequality is measured and in the type of survey on which estimates of e differences by estimating a cross-country inequality are based. We control for thes regression of the form: = β β β β β α + + INEQ SSA LAC SA EAP β ⋅ + + + EECA MENA + i i i i i i i 2 3 4 5 6 1 (1) 2 β β β + + + + + β β ε INC PER GRS YR YR , + 7 11 i i i i i i 10 9 8

4 4 where is the most recent estimate of in come or consumption inequality in INEQ i i ; are dummy variables equal SSA , EECA LAC , country SA , and EAP , MENA i i i i i i rica, Latin America and the Caribbean, South to 1 if the country is in sub-Saharan Af fic, Middle East and North Af rica or Eastern Europe and Asia, East Asia and the Paci INC Central Asia (as defined by the World Bank) respectively, and 0 otherwise; i income inequality, and 0 if it refers to equals 1 if the inequality estimate refers to PER consumption inequality; equals 1 if the inequality estimate refers to personal i GRS income or consumption, and 0 if it refers to household income or consumption; i equals 1 if the inequality estimate refers to gross income (befor e taxes and transfers), and 0 if it refers to net income (after taxes and transfers); is the year of the YR i is the residual term. ε inequality estimate, and i The results are shown in Table 2. When c ontrolling for the type and date of the tes, sub-Saharan Africa has the highest income or consumption inequality estima average Gini coefficient of all regions ( 20 points above the average in high-income countries), closely followed by Latin Ameri ca and the Caribbean (17 points above the average in high-income countries ). It also has the lowest average share of the poorest quintile in national income (3.3 percentage points below the average in high-income countries), again closely followed by afte r Latin America and the Caribbean (3.1 percentage points below the averag e in high-income countries). Some of the variability in inequality w ithin sub-Saharan Africa may also reflect differences in the type and year of the survey on which estimates are based. To control for this, we also examine the variance of the residuals from regression (1) across regions. The results, also shown in Table 2, show that th e amount of variation (as measured by the coefficient of variation) in levels of inequality within Sub- Saharan Africa remains high, relative to most other regions. When considering the income share of the poorest quintile, only Middle East and North Africa region shows more variation; when considering the Gini coefficient, only the Middle East and North Africa and Central Asia and East ern Europe show more variation.

5 5 ability of the inequal ity data in this as There are of course some questions the compar with any other cross country inequality study (see Atkinson and Brandolini, 2001, in dual country inequa the context of OECD countries). Indivi lity measures are ption, adjustments for price constructed using different a pproaches (income or consum holds and so on), and differences, whether the distribution is over individuals or house these differences can only be partly contro lled for in the above econometric analysis. lable for many countries, and the available In addition, inequality data is not avai years. The comparison though is based on estimates relate to a wide variety of estimates believed to be of high quality and broadly comparable (as assessed initially by Deininger and Squire, 1996, and updated by later authors). In other words, the e best data set curr ently available for this purposes. above comparisons are based on th two central issues addressed by this paper – Overall therefore, there it appears that the e variation of income and/or consumption inequality in the high average level and larg sub-Saharan Africa – are real features of the data Education inequalities Educational attainment is both itself and a determinant of an aspect of well-being in umption, via its effects on productivity and other dimensions of well-being (e.g. cons earnings). In this sense it is appropriate to consider here cross-country evidence qualities provided by Thomas et al. (1999). These authors relating to educational ine l attainment, measured by schooling years, calculate Gini coefficients for educationa among the population aged 15 and above, for 85 developed and developing countries between 1960 and 1990. Differences in this measure of educatio nal inequality between developing country regions are shown in Figure 3. The data rela se there is a clear te to 1990. In this ca difference between the two low-income regi ons SSA and SA – with high levels of educational inequality – and the two mi ddle-income regions LAC and EAP – with lower levels of educational inequality. Nevertheless, there is again a substantial amount of variation in levels of educational inequality within the SSA region. Health inequalities

6 6 ce on health inequalities]. [Insert discussion here of eviden 3 Understanding inequality in sub-Saharan Africa 3.1 Land-labour ratios Average land-labour ratios differ significa ntly between Latin America and sub- nd the Pacific on the Saharan Africa on the one hand, and South Asia and East Asia a e most obvious explanation other (Wood 2003, Table 1). Th ese differences provide th for the differences in average levels of inequality between re gions. Neo-classical economic theory predicts that countries w ith higher land-labour ratios will typically have higher levels of income inequality th ll else being equal), an other countries (a because land will account for a greater shar relative to labour, and e of national income 1 Support for this distributed asset than labour. because land is a less equally ound in recent empirical work (e.g. Leamer et al. 1999, prediction has been f et al. 1999). Spilimbergo There is also a large amount of variati on in land-labour ratios within sub-Saharan Africa: from 0.5 square kilometres of land pe r 100 adults in Rwanda to more than 50 square kilometres in Mauritania (Wood 2003, p. 166). It is also possible that this variation in levels of ine variation could account for the quality. Figure 4 shows that there is a clear positive rela tionship within the region betw een land-labour ratios and the Gini coefficient. affected by other factor endow Levels of inequality are also ment ratios. For instance, Leamer et al. (1999) show that levels of inequality te nd to decline as levels of human capital per worker (typically proxied by average years of schooling in the adult population) rise. This coul d explain why the average le vel of inequality in sub- 1 This need not necessarily be the case. The link between factor endowment ratios and the factor distribution of income depends on the various elasticities of substitution between factors in production. For instance, [in a two-factor model] a rise in the land-labour ratio will increase the share of land relative to labour in national income if th e elasticity of substitution between land and labour is greater than one, and reduce it if the elasticity is less than one. These elasticities are affected by technological change and by the extent of openness to trade. The link between the factor distribution of income and the distribution of income is also aff ected by inequalities in the ownership of assets (see Section 3.2).

7 7 Asia and the Pacific, where average years Saharan Africa is much higher than in East of schooling are much higher (Table 1, W ood 2003). However, it could not explain much higher than the level in South Asia, why the average level of inequality is also where average years of schooling are similar. Nor can it explain why the average level of inequality in sub-Saharan Africa is similar to that in Latin America, where ). Moreover, there is much less average years of schooling are much higher ( ibid. within sub-Saharan Africa which could variation in average years of schooling, account for the variation in levels of inequality within the region. correlated with ine One other factor endowment ratio which may be quality is the ratio ain, the hypothesis is of minerals (e.g. oil, metals, diamonds). Ag that countries with higher endowments of minerals per land area will tend to have higher levels of income inequality (all else being equal), because minerals will account for a greater rns to minerals are (like those to land) share of national income, and because the retu d. High measured inequality in some African countries typically unequally distribute (e.g. Nigeria, Sierra Leone, Botswana) may well reflect their large relative endowments of minerals. 3.2 Land inequalities The effect of high land-labour ratios on inco me inequality will be compounded if land ownership is highly unequal. Inequalities in land ownership are typically determined over time except during by historical factors, and change little or immediately after periods of substantial change (revolution, independence, war). But they do often vary across countries. It is ther efore possible that the high level of income/consumption inequality in sub-Saharan Af rica reflects particularly high inequality, and/or that variation in land inequality within the region account for the observed variations in income/consumption inequality. Estimates of land inequality in sub-Saharan Af rica are shown in Table 3. The data are taken from Deininger and Squire (1998) a nd IFAD (2001). They are clearly limited, in that estimates are availabl e for a subset of countries onl y 16 countries in the region only, and some of these are by now quite out of date. Notwithstanding this caveat, they show no evidence that inequalities of land ownership are sign ificantly higher in

8 8 Furthermore, there is little correlation sub-Saharan Africa than in other regions. between observed levels of la nd inequality and income inequality within sub-Saharan .g. Seychelles, Madagascar, and Tanzania) Africa. Some countries in the region (e have low levels of income inequality despite high levels of land inequality, while income inequality despite Leone) have high levels of others (e.g. Lesotho and Sierra low levels of land inequality. The lack of correlation is shown graphically in Figure 5. 3.3 Openness has been devoted to the issue of whether A large amount of attention in recent years migration – affects levels of inequality economic openness – to trade, investment, and is literature is prov ided by Anderson 2004). within countries. (A recent review of th s predictions regarding the effects of Economic theory does not provide unambiguou icts that, in countri es with high land- openness. Heckscher-Ohlin trade theory pred labour ratios and low skill-labour ratios, incr eased openness would raise the returns to land relative to labour, but re duce the returns to skilled relative to unskilled labour – leaving the effect on overall inequality uncertain. Do levels of openness in sub-Saharan Afri ca differ on average from those in other regions? The clearest difference in average levels of trade openness between regions, measured by the Sachs and Warner (1995) tr ade policy index, is between sub-Saharan Africa and South Asia on the one hand (less open), and Latin America and East Asia on the other hand (more open) (column 1, Table 4). Other measures of trade openness, such as the Frankel and Romer (1 999) predicted trad e share, differ only marginally between regions (column 2, Table 4). Furthermore, there is no apparent correlation between levels of trade openne ss measured in these ways and income inequality within Sub-Saharan Afri ca, as indicated by Figures 6 and 7. 3.4 Other Differences in political institutions between countries in the region may also explain differences in inequality. One hypothesis is that where the majority of citizens possess political and civil liberties, they are mo re able to prevent a rich minority from

9 9 l wealth. Evidence in support of this expropriating an excessive share of nationa (1998). The averag e level of political hypothesis has been found recently by Li et al. International is substantially higher rights as measured by the Freedom House (indicating less rights) in sub-Saharan Africa than it is in other regions (column 3 in Table 4). This could therefore account for the high levels of inequality in the region. However, differences in average levels of civil liberties between regions, which should also affect inequality, are smaller. Furthermore, there is no clear relationship between measured levels of political liberties within sub-Saharan Africa, as shown in Figure 8. It has been argued in previous ethnic diversity have had a work that high levels of sub-Saharan Africa, by contributing to negative impact on economic growth in Levine 1997; Mauro 1995). They may also conflict and corruption (e.g. Easterly and ls of inequality in the regi on and/or the have contributed to the high leve variations in inequality within it. Pressures for the redi stribution of assets and/or incomes may be higher and more successful when countri es are more ethnically homogenous. A common measure of ethnic dive rsity is the probabi lity that two individuals from a country will speak a different language. The av erage value of this va riable is in fact in South Asia (column 5, Table 4). no higher in sub-Saharan Africa than it is tionship between this variable and levels of inequality Moreover, there is no clear rela within sub-Saharan Africa, as shown by Figure 9. Finally, there is a geographical element to in ran Africa, in that equality in sub-Saha levels of inequality tend to be highest in Southern Africa, incl uding Namibia, South Africa, Lesotho, Botswana, Malawi and Zimbabwe. The relationship between countries’ and inequality, show n in Figure 10, is positive a nd marginally statistically significant. 3.5 Econometric analysis In this section we estimate the determinan ts of income/consump tion inequality using econometric analysis. We then examine the extent to which observable influences on inequality can account for the high average le vel of inequality in sub-Saharan Africa, as well as the large amount of variation within it. The explanatory variables are:

10 10 - the population aged 15-64, from World land-labour ratios (area divided by 2 Development Indicators), relative to the world average; the value of known (in 1990) metal, gas, coal and oil reserv es per unit of land - area, from Wood and Mayer (2001); average years of schooling in the adult population, fr om Barro and Lee (2000); - trade openness, as measured by the Sachs and Warner (1995) index of trade - mer (1999) predicted trade share; policy openness and the Frankel and Ro US$ PPP (from Dollar and Kraay 2002); - per capita GDP, and its square, in - distance from the equator, from Hall and Jones (1999); ethno-linguistic fragmentation, from - Easterly and Levine (1997); - index of political rights, from Freedom House International. We allow for the effect of land-labour ratios and average years of schooling on income/consumption inequality to vary with the level of trade openness, as neo- classical theory predicts. We estimate the equation using pooled cross- section and time-series inequality data r variables are measur used by Dollar and Kraay (2002). All othe ed for the closest year available. We include controls for the year and year squared, to allow for a potential common time trend ( linear or non-linear) in leve ls of inequality over the observations are derived. We period from which inequality also include controls for when income is measured net of taxes and transfers, and when inequality in consumption rather than income, or between households rather than individuals, is being measured. All variables are obs erved for a total of 163 country-year observations, of which 17 are from sub-Saharan Africa. We do not include land inequality in the regression, because th is limits the sample too much. The regression results when using the Gini coefficient as the dependent variable are shown in Table 5. Column (1) shows the av erage levels of the Gini coefficient in each region in this sample when controlling only for the year of observation and the 2 Ideally, we would measure each country’s factor endowment ratios relative to the effective world average, as Spilimbergo et al. (1999) do.

11 11 e remaining explanatory variables with the type of survey. Column (2) includes th exception of trade openness. Column (3 ) includes the Sachs and Warner (1995) acted with each of the factor endowment measure of trade openness, which is inter ratios. Column (4) includes the Frankel and Romer (1999) measure of trade openness the factor endowment ratios. instead, again interacted with each of The results are generally as expected. Cons idering first column (2), land-labour and e Gini coefficient, while capital-labour mineral-land ratios have positive effects on th variables are statistically significant at the ratios have a negative effect. All three s a positive effect on the Gini coefficient, 10% level. Ethno-linguistic diversity also ha tically significant. An improvement in civil liberties but the effect is not statis (represented by a fall in the Freedom House index) has a negative effect on the Gini, as expected, but the effect is not statis tically significant. Latitude has a negative effect on inequality – which is the oppos ite of the pattern observed within sub- Saharan Africa. There is a statistically si gnificant inverse-U shaped relationship – the famous Kuznets curve – between per capita income and the Gini coefficient. The results are quite similar when using the income share of the poorest 20% as the dependent variable (see Table 6). ents change little in size, although levels Turning to columns (3) and (4), most coeffici of statistical significance are sometimes lo wer. The interactions between average years of schooling and trade openness are positive, as expected, but they are not statistically significant. The interactions between land-labour and mineral-land ratios column (4), which is contrary to and trade openness are in fact negative in expectation, but again neither is statistically significant. Of particular interest is the size of the coefficient on the dummy for sub-Saharan Africa in each column. Comparing columns (1) and (2) suggests that roughly half of the higher average level of Gini coefficien ts in the region compared to developed countries can be accounted for by the addi tional explanatory variables included in column (2). Including trade openness and its inte raction with factor endowment ratios adds little if anything to this figure. The proportion of the lower average income share of the poorest 20% which can be expl ained is very similar (between 40% and 50%). Analysis of the residuals from th ese regressions suggests that the amount of

12 12 ub-Saharan Africa which can be explained by the variation in inequality levels within s the above variables is much smalle r, at between 10% and 20%. It seems therefore, that cross-country econometrics can provide only a partial explanation of the high but va riable levels of inequality in the region. What other factors are likely to be re sponsible? They are likely to include the amount of e amount of redistribution (or lack of it) carried out by inequality in land ownership, th ualities, and ethnic in equalities, all of governments, spatial inequalities, gender ineq ta been unable to include in the above which we have for reasons of lack of da these factors, and their own underlying regressions. Assessing the influence of ried out through case-study analysis. determinants, can most reasonably be car Case study evidence on inequality 3.6 A better understanding of the factors underlying high, and variable, levels of inequality in Africa is likely to need to be based on strong cas e study evidence. Survey data is now available for a large number of Africa countri es which can enable such an analysis, at a microeconomic le vel in terms of the explanatory factors typically available in such surveys. w studies of factors associated with the To date though there have been relatively fe countries, although there have been studies of levels of inequality in different African factors influencing changes in ular how these relate to inequality, and in partic economic reform – where there does not appear to be a systematic pattern (Christiansen, Demery and Paternostro, n.d.). This study drew on a number of country case studies using household survey data, and some of these provide more initial information about inequality with in countries, in particular comparing inequality levels across groups of hous eholds (e.g. by location and main economic activity – for instance Coulombe and McKa y, 2003, for Ghana). Other studies have used Demographic and Health Survey data to look at non-income dimensions of inequality (reported in Christ iansen et al, n.d.) and to l ook at urban-rura l inequality (Sahn and Stifel, 2002). A common finding from inequality decompositions in low income African countries is that only small proportions of inequali ty are explained by

13 13 the large majority of ine quality being among the groups inter-group inequality, with identified in these studies. 4 Conclusions and next steps Latin America is typically viewed as the region of the world in which inequalities esent the most pressing concern for policy- between rich and poor are highest, and repr makers and aid donors. This short paper high lights the fact that the average level of at least as high, if not higher, in sub- income and/or consumption inequality are ation within the region. The paper makes Saharan Africa, although there is much vari using cross-country econometrics. It finds an initial attempt to explain this finding gh level of inequality – around 50% – can be attributed to that much of the region’s hi its underlying factor endowments, notably its high ratios of land to labour ratios, its sometimes high ratios of known minerals to la rge proportion of its nd ratios, and the la land area which is located in the tropics. All these variables are shown in cross- country regressions are shown to be associat ed quite robustly with higher inequality. However, they account for only a small prop ortion of the variat ion in levels of inequality within the region, wh ich remains largely unexplained. These findings raise more questions than th actors account for that ey answer. What f part of the high average level of inequali ty in the region, and its variation, which econometrics is unable to explain? We specula to the distribution of te that they relate land ownership, the extent of government redi stribution (or lack of it), and of so- called ‘horizontal’ or inter-group inequalities. Th e most promising route of enquiry to assess the contribution of these factors to inequality in the region is through case- study analysis, as they are all difficult to measure in one let alone several different countries. From the point of view of policy, the firs t issue which needs exploring further is whether the case for addressing inequality explicitly is any way different in sub- Saharan Africa to the case in Latin America. The main difference, of course, is that Latin America is a predominantly middle-in come region, while sub-Saharan Africa is a predominantly low-income region. It can be and often is argued that low-income

14 14 countries and their governments do not have the resources for achieving much redistribution, unlike their middle-in come country counterparts. r is how the underlyi ng reasons for high Another issue which needs exploring furthe inequality in the region affect the choi ce of policies or instruments for addressing inequality. The results in this paper sugge st that high land-labour ratios, re lative to other regions are an important determinant of inequality in the region. The high level of these ratios relative to other regions is unlikely to change much in future years. The implication is that inequality in th e region may be highl y persistent without significant land reform or redistribution.

15 15 References Anderson, E. (2004). Openness and inequality in developing countries: a review of theory and recent evidence. Overseas Development Institute, London. Atkinson, A. and Brandolini, ... Barro, R. and Lee, J-W. (2000). ‘Interna tional data on educational attainment: updates and implications”. CID Working Paper No. 42, Center for International Development, Harvard University. Christiansen... Coulombe and McKay... Deininger, K. and Squire, L. (1996)... Deininger, K. and Squire, L. (1998). New ways of looking at old issues: inequality , 57: 259-287. Journal of Development Economics and growth. Dollar, D. and Kraay, A. (2002). Growth is good for the poor. Journal of Economic Growth , 7, 195-225. ica’s growth tragedy: Easterly, W. and Levine, R. (1997). ‘Afr policies and ethnic divisions’. , 107 (4), pp.1203.1250. Quarterly Journal of Economics 1999). ‘Does trade cause growth?’. Frankel, J. and Romer, D. ( American Economic Review , 89 (3), p.379-399. Hall, R.E. and Jones, C.I. (1999). ‘W hy do some countries produce so much more Quarterly Journal of Economics , 114 (1), p. 83-116. output than others?’. Attaining the International Development Hanmer, L. and Naschold, F. (2000). Targets: will growth be enough? Development Policy Review , 18 (1), 11-36. IFAD (International Fund for Agricultural Development) (2001). Rural Poverty Report 2001: The Challenge of Ending Rural Poverty. Oxford: Oxford University Press. Leamer, E., Maul, H., Rodriguez, S. and Sc hott, P. (1999). Does natural resource erican inequality? abundance increase Latin Am Journal of Development Economics , 59, 3-42. Li, H., Squire, L. and Zou, H. (1998). Explaining international and intertemporal Economic Journal , 108, 26-43. variations in income inequality. Mauro, P. (1995). Corruption and growth. Quarterly Journal of Economics, 681-712. Ravallion, M. (2001). Growth, inequality and poverty: looking beyond averages. World Development , 29 (11), 1803-1815. Sachs, J. and Warner, A. (1995). ‘Economic reform and the process of global integration’. Brookings Papers on Economic Activity , 1, p.1-118. Sahn and Stifel... Spilimbergo A., Londono J.L. and Szekely M. (1999). Income distribution, factor Journal of Development Economics , vol. endowments, and trade openness. 59, no. 1, pp. 77-101. Measuring education inequality: Gini Thomas, V., Wang, Y. and Fan, X. (1999). coefficients of education. World Bank: Washington D.C. Wood, A. (2003). Could Africa be like Am erica? Annual World Bank Conference on Development Economics 2003, 163-200. Wood, A. and Mayer, J. (2001). Africa ’s export structure in a comparative perspective. Cambridge Journal of Economics , 25 (3), pp. 369-394. World Bank (2003). Inequality in Latin America and the Caribbean: breaking with history? World Bank: Washington D.C.

16 16 Table 1 Income Income/ share of consum Person/ 1st -ption Gross/ house- quintile survey net Country Gini Year hold Burundi 33.3 7.9 1992 C P N 39.0 5.5 1995 C P N Burkina Faso Botswana 54.2 3.6 1986 C H N 61.3 2.0 1993 Central African Republic P N C Cote d'Ivoire 38.0 7.1 1995 C P N 49.0 . 1983 C P N Cameroon 38.1 . 1996 C P N Djibouti Ethiopia 44.2 7.1 1996 C P N Gabon 63.2 2.9 1977 Y H N Ghana 32.7 8.4 1997 C P N 40.4 6.4 1995 C P N Guinea 47.8 4.4 1992 C P n.a. Gambia, The Guinea-Bissau 56.1 2.1 1991 C P N Kenya 57.5 5.0 1994 C P N Lesotho 57.9 2.6 1993 C P N 43.4 5.8 1993 C P N Madagascar 50.5 4.6 1994 C P N Mali Mozambique 39.6 6.5 1996 C P N Mauritania 38.9 6.2 1995 C P N Mauritius 36.7 6.7 1991 C P N 62.0 . 1993 C P N Malawi 74.3 1.5 1993 C P N Namibia Niger 50.5 7.5 1995 C P G Nigeria 50.6 4.4 1997 C P N Rwanda 28.9 9.7 1984 C P N Sudan 40.0 5.6 1969 Y P G Senegal 41.3 6.4 1994 C P N Sierra Leone 62.9 1.1 1989 C P G Seychelles 47.0 . 1984 C P N Chad 35.0 8.0 1958 Y P G Tanzania 38.2 6.8 1993 C P N Uganda 39.2 6.6 1993 C P N South Africa 59.3 2.9 1994 C P G Zambia 49.8 4.2 1996 C P N Zimbabwe 56.8 4.0 1990 C P N Source: Dollar an d Kraay (2002)

17 17 Table 2 Unadjusted Unadjusted Adjusted Adjusted Average Coefficient of Average Coefficient of level* variation** level* variation** Gini coefficients Sub-Saharan Africa 0.23 19.7 0.22 13.8 5.0 East Asia and Pacific 9.5 0.17 0.16 South Asia 1.2 0.08 7.8 0.07 Latin America and the Caribbean 14.2 0.15 17.0 0.14 Middle East and North Africa 6.8 0.29 13.4 0.28 Eastern Europe and Central Asia 0.7 0.22 -9.5 0.48 Income share of poorest quintile (%) Sub-Saharan Africa -1.5 0.42 -3.3 0.39 East Asia and Pacific 0.0 0.27 -1.5 0.19 South Asia 1.6 0.08 -0.5 0.09 Latin America and the Caribbean -2.3 0.41 -3.1 0.27 Middle East and North Africa -0.5 0.42 -2.7 0.50 Eastern Europe and Central Asia 0.5 0.30 2.3 0.21 Notes: *Relative to high-income countries. ** Defined as the standard deviation divided by the mean.

18 18 Table 3 Income/ consumption Gini Land Gini Country Year Decade Source coefficient coefficient Cote d'Ivoire 38 1995 42.29 1970s IFAD Ethiopia 44.2 1996 47.01 1980s IFAD Guinea 40.4 1995 50.99 1980s IFAD Kenya 57.5 1994 77 1980s IFAD Deininger & Squire Lesotho 57.94 1993 36.2 1970s Madagascar 43.44 1993 80 1980s IFAD Deininger & Mali 50.5 1994 47.76 1960s Squire Mauritania 38.9 1995 58.58 1980s IFAD Deininger & Squire Niger 50.5 1995 31.77 1980s Sudan 40 1969 57.65 1960s IFAD Deininger & Squire Senegal 41.28 1994 49.27 1960s Sierra Leone 62.9 1989 47.74 1980s IFAD Seychelles 47 1984 82.06 1970s IFAD Tanzania 38.2 1993 78.99 1980s IFAD Deininger & Squire Uganda 39.2 1993 58.96 1990s Deininger & 1994 70.1 1960s South Africa Squire 59.3 No. of Regional Land Gini observations averages: Sub-Saharan Africa 57.3 16 East Asia and Pacific 52.1 10 South Asia 57.2 5 Latin America and the 78.3 25 Caribbean

19 19 Table 4 Sachs Freedom Freedom Frankel and House and House Romer index of index of Warner (1999) civil (1995) political Ethno- predicted liberties rights trade linguistic policy trade (1=highest, (1=highest, 7=lowest) measure share diversity 7=lowest) Sub-Saharan Africa -Mean 2.9 4.7 4.8 0.65 0.35 26 35 33 33 30 -No. of obervations East Asia and Pacific -Mean 0.67 4.3 3.9 0.55 2.3 9 6 13 13 -No. of obervations 10 South Asia 0.40 2.2 4.4 2.8 0.67 -Mean -No. of obervations 5 5 5 5 4 Latin America and the Caribbean -Mean 0.86 2.8 2.8 2.7 0.28 -No. of obervations 22 26 23 23 23

20 20 Table 5 1 2 3 4 (Constant) 66.07 -148.81 -168.57 -144.23 5.5 -3.2 -3.2 -3.1 1.39 1.37 Land-labour ratio 1.48 4.9 2.2 1.2 Average years of schooling -7.78 -2.54 -3.03 -1.9 -2.0 -2.1 Minerals-land ratio 0.45 3.09 1.62 0.2 0.7 1.7 -0.97 -1.87 Trade openness -0.7 -2.1 Land-labour ratio*trade openness -0.07 0.06 0.1 -0.2 2.05 Capital-labour ratio*trade openness 1.84 0.8 1.6 Minerals-land ratio*trade openness -0.56 1.41 0.6 -0.4 Latitude -0.11 -0.12 -0.13 -2.5 -2.7 -3.0 2.34 2.69 Ethnolinguistic diversity 0.90 1.2 1.4 0.5 0.26 0.07 Civil liberties (1=highest, 7=lowest) 0.27 0.7 0.7 0.2 Log average income (1985 US$ PPP) 52.99 58.17 53.95 4.9 4.6 5.0 Per capita income squared -3.33 -3.63 -3.40 -5.0 -4.6 -5.0 Income survey (1=yes, 0=no) 4.65 6.96 7.43 6.91 3.5 5.3 5.2 5.6 -2.58 -2.64 -2.90 Person survey (1=yes, 0=no) -2.79 -2.3 -2.2 -2.2 -2.4 Gross income (1=yes, 0=no) 1.73 1.40 1.00 2.52 2.1 1.4 1.1 0.7 -2.46 Years since 1950 -1.84 -1.81 -1.76 -3.5 -2.8 -2.8 -2.9 Years since 1950 squa 0.04 0.03 0.03 0.03 red 3.8 3.2 3.3 3.3 Sub-Saharan Africa 8.52 8.47 9.56 17.35 9.8 3.5 3.3 3.9 East Asia and Pacific 9.66 1.18 1.26 0.39 6.3 0.5 0.5 0.2 6.78 -0.90 -2.27 South Asia -0.70 3.8 -0.3 -0.3 -0.9 Latin America and the Caribbean 5.46 5.52 4.40 15.77 12.7 2.8 2.4 2.2 Middle East and North Africa 11.70 2.75 3.04 5.57 4.2 0.9 1.0 1.7 Adjusted R2 0.657 0.739 0.735 0.748 Notes: Dependent variable is the Gini coefficient. T-Statistics are shown below each coefficient.

21 21 Table 6 1 2 3 4 (Constant) -0.69 25.75 33.82 26.06 -0.2 2.3 2.6 2.2 -0.43 -0.25 Land-labour ratio -0.43 -5.9 -2.7 -0.9 Average years of schooling 0.47 0.14 0.31 0.4 0.9 0.5 Minerals-land ratio -0.87 -1.18 -0.55 -1.7 -1.1 -2.4 0.09 -0.07 Trade openness 0.3 -0.3 Land-labour ratio*trade openness -0.02 -0.07 -0.1 -0.7 Capital-labour ratio*trade openness -0.65 -0.12 -1.2 -0.4 Minerals-land ratio*trade openness 0.41 0.22 0.7 0.6 Latitude 0.03 0.03 0.03 2.8 2.9 2.8 0.09 0.09 0.06 Ethnolinguistic diversity 0.2 0.2 0.1 Civil liberties (1=highest, 7=lowest) 0.08 0.07 0.08 0.8 0.7 0.8 Log average income (1985 US$ PPP) -6.91 -9.08 -6.91 -2.6 -2.9 -2.5 Per capita income squared 0.58 0.44 0.44 2.7 3.0 2.6 -1.32 -1.39 -1.32 Income survey (1=yes, 0=no) -0.86 -2.7 -4.1 -4.2 -4.0 0.84 0.97 Person survey (1=yes, 0=no) 0.95 0.91 2.9 3.4 3.1 3.2 Gross income (1=yes, 0=no) -0.27 -0.25 -0.32 -0.71 -2.6 -0.9 -0.8 -0.9 0.53 0.46 0.47 0.45 Years since 1950 3.2 2.9 3.0 2.8 Years since 1950 squared -0.01 -0.01 -0.01 -0.01 -3.4 -3.3 -3.3 -3.1 -2.61 -1.55 -1.33 -1.47 Sub-Saharan Africa -6.2 -2.6 -2.1 -2.4 -1.17 -0.37 East Asia and Pacific -0.25 -0.15 -3.2 -0.7 -0.3 -0.4 South Asia 0.11 0.56 0.88 0.74 0.3 0.9 1.3 1.1 Latin America and the Caribbean -2.69 -1.08 -0.76 -1.01 -9.1 -2.2 -1.3 -2.0 Middle East and North Africa -1.46 -0.47 -0.20 -0.29 -2.2 -0.6 -0.3 -0.4 Adjusted R2 0.558 0.64 0.635 0.633 Notes: Dependent variable is the share of income r eceived by the poorest 20%. T-Statistics are shown below each coefficient.

22 22 Figure 1 80.00 70.00 60.00 50.00 40.00 Gini in most recent year 30.00 20.00 Latin America and Sub-Saharan South Asia East Asia and Pacific the Caribbean Africa Region in developing world Notes: The data are taken from Dollar and Kraay (2002 ), who update and extend of the authoritative Deininger and Squire (1996) database. They include only ‘high-quality’ observations: those which are based on nationally representative surveys, which cover all aspects of households’ or individuals’ income or consumption. Figure 1 shows the most recent observation for each country.

23 23 Figure 2 12.00 Suriname 10.00 8.00 6.00 4.00 2.00 Income share of poorest 20% in most recent year 0.00 Latin America and South Asia East Asia and Sub-Saharan Africa Pacific the Caribbean Region in developing world Notes: As Figure 1.

24 24 Figure 3 100.0 80.0 60.0 40.0 Gini coefficient of education attainment (Thomas et al) 20.0 Latin America and East Asia and Sub-Saharan South Asia Pacific the Caribbean Africa Region in developing world

25 25 Figure 4 80.00 Namibia 70.00 Gabon Sierra Leone Central African Repu Malawi South Africa 60.00 Lesotho Guinea-Bissau Kenya Botswana Zimbabwe Nigeria Niger Cameroon 50.00 Gambia, The Mali Zambia Ethiopia Madagascar Senegal Guinea Sudan Uganda Mauritania 40.00 Burkina Faso Mauritius Gini in most recent year Tanzania Djibouti Burundi Ghana Rwanda 30.00 20.00 -2.00 0.00 -1.00 1.00 2.00 -3.00 3.00 4.00 Land-labour ratio (log units, relative to world average) Notes: Figure shows 33 countries in sub-Saharan Africa with data on the Gini coefficient and the land- labour ratio in the corresponding year. The linear correlation coefficient between the two variables is statistically significant at the 5% level.

26 26 Figure 5 65.00 Sierra Leone South Africa 60.00 Lesotho Kenya 55.00 Niger Mali 50.00 Seychelles 45.00 Ethiopia Madagascar Gini in most recent year Senegal Sudan Guinea 40.00 Mauritania Tanzania Cote d'Ivoire Uganda 35.00 30.00 60.00 50.00 40.00 80.00 70.00 90.00 Gini coefficient of land ownership (average over decade) Notes: Y-axis shows the Gini coeffici ent for income/consumption in mo st recent year (years are those in Table 1). X-axis shows the Gini coefficient for land ownership in the most recent d ecade prior to the observation for income/consumption inequality.

27 27 Figure 6 80.00 Namibia 70.00 Sierra Leone Malawi Central African Repu South Africa 60.00 Lesotho Zimbabwe Guinea-Bissau Kenya Botswana Nigeria Mali Cameroon 50.00 Gambia, The Niger Ethiopia Seychelles Guinea Sudan Senegal Mozambique 40.00 Djibouti Cote d'Ivoire Gini in most recent year Burkina Faso Mauritania Tanzania Burundi Ghana Rwanda 30.00 20.00 3.50 3.00 4.00 4.50 2.50 2.00 Predicted trade share (Frankel and Romer 1999)

28 28 Figure 7 Sierra Leone Gabon Malawi 60.00 South Africa Zimbabwe Kenya Guinea-Bissau Botswana Nigeria Mali Cameroon 50.00 Zambia Gambia, The Ethiopia Senegal Guinea Burkina Faso 40.00 Cote d'Ivoire Mauritania Tanzania Gini in most recent year Mauritius Burundi Ghana 30.00 Rwanda 0 1 0.4 0.8 0.2 0.6 Sachs and Warner trade policy measure

29 29 Figure 8 80.00 Namibia 70.00 Gabon Central African Repu Malawi South Africa 60.00 Lesotho Guinea-Bissau Kenya Botswana Zimbabwe Mali Nigeria Zambia Cameroon 50.00 Gambia, The Niger Seychelles Madagascar Ethiopia Senegal Guinea Mozambique Burkina Faso 40.00 Cote d'Ivoire Mauritius Gini in most recent year Mauritania Tanzania Djibouti Burundi Ghana Rwanda 30.00 20.00 234567 Index of political rights (1=highest, 7=lowest)

30 30 Figure 9 Sierra Leone Gabon Central African Repu Malawi South Africa 60.00 Lesotho Kenya Zimbabwe Botswana Nigeria Niger 50.00 Cameroon Mali Zambia Gambia, The Ethiopia Madagascar Senegal Mozambique Uganda 40.00 Mauritania Guinea Cote d'Ivoire Sudan Mauritius Tanzania Gini in most recent year Chad Burundi Ghana 30.00 Rwanda 0.00 0.60 1.00 0.80 0.40 0.20 Ethnolinguistic diversity Notes: The linear correlation coefficient between the two variables is 0.21, but is statistically insignificant at the 10% level.

31 31 Figure 10 80.00 Namibia 70.00 Gabon Sierra Leone Malawi 60.00 Central African Repu South Africa Kenya Guinea-Bissau Lesotho Botswana Zimbabwe Niger Mali Cameroon 50.00 Gambia, The Seychelles Ethiopia Madagascar Senegal Guinea Mozambique Uganda 40.00 Djibouti Cote d'Ivoire Mauritius Gini in most recent year Burkina Faso Mauritania Tanzania Burundi Ghana Rwanda 30.00 20.00 15.00 10.00 25.00 20.00 5.00 0.00 30.00 Latitude (absolute value) Notes: The linear correlation coefficient between the two variables is 0.25, and is statistically significant at the 15% level.

Related documents

catalog e

catalog e

How to Buy. Wright Tool is proud to partner with thousands of distributors world wide to bring our products to tool users where they need them, Visit us at and when they need them. To locate a distrib...

More info »
NCDOT Current STIP

NCDOT Current STIP

May 2019 NCDOT Current STIP

More info »
Threads and Threading

Threads and Threading

Machinery's Handbook 27th Edition TABLE OF CONTENTS THREADS AND THREADING METRIC SCREW THREADS SCREW THREAD SYSTEMS 1783 American Standard Metric Screw 1725 Screw Thread Forms V-Thread, Sharp V-thread...

More info »
Main Street Marketplace Complete Application August 2018

Main Street Marketplace Complete Application August 2018

Main St. Marketplace Lot One Housing Project Application for Major Site Review, Conditional Use & Variance to Maximum Setback Submitted to: Town of Carbondale 511 Colorado Ave. Carbondale, CO 86123 Pr...

More info »
MCO 1200.17E MILITARY OCCUPATIONAL SPECIALTIES MANUAL (SHORT TITLE: MOS MANUAL)

MCO 1200.17E MILITARY OCCUPATIONAL SPECIALTIES MANUAL (SHORT TITLE: MOS MANUAL)

DEPAR T MENT THE NAVY OF ADQ UARTE UNI T ED ST ATE S MAR INE CORPS HE RS RINE COR N PS PENT 3000 MA AGO 20350-3000 NGTON, HI D.C. W AS 7E 00 .1 12 MCO 465 c AUG 0 8 013 2 ORDER 1200.17E MARINE CORPS C...

More info »
u7112 connectplus directories 2019

u7112 connectplus directories 2019

UCare Connect + Medicare Provider and Pharmacy Directory Introduction This Provider and Pharmacy Directory includes information about the provider and pharmacy types in UCare Connect + Medicare and li...

More info »
U7112 UCARE CONNECT + MEDICARE PROVIDERDIR MAY 2019 DATA.sv

U7112 UCARE CONNECT + MEDICARE PROVIDERDIR MAY 2019 DATA.sv

UCare Connect + Medicare Provider and Pharmacy Directory Introduction This Provider and Pharmacy Directory includes information about the provider and pharmacy types in UCare Connect + Medicare and li...

More info »
Nastran Dmap Error Message List

Nastran Dmap Error Message List

Overview of Error Messages NX Nastran displays User Information, Warning, and Error messages in the printed output. The amount of information reported in a message is controlled by system cell 319. Wh...

More info »
C:\Users\aholmes4\AppData\Roaming\SoftQuad\XMetaL\7.0\gen\c\H5515 ~1.XML

C:\Users\aholmes4\AppData\Roaming\SoftQuad\XMetaL\7.0\gen\c\H5515 ~1.XML

H. R. 5515 One Hundred Fifteenth Congress of the United States of America AT THE SECOND SESSION Begun and held at the City of Washington on Wednesday, the third day of January, two thousand and eighte...

More info »
dbi

dbi

REGISTERED DEPUTY BUILDING INSPECTORS LICENSE TYPE LICENSE NO. EXPIRE DATE DEPUTY BUILDING INSPECTOR P002118 ABBATOYE, CHRIS J. CASTAIC, CA 91384 (661) 312-9347 CONCRETE 09/01/2019 CONSTRUCTION 03/12/...

More info »
Operator Directory Listing

Operator Directory Listing

OPERATOR'S DIRECTORY SORTED BY OPERATOR NAME DATA SUPPLIED BY: FORM 1006 B CURRENT AS OF: Tuesday, April 16, 2019 Please notify Surety Department at ( 405 ) 521-2273 of any corrections or omissions th...

More info »
Los Angeles County Official Court Reporter Voicemail Directory

Los Angeles County Official Court Reporter Voicemail Directory

Los Angeles County Official Court Reporter Email and Voicemail Directory CSR No. Voice Mailbox Name 7925 1001 Tracy Aaron Hunt 1540 Flor Abadilla 9948 Sheri Abraham 3868 1003 Joy Adams 1005 8385 1025 ...

More info »
Los Angeles County Official Court Reporter Voicemail Directory

Los Angeles County Official Court Reporter Voicemail Directory

Los Angeles County Official Court Reporter Email and Voicemail Directory Name CSR No. Voice Mailbox 7925 Tracy Aaron Hunt 1001 Flor Abadilla 9948 1540 Sheri Abraham 3868 1003 8385 1005 Joy Adams Phyll...

More info »
old saybrook survey report july 2015

old saybrook survey report july 2015

HISTORIC AND ARCHITECTURAL PORTIONS OF THE RESOURCE SURVEY OF TOWN OF OLD SAYBROOK, CONNECTICUT The Town of Old Saybrook, as depicted on an 1893 atlas map s: ject Historian Pro David L. Taylor, M. A.,...

More info »
MSC MEPC.6 Circ.17 Annex (SOPEP)   31 January 2019

MSC MEPC.6 Circ.17 Annex (SOPEP) 31 January 2019

MSC - 1 7 MEPC.6/Circ. e 1 Annex, pag ANNEX LIST OF NATIONAL OPERATIONAL CONTACT POINTS RESPONSIBLE FOR THE RECEIPT, TRANSMISSION AND PROCESSING OF URGENT REPORTS ON INCIDENTS INVOLVING HARMFUL SUBSTA...

More info »
PowerPoint Presentation

PowerPoint Presentation

Deep Learning for Intelligent Video Analysis - Part II Tao Mei, Senior Research Manager Cha Zhang, Principal Applied Science Manager Microsoft AI & Research

More info »
wheretogo

wheretogo

WHERE TO GO WHEN YOU NEED 2018 EDITION HELP

More info »
OperatorHoursReport

OperatorHoursReport

John Bel Edwards Rebekah E. Gee MD, MPH SECRETARY GOVERNOR State of Louisiana Louisiana Department of Health Office of Public Health Certified Water and Wastewater Operators 2018 - 2019 Hours Hours li...

More info »