POLICY RESCARCH WORKING PAPER 1246 Determinants of Cross- An altemative hypothesis to explain why income Country Income Inequality inequality differs among countries. Inequality in richer An "Augmented" Kuznets' Hypothesis societies decreases not only An "Augmented" Kuznets' Hypothesis because of economic factors but also because societies Branko Milanovic choose less inc" jality as they grow richer. The World Bank Policy Research Department Transition Economies Division January 1994 1)()1 RSIAR(:I W(OxON; PAIPR 1246 Summary findings Why does inCcom inequality differ amiong countrecs? sector cnmploymniit, both of which reduce inequality. For Usinig a sample of 80 couintrics from thc 198(0s, this satmple, the reduction amounts to about a quarter of Milanovic shows that two types ot factors explain "given" inequality. variations in incomec incquality. The importance of sociaJ-choice factors rises as the The first are factors that are, in the short term, level of incom11e rises. The divergence between actual indepceident of cconoimic policies and are includcd in thf inequality and the inequality predicted by the standard sranidard formulation of the Kuzncts' curve: the level of Kuzncts' curve therefore systematically widens as a per capita inconic and thc country's regional societ; develops. heterogeincity. From the viewpoint of economic policy, The discrepancy is systematic, Milanovic contends. these are "given" factors, resulting in a "givenI Inequality in richer societies decreases not only because incquality." of econormiic factors but also bccause societies choose less The second group of factors are the social-choice inequality as they grow richer. factors reflected in the size of social transfers and of statc This paper - a product of the Transition Economics Division, Policy Research D)epartment - is part of a larger effort in the department to study detcrminants ot income distribution and poverty. Copies of the paper are available free from the World Bank, 1818 H Strect NW, Washington, DC20433. Please conitact Febecca Martin, room NI 1-043, extension 39065 (62 pages). January 1994. 7'he Policy Research Working Paper Series disse'ninates the fininJigs of wuork in progress to encourage the exchange of ideas about dezilopmeni issues. 4n objectire of the senes is " get the findings out quickly, even if the presentations are tess than fu41y pilished T7he papers carry the narnes o/the authors and should be sed and cited accordingly. 'ihe findings, intrerpretat ions, and conclusions are the authors' owrn and should not be atributed te the World Bank, its Lxecutite Board of Directors, or any of its member countries. P'roduced bv the Policv Research Dissemination Center Determinants of Cross-Country Income Inequality An "Augmented" Kuznets' Hypothesis Branko hilanovic DETERMINANTS OF CROSS-COUNTRY INCOME INEQUALITY: AN "AUGMENTED" KUZNETS' HYPOTHESIS Branko Milanovic I TABLE OF CONTENTS Section 1. Introduction 1 Section 2. The Background: the Kuznets' Relationship 1 Section 3. A New Hypothesis 3 Section 4. Testing the New Hypothesis 8 The Data 8 Empirical Analysis 12 Is Asia Different? 23 What Explains the Differences in Inequality? 26 How Important are Social Factors? 29 Section S. Conclusions and the Implications of the Findings 33 References 36 Annexes Annex Table 1. State sector employment as percentage of labor force or economically active population 40 Annex Table 2. Social transfers (cash and in-kind) as percentage of GDP 46 Annex Table 3. Within-country regional heterogeneity (ratio of incomes between most developed and least developed region) 51 Annex Table 4. The Gini coefficients 56 Annex Table 5. Income data 62 1I would like to thank Yvonne Ying and Vesna Petrovic for excellent research assistance. I would also like to thank Annette Brown, Alan Gelb, Bill Easterly, Ravi Kanbur, Klaus Schmidt-Hebbel, Gur Ofer, Milan Vodopivec and Mike Walton for very useful comments. 1. Introduction This paper presents an alternative hypothesis why income inequality differs between the countries. The only currently existing hypothesis was formulated by Kuznets (1955). Kuznets' hypothesis is briefly reviewed in Section 2. It provides an indispensable background to our "augmented" Kuznets' hypothesis which is formulated in Section 3. The empirical assessment of our hypothesis is presented in Section 4. The hypothesis is t:sted on a cross-sectional sample of 80 countries including all OECD countries, all European (former) socialist countries, and 50 A^rican, Asian, and Latin American countries. The data are from the 1980s. Section 5 spells out the main conclusions and implications of our hypothesis. 2. The Background: the Kuznets' Relationship When it comes to factors that explain differences in size income distribution between the countries, there exists only one broad hypothe.sis, proposed almost 40 years ago by Simon Kuznets (1955). It became famous as the Kuznets' inverted U curve. The hypothesis states that at very low levels of income, income inequality must also be low, as practically everybody lives at, or close to, subsistence level. There is no room for increased inequality because <.' the small size of overall output increased inequality would push many people below the subsiatence level. As the process of growth begins, income inequality increases. People migrate from the traditional agricultural sector where incomes are low to the modern industrial sector where both the (expected) wage is higher and wage differentiatior. is greater. Kuznets' model is thus also consistent with the Lewis-type pattern of growth. At the early stage of development, both physical and human capital are scarce and unequally distributed (that is, heavily concentrated among the few), and owners of human and physical capital are able to command high returns. As the two types of capital accumulate and become more diffused among the population, the rate of return on the physical capital declines while wage differentials between skilled and unskilled labor diminish. Incor,ae distribution becomes more equal. The process 'v.as summarized as follows by Kuznets (1966, p. 217): "It seems plausible to assume that in the process of growth, the earlier periods are characterized by a balance of counteracting forces that may have widened the inequality in the size distribution of total income for a while because or the rapid growth of tl- non-A [non-agricultural] sector and wider inequality within it. Tt is even more plausible to argue that the recent nirrowing in income inequality observed in the developed countries was due to a combination of the narrowing inter-sectoral inequalities in produc, per worker, the decline in the share of property incomes in total incomes of households, and the institutional changes that reflect decisions concerning social security and full employment." Ktuznets' empirical relationship has been extensively studied in both the cross-country and inter-temporal contexts. It remains the subject of controversy.2 The controversy has centered on: (1) the very existence of the relationship (it was argued that the Kuznets relationship critically depends on Latin American countries which are at an intermediate stage of develbpment, and for reasons peculiar to them, exhibit high inequality),3 (2) its validity for different countries and regions,4 and (3) its validity for different epochs. Kaelble and Thomas (1991, p.32) have recently thus summarized the empirical results of the Kuznets hypothesis: "Incomne levels explain cnly a small part of the variance of the inequality measures. This suggests that national characteristics (whether in terms of economic structure, political institutions, socio-cultural heritage, or whatever) play an important part in determining exactly what level of inequality is to be found at any particular level of modernization." No comprehensive alternative hypothesis regarding determinants of income inequality has so far been suggested, however. 2Reviews of theory and evidence on the Kuznets curve are extremely numerous. A particularly useful subset would include Lindert and Williamson (1985), Kaelble and Thomas (1991), Williamson (1991a), Polak and Williamson (1991), Paukert (1973), aid Lecaillon et al. (1984). Williamson ""91) provides a useful summary of the country studies and tries to determine if there is historical evidence for the Kuznets curve in Great Britain; Dumke (1991), Soderberg (1991), and Thomas (1991) in the same volume do the same thing respectively for Germany, Sweden, and Australia. Ram (1991) applies the Kuznets hypothesis to the states of the Us. 3See, for example, recent criticism by Atldnson and Micklewright (1992, p.35). 4For the denial of its validity in Asia, see Ushima (1991, p.121); for the absence of the Kuznets curve in Japan, see Lindert and Williamson (1985, p.354). s/ 2 It is worth pointing out, in light of the alternative hypothesis proposed here, that the Kuznets' hypothesis puts at center stage the role of economic factors, that is, of the supply of, and demand for, various factors of prodcCLAun.S The forces of economic development determine the shape of income distribution. Societies do not choose the income distribution that they would like to have. T}.' process is led by inexorable economic forces, and deviations frorr. the income distribution that a country must have at a certain level of development are small and non- systematic. 3. A New Hypothesis Here, I propose an "augmented" Kuznets' hypothesis. I argue that size income distnbution is determined (1) by factors that are in the short-run, from the point of view of policy makers or society as a whole, "given", and (2) by social (or public policy) choice. The "givens" are (1) the level of income and (2) the regional heterogeneity of a country. Neither of these factors can be influenced strongly in the short-run. The level of development (level of ir.come) is obviously a variable that changes slowly; so is, and for the same reasons, the inherited regional inequality. No amount of government redistribution will transform, in a few years, Sicily into Lombardy, nor, in the former Soviet Union, Kyrghzystan into Estonia. The public policy factors are (1) the percentage of workers employed in the state and the para-statal sector, and (2) the extent of government transfers, measured as a share of a country's GDP. These two factors are the products of political decisions, both current and past (e.g. a country might have a large s. .te sector because of a strong past influence of socialist parties). In the empirical section that follows, I will address two key questions: (1) Are social choice factors statistically significant "explanators" of cross-country income inequality? and (2) If so, how large is their influence? The "given" factors are not new. They have already been included in the numerous studies of cross-country income inequality. This applies not only ,o income as in the strong variant of the Kuznets hypothesis where income alone determines income inequality, but also to regional 'I use thc qualifier "at center stage" because Kuznets' was indeed aware, as the earlier quotation make-s clear, of the role of institutional factors in income distribution. / 3 heterogeneity. The point was made in earlier studies that the heterogen'eity of the country will have an impact on income inequality. The total population (Pryor, 1973, pp. 83ff.) or the geographical size of a country were used as control variables, assuming that larger or more populous countries will tend to be more heterogeneous. These assumptions are dubious. For example, in the former Yugoslavia, equal in size to the state of Oregon and in population to California, the ratio of per capita income between the richest and the poorest republic was almost 8:1, whereas in the much larger United States the ratio between the richest and tUe poorest state in 1980 was only 2:1. More exact indicators than geographic size would need to be use4 to reflect a country's heterogeneity. In the empirical part of the paper, I use, for each country, the ratio in average income between its richest and its poorest territorial unit (state, republic, province, ldnder in the case of federal states; prefectures, counties, etc. in the case of unitary states).6 The heterogeneity of the country, however, requires special attention. If we consider regional difference as a datui1m, in the sense that i, reflects iong-standing and slow-changing features of different regions that are not significant y influenced in the short-run by social policy, the inclusion of regional heterogeneity as an eWxplanatory variable is appropriate. Thus, if we take the former Soviet Union or Brazil as examples, it could be argued that, everything else being the same (income, social transfers, state-sector employment etc.), these courtries could be expected to have a more unequal income distribution than some others, such as France or Sweden, owing to historically different regional income levels. One would also expt-ct that this year's social policy (or that of the last several years) would have almost no effect on the ratio of average incomes between (say) Russia and Tajikistan, and Sao Paolo and Rondonia. If nothing else changes except that a country splits up, as happened with the Soviet Union, size income inequality within each of the new countries will decrease precisely because regional differences will be less. The inclusion of a variable that captures regional heterogeneity is then 'Clearly, this is not a perfect measure either. Heterogeneity will increase the smaller the size of the units. There also the usual problems associated with the use of extreme values only. However, as can be observed in the Annex, the variable seems to reflect relatively well the heterogeneity of the countries. 4 legitimate. However, if one believes that regional inequality is also influenced by the variables which we hold to determine personal income distribution, then the model may be misspecified. Regional inequality may, in effect, be the dependent variable, explained by the same factors as personal income inequality. If the first hypothesis is true (regional differences are 'given"), then thc correlation between regional inequality and other explanatory variables must be low, and significantly lower than the correlaticn between the other explanatory variables and size income inequality. In the empirical section, I shall therefore always present two versions of each equation: with and without the rngionJ heterogeneity variable. What is new in our "augmented" Kuznets hypothesis is the role of social choice. Our nypothesis says that -- once the "given" elements are accounted for -- there is still siz^able discretion regarding income inequality. Income 4istribution is viewed also as the product of social choices mediated through elections, lobbying of various social groups, societal preferences or historical devclopments. Thus, some countries may have a greater proportion of state-sector workers because socialist or Communist parties were historically stronger; or the population may have a high preference for eradicating povert, and redistributing income through transfers; or the middle classes which decisively determine the size of transfers in developed democracies may have had experience of downward mobility and may regard transfers as an insurance proposition (lest they become poor) as argued by Lindert (1989 and 1991). In any case, variables such as the size of the state sector and the size of transfers will be determined through the interaction of social forces, or put rmore broadly, by the political economy of the country. Consider now the influence of the two "social choice" elements in more Jetail. The large size of the state sector will tend to reduce inequality because of a more compressed wage distribution existing in the state compared to the private sector. More bureauciatic structures, in which earnings are largely determined by seniority and academic credentials, are believed to reward those at .he top relatively less and to pay relatively more to those at the bottom. This is confirmed by empirical studies. Bishop, Formby, and Thistie (1991, p.430) find that wage distribution in the U.S. government sector is consistently more egalitaria. than in manufacturing, / 5 services or agriculture (all of which are entirely private). Meron (1991) obtains the same result for France. Blank (1993, pp. 29-30) writes: "Pu',lic sector workers [in the U.S. and the UK] face more compressed wage distribution than do private sector workers. For almost every occupation in cvery year in both countries, M'th the 10th percentile and the 90th percentile of wages in the public sector are closer to the mean public sector wages than are 10th percendle and 90th )ercentile of wages iP the private sector." Further confirmation of the levelling tendencies present in state-owned enterprises is provided by socialist countries, where the majority of workers (outside agriculture) were or are employed ir. the state sector. Wage distribution in socialism, adjusted for the heterogeneity of the country, tends to be more equal than in capitalism. Thus Phelps-Brown (1988, p.303) writes that lower inequality in Soviet-type economies "arises mainly from a slower rise of income above the' median, that is, t-oadly: the more skilled manual occupations and still more .he higher clerical, the professional and administrative, are paid less than in the West relatively to the bulk of manuial workers."7 There is yet another reason wh.y a high level ot state involvement in the organization of an economy may lead to lower inequality. The point was made by Hirschman (1973, p.558) "[i]f decision-making is perceived to be largely decentralized, individual advances are attriouted to chance, or possihlv merit (or dement). When decision making is known to be centralized, such advances will be attributed to favoritism.... [Centralized systems] will strain to be more egalitarian not just because they want to, but also because they have to: centralization of decision making largely deprives them of tolerance for inequality that is available to more decent-alized systems". I am not aware of previous attempts to link explicitly, at the economy-wide level, the share 'See also Phelps-Brown (1977, p.286) and Lydall (1968). Atkinson and Micklewright (1992, pp.81ff.) show that Czechoslovakia, Hungary and Poland have consistently lower earnings inequality than the UK. The USSR and the UK have about the same level of inequality of earnings; the former is, however, regionally much more heterogeneous. Comparisons are, of course, strewn with many problems. State sector wages in socialism are almost always on net basis, wages in capitalism are gross. This imparts an upward bias to income inequality in market economies. The opposite bias, however, has to do with the absence of unemployment in socialist countries. This means that even those with low productivity, often unemployed in market economies, will be wa".e earners in socialist economies. 6 of the state-sector employment to size .ncome ineqluality. Some indirect attempts were made -- for example, through the introduction of the dummy variable for socialist countries. In some studies (e.g. Kaelb!e and Thortlas, 1991, or Ahluwalia, 1976) the socialist dummy variable was found to be significant (lowering inequality) while in others its effeot was negligible (Dye and Ziegler, 1988). Here, however, I propose to use a continuous variable that spans almost the entire theoretical spectrum from 100 nercent of state employment (USSR and Czechoslovakia before the change of the regime) and almost 0 ?prcent (e.g., 3 percent for Madagascar and Senegal). The extent of government transters will also tend to reduce inequality. The relationship, however, is not unambiguous, because the reduction in ineqjality achieved by a given amount of government transfers will vary. The reduction of inequali+, will depend on the cxtent to which transfers are focused on the poor. If most transfers are captured by those who pay taxes out of which the trans-ers are financed, the reduction in inequality may be small (the theory of the middle class capture of benefits argued by Le Grand, 1982 and Sawyer, 1982). However, on balance, the larger the transfers are, the greater will be the reduction in inequality, even if the relationship may be concave, that is, additional increases in transfers may lower inequality by less and less. 7 4. Testing the New Hypothesis The D- i The sample consists of 22 OECD countries, 8 socialist European countries including the former Soviet Union 16 African, 17 Asian, and 17 Latin American countries. For these 80 countries I have been able to col!ect the necessaiy information, compatible in both the definition of the variables and the time-period (mostly early to mid-1980s). These 80 countries acccJnt for 98.8 percent of world GDP and 90 percent of world population.8 The list of the countries, the data, and their sources are given in Annex Tables 1-4. in undertakings of this scope, the data represent a particular problem. It is therefore important to discuss them in some detail. Income distribution data are gei rally thought to be among the least reliable types of macroeconomic data. The problems that hinder comparability are numerous. The most frequently mentioned are the fcllowing: How representative are household rurveys on the basis of which income inequality is estimated? What is the type of incorrei (original, gross, or disposable)? Who are the recipients (households, families or individuals)? How are they ranked (by total household income or by household per capita income or by equivalent household income)? Therefore, in Annex Table 4, I have indicated exactly the type of income and recipient from which the Gini coefficients are calculated. A general requirement, satisfied for al! the countries, was twofold: the data should be derived from household surveys and they should be nationally-representative. For the OECD countries, I have relied heavily on the Luxembourg Income Study (LIS) where a special effort was made to gencrate consistent data across the countries. For most of the OECD countries, the Gini coefficients are calculated for disposable (after both transfers and personal taxcs) per capita income. The recipients are individuals. This means that each individual in a household is assigned the same, household per capita, income. The same principle was applied to Eastern Europe and the former Soviet Union, where most of direct taxation is in the form of 'GDP figures exclude the Soviet Union and East Germrany. Taiwan and Hong Kong are not included in either population or GDP figures. 8 payroll taxes. Most of East European data were directly calculated from the published household surveys. For the Latin American and Caribbean (LAC) countries, .he majority of the data come from a single source (Psacharopoulos et al., 1992) which itself is based on household surveys of very similar design as those used for OECD countries and Eastern Europe (distribution of individuals by their per capita income). However, income is almost always gross (i.e. inclusive of transfers, but not of personal taxes) rather than disposable income: since personal taxation is minimal in LAC countries, the two measures do not differ by much. Full comparability was more difficult to ensure for Africa and Asia. The problem here is less the income concept -- gross and disposable income are practically the same -- but rather the reliability of the surveys. I have used Iublished results which I have tried to render as consistent as possible, often by using the data from the same source (e.g., a single comparative paper). The problems, however, remain: it is mostly households, rather than individuals, that are treated as recipient units. This imparts an upward bias to the aata. Finally, regardirg the time-period: for all but 10 countries, the Gini coefficients are from the 1980s (including 1979). The reader can check how close the definitions and the time-periods are in Annex Table 4. I believe that the data represent the most consistent set of the Gini coefficients existing at present. Among explanatory variables, social transfers as a percentage of GDP and GDP per capita in equivalent purchasing power are relatively easily available. OECD and ILO data are the source for cash and in-kind social expenditures for most of the countries; these data were complemented by various World Bank, IMF and individual countries' publications (see Annex Table 2). For practically all tne countries, the data refer to the year 1985 or the 1980s average. The purchasing power equivalent GDr per capita in 1988 or 1985 is obtained for practically all the countries from Summers and Heston (1991). The exceptions are several East European countries that were not included in the Summers-Heston sample. Estimates for these countries are made by the World Bank. Since both income concepts (disposable and gross income) used for the calculation of the Gini coefficient include transfers, size of transfers will, it is argued, directly influence both types of GINI. But, in addition, there may be also indirect effects of social transfers. As documented (see Danziger, Haveman and Plotnick 1981 for a review of the U.S. experience, or Atkinson 1987 and 9 Atkinson et al., 1984 for the UK experience) the existence of transfers leads to changes in behavior of firms and individuals and thus affects their pre-fisc income. For example, existence of unemployment insurance may reduce willingness to work and reduce person's labor income. If that person is poor and his overall income, equal to income from unemployment allowance, is less than would be his income from labor (in absence of unernplpoyment insurance), a perverse situation may appear where increased transfers -- existence of unemployment insurance -- lead to greater inequality. I cannot account for this effect. I must assume that the indirect effect is sufficiently small to be swamped by the direct effect of transfers on income distributuon. The size of the state sector is more difficult to obtain. Again, for the OECD countries, the OECD publications are the best source (even if such publications are not as exhaustive and up-to- date as one would expect). East European countries generally provide, in their statistical yearbooks, very detailed data on the size of the state sector (and the cooperative secor). For the LAC countries, Psacharopoulos et al. (1992) has also been used extensively because household surveys provide information on the employer (state, private, own-account) of the interviewed individuals. For Africa and Asia, the main sources were countries' statistical yearbooks. In almost all cases, the denominator (state sector as percentage of what) wis the labor force or the economically active population. Both include the officially unemployed and agricultural underemployment; both exclude students, housewives, etc. that is, people of working age who are not economically active outside their household.' Aimost all of the data refer to the 1980s (see Annex Table 1). The heterogeneity variable is not only the most problematic in analytical sense; it is so in an empirical sense as well. I have tried to use the ratio of household incomes (per capita or total household) between the richest and the poorest region as the preferred indicator. But even if such data existed for all the countries, changing administrative divisions alone would produce changes in the results. Clearly, the greater is the number of administrative units in a country, the larger is the ratio. I have therefore indicated, in Annex Table 3, the number of administrative units The distinction is, of course, somewhat artificial in the case of countries with agricultural underemployment. /' 10 which are being compared (e.g. 10 regions or 16 provinces or 24 departments). In addition, ratio in incomes was not always available. I have then had to resort to proxies like consumer expenditures, wage bills per region, or even, in some cases, per capita consumption of electricity (for five countries) or per capita ownership of consumer durables such as cars or TV sets (for six countries). The explanatory variables are therefore the following: INCOME = the country's purchasing power 1988 GDP per capita (in thousands of 1988 international dollars); RATIO = the ratio of average incomes between the richest and the poorest region within a country; STATE - the percentage of all employed who work in the state sector (inclusive of government administration); and TRANS = the percentage share of cash and in-kind social transfers (pensions, maternity and family allowances, temporary sick pay, unemployment compensations, education, and health) in the country's GDP. The dependent variable is the Gini coefficient of disposable income (GINI) expressed for convenience in percentages: Gini coefficient of 30 (instead of 0.3). Two further points need to be clarified. An apparent inconsistency may be detected between the inclusion of in-kind transfers like education and health in the TRANS variable, and concern with disposable income inequality (which excludes public in-kind transfers) in the GINI variable.10 The rationale fcr this is that public expenditures on health and education are conducive to more equal distribution of human captal which, in turn, reduces the inequality of disposable income: for example, more widely spread public education is likely to reduce wage differences. Second, the analysis is conducted in per capita terms rather than in terms of equivalent consumption units. There are several reasons for this. There are practical ones, because most of the income distribition data for non-OECD countries are expressed in per capita terms; also, even when income distribution is done in terms of equivalency units, the weights used in different studies are different. There are also more substantive reasons for using the per capita measure. If we require that GINI be reported in equivalent units should we not require the same for GDP? "'Disposable income includes money income plus in-kind consumption. /' 11 The most compelling reason is that the very idea of equivalency units is country-dependent (or rather price-structure) dependent. If rents, for example, are subsidized, then economies of scale are much less important than if they are not; if education is private, the cost of children is much higher (their weight may be higher than the weight of an adult) than if education is public and free. In consequence, the use of per capita terms has both practical and substantive adventages. Empirical Analysis The regression with the expected signs of the coefficients is given below. The income variable is quadratic, since we test for the existence of an inverted U-shaped relationship. GINI = fct [INCOME, RATIO (+) ,STATE (-) ,TRANS (-)]. The expected negative sign of TRANS deserves a further comment. As has been ar increased social transfers will tend to reduce the inequality of disposable or gross income. b . ln some recent studies (e.g., Alesina and Rodrik, 1992; Persson and Tabellini, 1992) whic;. concerned with determinants of social transfers, higher income inequality is shown to lead, it conditions of wide franchise, to high redistribution. It would hence appear that inequality and transfers are positively related. The example underscores the ambiguity with which the term 'income inequality" is used. The positive relationship between income inequality and transfers makes sense only if one has in mind inequality of market income (before government redistribution)."1 It is then logical to assume that if market incomes are distributed unequally, people (i.e. the median voter) will vote for large redistribution because they will thereby gain. But both Alesina and Rodrik (1992) and Persson and Tabellini (1992) use measures of income inequality after government cash transfers. Consequently, the cross-country relationship between market income inequality and TRANS may be positive (because taxes are higher in more unequal countries), while the cross-country relationship between TRANS and disposable or gross income "Market or original income is the income prior to any government redistribution (ideally, it should be even prior to payroll taxes deducted at source). Gross income is equal to market income plus all cash government transfers. Disposable income is equal to gross income minus all direct taxes. 12 inequality may be negative (because transfers paid out of taxes lower inequality). The two income inequalities -- pre- and post-government -- are in effect two entirely different variables. Table 1 gives summary statistics for the five regions. The most important conclusions are the following. (1) In terms of income inequality, the five regions have distinctly different averages: inequality is highest in Africa (Gini of 52), closely followed by Latin America (49), then Asia (41), OECD countries (31), while the European socialist economies are the most equal (25). (2) Eastern Europe and the former Soviet Union have a much larger share of state sector employment than does any other region (90 percent); the African and Asian samples have the lowest share (11 to 12 percent of the labor force). (3) The size of social transfers is much greater in OECD and socialist countries than elsewhere. (4) Regional heterogeneity within countries is largest in Latin America, followed by Africa; OECD countries are the most homogeneous. Table 1. Summary statistics for the five regions Region GINI STATE TRANS RATIO INCOME Number OECD 31.2 21.2 22.6 1.8 12501 22 E. Europe 24.8 90.0 17.2 2.5 6234 8 Africa 52.3 11.3 5.7 4.8 1778 16 Asia 41.0 12.6 6.8 3.3 4851 17 L.America 49.2 19.3 7.6 7.0 4156 17 Note: All the statistics are unweighted averages. Definition of the variables: Region: For the list of countries see Annex. Algeria, China, and Cuba, although socialist, are included in their respective regions. GINI : Gini coefficient of disposable income (for OECD and socialist economies); Gini coefficient of gross income for Africa, Asia and Latin America. Gini coefficients are expressed in percent. STATE: Share of state sector workers (general government and state-owned enterprises) in total labor force. TRANS: Share of cash and in-kind social transfers in GDP, in percent. RATIO : Ratio of per capita income between the richest and the poorest administrative unit (province, republic, state) within a country. INCOME : Purchasing power GDP in international dollars for 1988. Number: Number of countries included. 13 The relationship between RATIO and other explanatory variables is of particular importance because of the two possible interpretations of regional heterogeneity mentioned above. In order to include RATIO in our regressions we need to satisfy two conditions. First, the correlation between RLATIO and GINI, while existent, should not be close to unity (as it would be if RATIO and GINI were practically the same variable); and second, the correlation coefficients between the other explanatory variables and RATIO should be small (ideally close to zero) and in any case smaller than the correlation between these explanatory variables and GINI. Table 2 shows the results. The correlation between RATIO and GINI is -t0.54, which is the weakest of any explanatory variable and GINI. This argues that RATIO is not a proxy for GINI. The correlation between other explanatory variables and GINI is always two to three times stronger than the correlation between the same explanatory variable and RATIO, thus implying that RATIO is not determined by the same set of factors as GINI. RATIO can therefore be included in our regressions. Table 2. Testing RATIO: Zero-order correlation coefricients [_____________ STATE TRANS INCOME GINI RATIO -0.20 -0.39 -0.39 +0.54 [ GIN! -0.63 -0.73 -0.60 ____ Figures la-Id display the relationship between GINI and the four explanatory variables. We test first the "canonical" equation given above. This is equation (1.0) displayed also in Table 3. The observations in all the regressions are arranged in ascending order according to INCOME. GINI = fct [STATE, TRANS, RAT70, LN(INCOME), LN(INCOME)2]. All the coefficients have the predicted sign and are statistically significant at either 1 percent 14 FI;ure 2a. Relationship between GINI and STATE Go0_- :Utz HON A . PEUAN 55 - uDSaL s 50 - SL& R 45- _ 35 - Be iy- nL 250- FIN +ze 20 IS 10 20 30 40 50 60 70 00 go too1 STATE L, - ZAK BSA 2 - MAL HK 3 - LQ. M x 4 - BK MY Note: Steeper regression excludes socialist countries. 15*~~~~~~~1 Figure 2b. Relationship between GINM and TRANS 65 ,-.. a a PERU W*WAZN Ix PAN 55- 50 - 10 IF 13~~~~1 45 - ~40 - P 4 4o GILA 30- e1C DE 25 -EL 20 CDR Is.& 0 5 10 15 20 25 30 35 16 Figure 2c. Relationship between GINI and RATIO 45-- Cal LvA so - - 55- Wp126 AL aso- cm 45- 30 - 25- 20 - 0 2 4 6 U 10 12 14 14 16 20 22 24 26 RATIO 17 Figure 2d. Relationship between GNI and INCOME o *- 1E,BN GAL EA SWA PERpAN S _40- GIA ' ' * 355- BAst INR D,E Cf _SZ 1 20- , 15 , , 0 E 4 8 8 10 12 14 16 18 20 SNCOME Notes: Socialisit countries shown in the circle. Regression is: GINI - constant + B0 l(INCOME) r 1 InACOMD 18 (STATE, TRANS, RATIO) or 5 percent level (inINCOME and squared InINCOME).'2 The intercept is not statistically significantly different from zero. This means that, for a sufficiently low per capita income (at the limit for INCOME=0) and in the absence of state sector employment and transfers, the Gini coefficient would be close to zero: i.e., no inequality would exist. The coefficient of determination is 0.76. The interpretation of the results is as follows. Each ten percentage point increase in the share of state sector workers reduces inequality, on average, by 2.09 Gini points; each increase in social transfers by 10 GDP percentage points lowers inequality by 3.8 Gini points; each increase in country's heterogeneity by I (say, from 3 to 4) increases inequality by 0.65 Gini points. Finally, the relationship between income level and inequality is quadratic: at first, inequality rises with income and then declines. The turning point is reached for $2,100 per capita (at 1988 international prices) which is broadly the level of income of the Philippines, Swaziland, or Sri Lanka.'3 There are two potential problems with equation (1.0). The first is that of heteroskedasticity. It was observed in the literature (see Lindert and Williamson 1985, p. 344; Lecaillon et al., 1984, p.40) that the dispersion of the Gini is greater at low than at high income levels. One can therefore expect some heteroskedasticity because standard errors would systematically decline with increase in income level. Indeed, this is exactly the case, as shown in Figure 2, where residuals from equation (1.0) are plotted against income levels. Regression (1) is the same as (1.0) except that I correct for heteroskedasticity by running OLS with Whites' heteroskedastic-consistent standard errors. This does not affect STATE, TRANS, or RATIO but does affect the two income "2I have experimented with a number of other formulations, some of them suggested recently by Anand and Kanbur (1993). The log-squared gives the best results. This was the original formulation used by Ahluwalia (1976). 13This is somewhat higher than the turning point shown in Figure ld (about $1,800) where GINI is a function of INCOME alone. Ahluwalia (1976) finds the turning point at $468 per capita at 1970 prices and current exchange rates. On the basis of a somewhat smaller sample, Kaelble and Thomas (1991) find the turning points to range, depending on the measure of inequality used, between $322 and $489. Converting these values to 1988 prices and then applying the ratio between the purchasing power parity exchange rate and the current exchange rate from Summers and Heston (1991), we can express the turning points in 1988 purchasing power GDP per capita (as in our sample). Ahluwalia's value is then equivalent to $3 070, and Kaelble and Thomas's range turns out to be $2,175 and $3,176. ,'' 19 terms that become statistically significant only at a 10 percent level (instead of 2-3 percent level in regression 1.0). Since the same problem exists in all equations, all regressions will henceforth be run with the correction for heteroskedasticity. The second problem is the role of RATIO. As indicated, we need to be sure that the model is correct even if RATIO is left out. Thus, regression (1A) is the same as (1) except for RATIO which is now deleted.'4 Omission of RATIO raises the coefficients and the significance of all the remaining variables. This produces an important effect on both inccme terms which now again become statistically significant at 2-3 percent level. The coefficients of STATE and TRANS remain stable. They rise in absolute amounts but by relatively little (e.g., STATE rises from -0.21 to -0.22). The intercept remains not significantly different from zero. R2 decreases by very little, from 0.76 to 0.71. We can therefore conclude that the omission of RATIO does not affect the results except that it brings out the role of income more strongly. Flgure 2. Residuals from equation 1.0 as a function of INCOME 10. + + -0 , ._ , 4. +4 + $4 + 44 .. . . . . . .. ..4 .4 . . . . 4 . .. . .. . 0 5000 10000 lS000 206Q00 PPP Are our results, and in particular the role of STATE, perhaps driven by the presence of "nTis notational rule will be followed throughout: equation number followed by A denotes the same equation save for the elimination of RATIO. 20 socialist countries and their high share of state-sector employment? Regression (2) is the same as regression (1) except that all socialist countries (7 from Eastern Europe, the Soviet Union, Algeria, China, and Cuba) are dropped. The values of the coefficients change but slightly: the coefficient of STATE becomes, in absolute terms, greater, rising from -0.21 to -0.32 (see also Figure la where the regression lir.e becomes steeper when socialist countries are omitted) and the coefficient of TRANS becomes smaller. Both income coefficients increase and their statistical significance rises. R2 decreases from 0.76 to 0.7. Overall, the inclusion or exclusion of socialist countries makes little difference. The steeper relationship between STATE and GINI when socialist countries are omitted requires an explanation. It implies that decreases in inequality recorded by socialist countries are small compared with the huge size of the state sector in their economies. Ind od, even from the summary Table 1, it can be seen that while the difference in GINI between East European countries and (say) OECD is only some 6 Gini points (or differently, inequality in OECD is about a quarter greater than in Eastern Europe) employment in the state sector is more than four times greater in Eastern Europe. Therefore, when socialist countries are dropped from the sample a given increase in state sector share produces larger decreases in GINI. Regression (2A) is the same as (2) except for RATIO which is omitted. No major differences between the two regressions exist except (as before) that income terms are larger and statistically more significant. 21 Table 3. The Regressions: 80 countries; except equations (2) and (2A), 69 non-socialist countries only Regr Constant STATE TRANS RATIO INCOME INCOME2 DUMMY EDUC R2 (F) SE(DW) 1.0 -69.08 -0.209** -0.381** 0.646** 31.21* -2.036* 0.76 5.947 (0.22) (0.000) (0.003) (0.000) (0.028) (0.020) (46.5) (1.95) 1 -69.08 -0.209** -0.381** 0.646** 31.21 -2.036 0.76 5.947 (0.37) (0.000) (0.000) (0.000) (0.092) (0-G64) (46.5) (1.95) 1A -97.08 -0.223** -0.416** 39.80* -2.608* 0.71 6.449 (0.21) (0.000) (0.000) (0.C35) 0.002) 146.4) (1.99) 2 -84.72 -0.320** -0 297** 0.652** 3--.35 -2.293* 0.70 6.242 (0.29) (0.001) (0.005) (0.001) (0.065) (0.043) (28.8) (1.94) 2A -113.2 -0.288** -0.343** 44. 13* -2.888* 0.64 6.779 (0.16) (0.004) (0.002) (0.124) (0.013) (27.9) (2.02) 3 -7 l.48 -0.182** -0.386** 0.b42** 31.87 -2.084 -2.079 0.76 5.980 (0.36) (0.005) (0.000) (0.001) (0.086) (0.059) (0.650) (38.4) (1.94) 3A -100.3 -0.185** -0.423** 40.66* -2.671 * -2.949 0.71 6.479 (0.20) (0.000) (0.000) (0.032) (0.018) (0.413) (36.9) (Q.98) 4 -48.99 -0.190** -0.292** 0.672** 24.60 -1.449 -1.247** 0.78 5.680 (0.528) (0.000) (0.002) (0.000) (0.186) (0.191) (0.002) (43.9) (2.03) 4A *79.71 -0.206** -0.336** 34.06 -2.092 -1.144** 0.73 6.259 (0.309) (0.000) (0.001) (0.072) (0.064) (0.013) (40.6) (2.06) .1 5 -91 1 -0.230** -0.512** 0.498** 37.22* -2.376* -7.128** 0.81 5.306 (0.165) (0.000) (0.000) (0.001) (0.020) (0.014) (0.000) (52.1) (1.82) 5A -115.1 -0.244** -0.558** 44.47** -2.849** -8.199** 0.78 5.625 (0.076) (0.000) (0.000) (0.005) (0.003) (0.000) (53.8) (1.93) 22 Notes to Table 3: Values in parenthesis are the complements of the level of confidence with which the null hypothesis is rejected. Two (one) astensks indicate that coefficient is significantly different from zero at less than 1 (5) percent level. Variable INCOME is In (purchasing power per capita GDP). Variable INCOME' is INCOME squared. In regressions (3) and (3A), DUMMY variable takes va!t 1 for socialist countries, zero for others; in regressions (5) and (5A), DUMMY variable takes value I for Asian coumtIies, zero for others. Another issue is whether our STATE variable really adds something to the common practice of using a dummy variable for socialist countries in income distribution studies. We argued above that STATE is more general because it covers the whole spectrnm of values from 0 to 103, and thus differentiates also between various capitalist (or even socialist) countries. In regressions (3) and (3A) I introduce both STATE and a socialist dummy variable (otherwise the regressions are the same as 1 and 1A). The equation is therefore GINI = fct [STATE, SOCIALIST DUMMY, TRANS, RAT70, LN(INCOME), LN(INCOME)J. The regression coefficients are practically unchanged. Only the coefficient of STATE decreases somewhat (from -0.21 to -0.18) but remains highly significant. We can safely reject the hypothesis that the dummy variable is statistically significant in the presence of STATE. Is Asia Different? From Figure 3a, which displays residuals from regression (1), it emerges that in the case of Asian countries the actual level of inequality is often smaller than the predicted. Out of five countries whose actual inequality is more than 10 Gini points less (about one-and-half standard deviations less) than tie predicted inequality, four are Asian (Bangladesh, Pakistan, South Korea and Taiwan)." Also, out of 17 Asian countries (the dots in the Figures), in only four is the actual inequality higher than the predicted inequality. Differently, in African and Latin American economies inequality seems to deviate upward from the predicted values. "The only other one is Ghana. 23 * 01-.0~~~~~~~~~~~~~~~~~N 1- J~~~~~ I......,C-0 g2. . . C.... ......... . . . . . . . .... .................. ........... ............... .. . ......... ......... .. ;. ........ . . . . . . . . . IN nvd~~~~~~~~~~ (g ao mwj :q[ aj21 (1) wpba Joi w 92 '-1 *0 Several possible explanations for the contrast between Asia and other continents can be adduced. For example, more equal distribution of physical and human capital in Asian countries may result in lower market (pre-government involvement) inequality. Then, even if transfers are small, inequality in disposable income (i.e., after transfers and taxes) will be less than in the countries in which the underlying market distribution of income is skewed. Take, for example, Taiwan and Uruguay, both probably the most highly educated and among the most developed countries in their respective regions. The per capita GDPs of these countries are very close ($6,500 for Taiwan and $5,800 for Uruguay). Uruguay's share of state sector workers is twice as high as Taiwan's (21 vs. 10 percent), and social transfers are greater (10.5 percent of GDP vs. 8.1 percent). Yet Taiwan's Gini coefficient is 32 and Uruguay's is 42. But the average number of years of education completed by the population over 25 years of age, is 9.2 years for Taiwan and 7.8 years for Uruguay. The high premium placed on education in Taiwan is also reflected in the structure of social transfers: while tota; social transfers, in terms of GDP, are smaller in Taiwan, public education expenditures are three times as high: 4.6 percent of GDP in Taiwan and 1.5 percent in Uruguay. Another indicator of the high dispersal of assets in Taiwan is the proportion of stock-owning population, which at 27 percent is twice as large as in most West European countries and about the same as in the United States. One possibie explanation of the lower (than predicted) inequality in Asia may lie then in a more equal distribution of physical and human capital. The former is extremely difficult to approximate; the latter can be approximated by the spread and depth of education. I introduce the average number of school years completed by the population 25 years of age or older (EDUC).1 The equation (4) is therefore GINI = fct [STATE, TRANS, RATIO, LN(INCOME), LN(INCOME)2, EDUCI. However, because of the strong collinearity between education and income, no new insight is obtained. These two variables can be used practically as substitutes. The introduction of education renders both INCOME terms statistically insignificant (see regression 4). Moreover, EDUC does not reduce the downward deviation of GINI observed in Asian countries (not shown n6The data come from the United Nations Development Program (1992). 25 here). The omission of RATIO (regression 4A), as in earlier regressions, increases all the coefficients and raises the statistical significance of both INCOME terms; however they still remain statistically insignificant at a 5 percent level. Education, therefore, does not provide an independent explanation (i.e., an explanation that is different from what is implied by income) for the lower inequality in Asia.'7 We are left with the altemative of introducing a dummy variable for Asian countries (equations 5 and 5A in Table 3). The equation becomes GINI - fct [ST,1TE, TRANS, RATIO, LN(INCOME), LN(INCOME9, ASIA DUMMY]. This improves the fit and eliminates the systematic negative residuals for the Asian countries (Figure 3b). All the coefficients, including those of both INCOME terms, are statistically significant at less than 2 percent level. This is the first time that in the presence of RATIO both IN iCOME terms are statistically significant. The dummy variable has the expected negative sign and is highly significant: Asian countries have, all other elements being the same, an income inequality that is some 7.1 Gini points less than that of non-Asian countries." This, of course, is not an entirely satisfactory conclusion because we are unable to explain what real factors lie behind the observed lower inequality in Asia. What Explains the Differences in Inequality? On the basis of these results we can find the causes for the difference in the levels of inequality between the five groups of countries. OECD countries are used as a yardstick and the difference in GINI between them and the other groups is explained by the differences in social choice variables (state sector employment and transfers), "given," variables (income levels and regional heterogeneity), and an 'Asian element' variable. I use regression (5) for the c&culations. "7Different formulations using INCOME and EDUC were tried; none dispenses with the need for a dummy variable. "As usual, the exclusion of RATIO in equation SA does not affect our results. 26 The results are displayed in Table 4. Table 4. Factors explaining the difference in inequality compared to OECD countries (in Gini points) Due to: Socialist Africa Asia LAC State sector -15.8 +2.3 +2.0 +0.4 Size of transfers +2.8 +8.7 +8.1 +7.7 The Asia dummy -7.1 Social choie 1 - 13.0 +1 L30: j Regional inequality r0.3 +1.5 +0.8 +2.6 Income level +4.1 +5.8 +5.1 +5.5 "Giv-a hfactors +4.4 _ 7 _ 3 Unexplained +2.2 +2.8 +0.9 +1.8 A7 ua i-6.4 +2. +.S Note: Calculated from regression (5) in Table 3. Negative sign indicates that a given element reduces inequality in the region in comparison with inequality in OECD countries. In the case of Latin America, Asia and Africa, the main causes of greater inequality, in comparison with OECD countries, are lower transfers (which explain between 7.7 and 8.7 additional Gini points) and lower income (which explains between 5.1 and 5.8 additional Gini points). These two elements alone would make inequality in Africa, Asia, and Latin America some 13 to 14 Gini points greater than in OECD. It is interesting to observe that despite other differences Africa and Latin America display very similar patterns in the ex.-ination of inequality. Asia, however, is different because the Asia dummy variable lowers inequality from the levels predicted by the four general variables by about 7 Gini points. We also conclude that the existing lower state sector employment and greater regional heterogeneity do not alone produce much greater inequality in the three continents compared to the OECD countries. Because of lower state sector employment, the Gini coefficient in Africa and Asia would be greater by about 2 points, and by only 0.4 Gini points in Latin America. Greater regional heterogeneity similar!y adds only between 2.6 and less than 1 Gini points (the latter in Asia) to inequality. These are all very small 27 differences. In the case of Eastern Europe, by far the most important factor explaining lower inequality than in the OECD countries is the greater share of state sector workers: this lowers the Gini coefficient by 15.8 points on average. All other elements point to a greater inequality in Eastern Europe than in OECD but their impact is not sufficient to offset the impact of the large state sector. The debate about the lower income inequality in socialist economies (Ahluwalia 1976, Morrison 1984) can now be placed within a larger context of factors which explain income inequality in general. Socialist economies display lower inequality owing to the key feature of their system: the high share of state sector employment. This tendency is partly offset by capitalist countries' higher social transfers and higher income levels. Regional heterogeneity plays practically no role. An important distinction to be made is between the effect of social choice and of 'given' variables. If income level and regional heterogeneity were the same in Africa and Latin America as in OECD, inequality would still be greater on two these continents by 8.1 (Latin America) and 11 (Africa) Gini points. In consequence, social choice elements -- principally transfers -- seem the chief "explanators" of greater inequality in Africa and Latin America. The Asian situation is different because of the ambiguity of the "Asian variable": if it is a social choice variable, as it is logical to assume, then the difference between the importance of social choice elements in the OECD countries and in Asia is very small. However, while in the OECD countries social choice operates through high transfers and state-sector employment, in Asia, social choice takes the form of relatively equal asset endowments (presumably captured by the dummy variable). If this interpretation is correct, then Asian countries can afford to have low transfers since other factors (e.g., even distribution of assets) produce relatively equal distribution of original income (pre- government redistribution). Overall, the greater income inequality in Asia -- compared with that in OECD countries -- is explained primarily by the difference in income level. In conclusion, how do we explain the higher inequality in less developed countries and the lower inequality in Eastern Europe, compared to OECD? For Africa and Latin America, inequality is higher because of lower social transfers and lower income; for Asia, inequality is 28 higher only because of lower income; and for Eastern Europe, inequality is lower because of the high share of the state sector. How Important Are Social Factors? Our next question is: What is the importance of social factors compared with "given" factors? This is an important question because it is only after we empirically know the relative importance of social factors that we can make a judgment about the extent to which the standard Kuznets hypothesis needs to be modified. If social choice variables reduce income inequality by only a few Gini points, then the general validity of the standard Kuznets hypothesis cannot be seriously questioned. Societies can at the margin tamper with income distribution, but it is overwhelmingly determined by the factors that they cannot influence in the short-run, and in particular by their level of income. Differently, if social choice variables lower income inequality significantly, then the standard Kuznets' hypothesis needs to be substantially altered. This would mean that societies can affect income distribution: the economic determinism implicit in the standard formulation of the Kuznets' hypothesis is then seriously weakened. The solid line in Figure 4 shows the calculated Gini coefficients that are solely the result of "given" factors: the line shows income inequality that would obtain if only income and regional heterogeneity determined inequality.'9 An upward and short bulge in inequality is followed by a prolonged and slow decrease in inequality as income levels rise. The Figure also shows that, if "givens" alone mattered, the differences in inequality between rich and poor countries would be relatively small. While the standard deviation of the actual GINI in our sample is 11.7, the standard deviation of the thus calculated GINI is only 4.1 (see Table 5). '9The calculation is made by using the coefficients from regression 5 for income and regional heterogeneity, and setting transfers and state sector employment=0. / 29 Figure 4. "Given" GINI and the actual values of GINM GINI 65 55 Poit rent 35 d a h t 25 * . 1 5~ ~~~~ o 2 4 6 8 1 0 1 2 1 4 1 6 1 8 20 22 INCOME ($ 000) Note: Points represent the actual GINIs. The distance between the solid line (the "given" Gini) in Figure 4 and the actual Gini points is due, save for the statistical discrepancy, to the role of social choice variables. The distance widens around $6,000 per capita. For all countries with higher incomes (except for Hong Kong), the divergence, and hence the role of social factors, is substantial. One can therefore propose two tuming points of inequality: the first would occur at the level of approximately $2, 100 where, as noted before, the standard Kuznets' curve linking income and GINI begins to turn downward. The second occurs at around $6,000 when social choice variables become significantly more important than before and reinforce the downward trend in inequality. / 30 Table 5. The role of social choice variables Level of (1) (2) Effect of Due to: Due to: income ($ "Given" Actual social STATE TRANS PPP) GINI GINI choice:(2)-(1) Less 1500 56.0 50.7 -5.3 -1.9 -2.3 1500-3000 56.8 46.0 -9.2 -3.7 -2.9 3000-4500 57.1 46.2 -10.9 -5.9 -4.5 4500-6000 55.4 41.2 -14.2 -8.6 -5.7 6000-10,000 52.6 29.8 -22.8 -10.2 -8.6 Over 10,000 48.8 31.4 -17.4 -5.8 -11.0 Total 53.9 40.7 -13.2 -5.5 -6.2 Standard 4.1 11.7 deviation Sotes: Given' GMN: Calculated from re-ression (5) by setting STATE and TRANS-O. Effect of STATE and TRANS: Calculated from regression (5) by multiplying the corresponding coefficients with the actual values of STATE and TRANS. All of the difference in column (3) is not explained by STATE and TRANS. Some of it is explained by the Asia dummy and some is unexplained because of the discrepancy between the values predicted by the regression and the actual GINs. All values a-e unweighted averages. The difference between the unweighted "given" Gini and the actual Gini in the whole sample amounts to 13.2 Gini points (Table 5). This is, therefore, the joint effect of social transfers and state sector employment: a reduction of the Gini coefficient from almost 54 to 41. How important is this effect? How big is it in practical terms? It is equivalent to transforming Bolivia or Cote d'Ivoire (both with actual Ginis of about 54) into Sri Lanka or Uruguay (Ginis of 41). The 13.2 Gini point reduction is almost evenly shared between the effect of state sector employment and social transfers: state employment reduces inequality, on average, by 5.5, and social transfers by 6.2, points. The effect of the social choice variables is not independent of the level of income. At low levels of income, less than $1,500 at purchasing parity, the "given' and actual Gini differ by very 31 little: by about 5 Gini points with STATE and TRANS being of about the same importance in reducing inequality. Between $1,500 and $4,500, social choice variables reduce inequality by some 10 Gini points. The state sector now becomes more important than transfers. After $4,500, the importance of social choice variables further increases, reducing the "given" GINI by between 15 and 20 Gini points or, put differently, cutting the level of inequality by more than a third. The importance of STATE remains greater than that of TRANS reaching its peak for the countries with incomes between $6,000 and $10,000 where almost all socialist countries are located. Finally, for the richest countries, the reduction in inequality, equal to 17.4 Gini points, owes much more to transfers than to state sector employment. Two conclusions can be drawn. First, variables which represent social choice have an important role in determining the degree of inequality. On average, social choice variables reduce the unweighted Gini coefficient in our sample by some 13 Gini points (i.e., by a quarter). Second, the importance of social choice variables increases with level of income. Social choice variables do not matter very much at low levels of inco ie, but as income rises, society's preference for policies that reduce inequality seems to increase. Equality seems to be a superior good. The strong fermulation of the Kuznets hypothesis is therefore less valid as income increases and non- economic factors -- compared with strictly economic factors -- become more important in shaping personal income distribution. 32 5. Conclusions and Implcation of the Findings We have set out to answer two questions. First, do social choice variables -- jointly with the purely econiomic variables included in the standard formulation of the Kuznets' hypothesis - determine income inequality? The answer to this question is Yes. We have found that social choice variables (social transfers and state sector employment) uniformly, in all formulations of the regressions, show a statistically significant negative impact on inequality. The second question is, how important is the effect of social choice variables? Here we have found that, for the sample of 80 countries in the 1980s, the social choice variables reduce inequality by some 13 Gini points. Actual inequality is, on average, only about three-quarters of what it would be if social variables were not operative. But this relation is not uniform with respect to income level. At a low level of income, the role of social choice variables is almost negligible. As income rises, their importance becomes greater. This finding cannot be interpreted by arguing that, at a low level of income, social choice has no role to play because there is nothing to redistribute as everyone is poor. This is patently not true because at low levels of income inequality is relatively high.2" Thus, social choice variables could, a priori, play a significant role even at low income levels. Why they do not do so can only be conjectured now. My hypothesis is that society's preferences change in the process of development and that people, as average income rises, tend to place greater emphasis on equality. The preference for social equality is therefore income-elastic. But, whatever the cause for the increasing role of the social choice variables, the implication of our results is that the validity of the strong formulation of the Kuznets' hypothesis diminishes as society develops. The level of inequality that a society charts in its development diverges increasingly downward from the level predicted by the Kuznets' curve. The discrepancy is therefore systematic. This is so because inequality in richer societies does not decrease because of economic factors, but also because societies choose less inequality. IOAt some possibly mythical extremely low level of income everyone would be equally poor. But this is not true at the actual low levels of income which we observe in our sample. // 33 We also find that Asian countries, once all these elements are taken into account, tend to have a lower than predicted inequality. The difference amounts to some 7 Gini points. Further research may be needed to find out just what accounts for the lower inequality. One hypothesis has been that the distribution of physical and human capital may be more equal in Asian countries -- for a given level of income -- than elsewhere. If this is the case, then government redistribution via transfers and taxes need not be as extensive in Asia as in other regions with more unequal personal distribution of assets. Equal distribution of assets, if confirmed, may be that missing "social choice" variable that not only explains lower inequality in Asia (compared to what "it should be") but provides a potential clue for high growth rates recorded by some Asian countries. Recent literature on the link between economic growth and political economy (e.g., Alesina and Rodrik 1991; Perotti, 1991 and 1992; Persson and Tabellini 1992) argues that the size of transfers is determined by the political process, in short, by the gain that the median voter expects from redistribution. Thus the population in countries in which assets are highly unequally distributed and in which, consequently, inequality in original income is high, will have an interest to vote for large social transfers. To the extent that transfers reduce the incentive to accumulate wealth and to work hard, either economic growth will be slow or democracy will be impossible to achieve. The dilemma, familiar from the 19th century Europe, was eloquently summarized by the Spanish statesman Canovas del Castillo: rebutting those who complained about electoral fraud, he wrote: 'To have I o choose between the permanent falsification of universal suffrage and its abolition is not to have to ^hoose between universal suffrage and preservation of property" (quoted in Ubieto et al., 1972, p. 731). But if a country's assets are relatively widely distributed and market- generated inequality is moderate, then large, particularly cash, transfers are not needed. Fast growth becomes compatible with democracy (as the median voter does not have an interest to vote for high taxes) and relatively equal distribution of income. Our "augmented Kuzrets" hypothesis cam also be considered in a historical continuum. Pareto was the first economist who studied personal income distribution. On the basis of his empirical research, he was led to formulate the "iron rule of inequality."2' Pareto held that, whatever the 21Pareto's law of income distribution appears for the first time in print in 1896. The sample contains seven countries or cities. The next year Pareto (1897) published his famous article in 34 social system, level of deveiopment, or type of elite in power, size income distribution had the same shape: only different people may be rich in one system (say, owners of capital) than in another system (for example, party bureaucrats or lana-owners). After numerous disputes, Pareto's "iron law" was generally rejected. The most favorable conclusion that can be made is that the upper tail of income distribution (top 1 to 2 percent of recipients) tends to display features observed by Pareto and embodied in the density function bearing his name. The second general theory of income distribution was propounded by Kuznets (1955). The unmovable "iron law" of income distribution took the form of an economic "iron law," whereby size income distribution changes with development but does so in a predictable way and shaped by economic factors. The forces that determine the distribution of personal income, although knowable, are not alteiable by human design (unless, of course, a society decides not to "develop"). This is so because the level of inequality is chiefly determined by economic factors: by the level of development and the attendant scarcity and the concentration among the individuals of various grades of skills, capital and land. The hypothesis advanced here mitigates the economic determinism implicit in the standard formulation of the Kuznets' hypothesis.22 Size income distribution is determined also by social choices. Societies can choose, within limits imposed by the "objective" circumstances, whether they want to have a more or a less equal income distribution. And they tend to choose less inequality as they grow richer. which his original sample is extended by a further ten countries. See Creedy (1985, p.22). 22Kuznets himself was aware of the role of social factors. See the quotation above. 35 REFERENCES Ahluwalia, Montek S. (1976), "Inequality, Poverty and Development", Journal of Development Economics, No.3. Alesina, Alberto and Dani Rodrik (1992), "Distributio-, Political Conflict and Economic Growth: A Simple Theory and Some Empirical Evidence", in Alex Cukierman, Zvi Hercowitz and Leonardo Leiderman (eds.), Political Ecoromy, Growth and Business Cycles, Boston, Mass.:MIl' Press, pp. 23-50. Anand, Sudhir and Ravi Kanbur (1993), "Inequality and Development: A Critique", Journal of Development Economics, 41, pp. 19-43. Atkinson, Anthony B. (1987), "Income Maintenance and Social Insurance" in A.J. Auerbach and M. Feldstein (eds.), Handbook of Public Economics. Atkinson, Anthony B. and John Micklewright (1992), Economic Transformation in Eastern Europe and the Distribution of Income, Cambridge: Camb-idge University Press. Atkinson, Anthony B et al., (1984), "Unemployment Benefit, Duration and Incentives in Britain: How Robust is the Evidence", Journal of Public Economics, 23, pp. 3-26. Bishop, John A, John P. Formby and Paul D. Thistle (1991), "Changes in the US Earnings Distributions in the 1980s", Applied Economics, 23, pp. 425-434. Blank, Rebecca M. (1993), "Public Sector Growth and Labor Market Flexibility: the United States vs. the United Kingdom", National Bureau of Economic Research Working Paper No. 4339, April. Creedy, John (1985), Dynamics of Income Distribution, London: Basil Blackwell. Danziger, Sheldon, Robert Haveman and Robert Plotnick (1981), "How Income Transfer Programs Affect Work, Savings, and the Income Distribution: A Critical Review", Journal of Economic Literature, 19, pp. 975-1028. Dumke, Rolf (1991), "Income Inequality and Industrialization in Germany, 1850-1913: the Kuznets IHypothesis Re-examined', in Y.S. Brenner, Hartmut Kaelble and Mark Thomas (eds.), Income Distribuion in Historical Perspective, Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de l'Homme. Dye, Thomas R. and Harmon Ziegler (1988), "Socialism and Equality in Cross-National Perspective", PS:Political Science and Politics, Winter, pp. 45-56. Hirschman, Albert 0. (1973), "The Changing Tolerance for Income Inequality in the Course of / 36 Economic Development: With a Mathematical Appendix", Quarterly Journal of Economics, 87(4), pp. 544-565. Kaelble, Hartmut and Mark Thomas (1991), "Introduction", in Y.S. Brenner, Hartmut Kaelble and Mark Thomas (eds.), Income Distribution in Historical Perspective, Cambridge and Paris:Cambridge University Press and Editions de la Maison des Sciences de l'Homme. Kuznets, Simon (1955), "Economic Growth and Income Inequality", American Economic Review, 45:March, pp. 1-28. Kuznets, Simon (1966), Modern Economic Growth: Rate, Structure and Speed, New Haven: Yale University Press. Lecaillion, Jacques, Felix Paukert, Christian Morrisson and Dimitri Germidis (1984), Income Distribution and Economic Development: An Analgtical Survey, Geneva:ILO. Le Grand, Julian (1982), The Strategy of Equality: Redistribution and the Social Services, London: Allen and Unwin. Lindert, Peter H. (1989), "Modern Fiscal Redistribution: A Preliminary Essay", Agricultural History Center University of California, Davis, Working Paper Series No.55, June. Lindert, Peter H. (1991), "How Welfare Spending Evolves", Agricultural History Center University of California, Davis, Working Paper Series No.66, July. Lindert, Peter H. and Jeffrey G. Williamson (1985), "Growth, Equality, and history", Explorations in Economic History, vol. 22, pp 341-377. Lydall H.F., The Structure of Earnings (1968), Oxford: Oxford University Press. Meron, Monique (1991), "La dispersion des salaires de l'Etat 1982-1986", Economie et Statistique, No. 239, January, p.37. Morrison, Christian (1984), "Income Distribution in East European and Western Countries", Journal of Comparative Econcmnics, vol. 8, No.2, June, pp. 121-138. Oshima, Harry T. (1991), "Kuznets' Curve and Asian Income Distribution", in Making Economies More Efficient and More Equitable: Factors Determining Income Distribution, ed. Toshiyuki Mozoguchi, Economic Research Series No.28, The Institute of Economic Research, Hitotsubashi University, Tokyo: Kinokuniya Company Ltd and Oxford University Press. Pareto. Vilfredo (1897), Cours d'Economie Politique, Lausanne. Paukert, Felix (1973), "Income Distribution at Different Levels of Development: A Survey of Evidence", Internationa' Labor Review, Vol. 108, Nos. 2-3, pp. 97-125. 37 Perotti, Roberto (1991), "Income Distribution, Politics and Growth: Theory and Evidence", mimeo, Columbia University, 1991. Perotti, Roberto (1992), "Income Distribution, Politics, and Growth", American Economic Review Papers and Proceedings, May, p.311. Persson, Torsten and Guido Tabellini (1992), "Growth, Distribution, and Politics", Thieory and Evidence", in Alex Cukierman, Zvi Hercowitz and Leonardo Leiderman (eds), Political Economy, Growth and Business Cycles, Boston, Mass.:MIT Press, pp. 3-22. Phelps Brown, Henry (1977), 7he Inequality of Pay, Oxford: Oxford University Press. Polak, Ben and Jeffrey G. Williamson (1991), "Poverty, Policy and Industrialization: Lessons from the Distant Past", World Bank, Policy, Research and External Affairs Working Paper No. 645, April. Pryor, Frederick L. (1971), "Economic System and the Size Distribution of Income and Wealth', Indiana University, International Development Research Center. Pryor, Frederick, L. (1973), Property and Industrial Organization in Communist and Capitalist Nations, Bloomington and London:Indiana University Press. Psacharopoulos et al (1992), Poverty and Income Distribution in Latin America: The Story of the 1980s, Washington, D.C.: World Bank Latin American and the Caribbean Technical Department, Regional Studies Program, Report No.27, December. Ram, Rati (1991), "Kuznets's Inverted U-Hypothesis: Evidence from a Highly Developed Country", Southern Economic Journal, vol. 57, April, pp. 1112-1123. Sawyer, Mark (1982), "Income Distribution and the Welfare State", in A. Boetho, The European Economy, Oxford: Oxford University Press. Soderberg, Johan (1991), "Wage Differentials in Sweden, 1725-1950", in Y.S. Brenner, Hartmut Kaelble and Mark Tnomas (eds.), Income Distribution in Historical Perspective, Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de l'Homme. Summers, Robert and Alan Heston (1991), "The Penn World Table (Mark 5): An Expanded Set of International Comparisons", Quarterly Journal of Economics, vol. 106, No. 2, May, p.327. Thomas, Mark (1991), "The evolution of Inequality in Australia in the Nineteenth Century", in Y.S. Brenner, Hartmut Kaelble and Mark Thomas (eds), Income Distribution in Historical Perspective, Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de l'Homme. Ubieto, Antonio, Juan Regla, Jose Maria Jover and Carlos Seco (1972), Introduccion a la Historia X 38 de Espana, Editorial Teide, Barcelona, p.731. United Nations Development Program (1992), The Human Development Report, New York: UNDP. Williamson, Jeffrey (1991), "British Inequality during the Industrial Revolution: Accounting for the Kuznets curve", in Y.S. Brenner, Hartmut Kaelble and Mark Thomas (eds.), Income Distribution in Historical Perspective, Cambridge and Paris: Cambridge University Press and Editions de la Maison des Sciences de l'Homme, 1991. Williamson, Jeffrey (1991a), Inequality, Poverty, and History, Cambridge, Mass.:Basil Blackwell. / 39 Aman Tae 1. State sector employmrm as perntage of bhar fote or seo r cotneiy active popudtieo COUN1RY STATE YEAR a/ COMPONENTS SOURCES SECTOR date ector employment (all enployed) EMPLOY. OECD Australia 29.3 1986 governmend + SOEA (alI=labor force) Australia satistical yearbook 19S9 (p.171) Austria 37.9 Avg75-SO(G) idem OECD Economic Studies, Spring 1915, No.4 ._______________ Avg7S-79(S) delgium 22.5 Avg75-tO(G) idem OECD Economic Studies, Spring 1985, No.4 Avg75-79(S) Canad 24.1 Avg75-80(G) idem OECD Economic Studies. Spring 19S5, No.4 Avg75-79(S) Denmttrk 9.4 Avg75-80(G) idem OECD Econonmic Studies, Spring 19S5, No.4 Avg75-79(S) _ Fusland 2S.7 19S9 public ector: productive + non-productive Fmnband scUtistical yearbook 1991- .361) _________________ (all labor force) France 21.2 1984 government + S0ESs + health, education and France saatistical yearbook 19SS welfare (alllabor force) W. Gennany 22.3 Avg75-S-(G) idem OECD Economic Studier. Spring 1985, No.4 Avg75-79(S) Greece 10.7 1986-87 except general govt + health and education, Iransport Greece statistical yearbook 1988; Rutkowska (1991) govt 1975 and telecom workers (all=econ. active popul.) Ireland 19.6 Avg75-SO(G) idem OECD Economkic Studies. Spring 1915, No.4 Avg7S-79(S) Italy 20.9 Avg75-SO(G) idem OECD Economic Studies, Spring 1985. No.4 ___________ Avg75-79(S) Japan 9.5 1986 SOEs + public health and educalion (a l=labor Japan statistical abtract 1991 force) _ Netherlands I 5.0 1987 public sector: productive + non-productive Netherands statistical yeagbook 1981 (pp. 133, 140) (all labor force) New 7etuad 24.7 Avg75-80(G) idem OECD Economic Studies Spring 1985, No.4 Avg75-79(S) Norway 24.8 Avg75-80(G) idem OECD Economnic Studics Spring 1935, No.4 Avg75-79(S) Portugal 14.2 19S1 general govt + SOEs OECD Economic Studies, Spring 1915, No.4; Portugal statistical yearbook 1982 (pp.41,62) Spain 13.7 19S2 general govt OECD Econonic Studies, Spring 9115, No.4 40 COUNTRY STATE YEAR .1 COMPONENTS SOURCfS SECTOR tate sector enployment (all employed) EMPLOY. Sweden 36.2 Avg75-80(G) idem OECD Economic Studies, Spring 19S5. No.4 Avg75-79(S) Switzeritand 10.4 1982 general govt (all=labor forxe) OECD Economic Studies, Spring 1915. No.4 Turkey 13.6 1990 govt + SOEs (ail=emaployed) World Bank Turkey Data Base United Kingdom 22.S 1919 general govt + SOEs (&II= labor force) UK Cenmral Statistical Office, Social Trends No-21 (1991) United Stes 5.1 1985 govt employment (all=lahor force) Esping-Anderwen(1990; p.202) Eastn Europe Bulgaria 91.5 1918 acisliat sector (all =labor force) World Bank Country Study, Bulgaria: Crias and Transition to a Market Economy (1991, p. 331). Czechoslovakia 91.3 19S9 Mate sector + cooperatives (all= labor fomce) Czechoslovakia statistical yearbook 1990 (p 19S) Hungart 93.9 1981 state sector + cooperatives (all=labor fomce) Hungary statistical yearbook 19S (pp. 66-67) Poland 70.4 1919 scialized sctor (all= labor force) Poland statistical yearbook 1990 (p.93) Rotania 95.2 1919 state scctor + cooperatives (all= labor force) World Bank Country Study, Ronsanis: Tle Challenge of Transition (1991, pI). Forner Yugoslavia 71.9 1919 socialized asctor (all=labor force) Yugoslavia statistical yearbook 1990 Former USSR 96.3 1111 staue sctor + cooperatives (11= labor force) Soviet Union sttistical yearbook 1988 (p.33) E. Gernmny 94.7 1987 state sector + cooperatives (all=labor foce) East Germany satistical yearbook 19S(p. 112) Iem indicates that dhe conpoaenta are the sarne as in the entry under Austnlia, i.e. general government plus public ector. SOEs e- ate-owned enterpries. 'These are public sector enterpriesa as defined in each country. eI Avg 75-10 (G) dewotes the r crige governient employntr (G) in the period 197S-S0; Avg 75-79 (S) denotec the average employment in state-owned eaterprises or public sector (S) in the period 75-79. 41 Amnt Tabh I 60.) COUNTRY STATE YFAR COMPONENTS SOURCES SECTOR ae setor enpboyrnent (all enployed) EMPLOY. Ah*a Algenia 50.2 87 public ctor (all-econ.active pop.) Algeris satiskal yatrbook 1990 (p.47) and FAO production yearbook 1987 Egpt 19.3 79 non-financial public cal.+ gereral govtw(all1eon-activepop.) Hellcr & Tait (1983, p.40) and FAO production ycerbook 1987 Gabon 8.4 S9 govt + SOEc (all-labor force) Cabon Direction GCnkrsle de I'Econonue (1990, p.14), Wodd Bank Social Indcaton of Developamet 199 1-92 and The Word Factbook 1992. p. 128 Ghana 12.4 a5 govt + SOEa (all-labor force) Ghana Quarterly Digt of Statistics. Septenmber 199, p.48 and The World Factbook .1992p.102 Cose d'lveire 11.3 36 publi ector cnployces (all -econ-active pop.) Cakulatcd fiom Appkton, Collier & Honrhell (1990, pp. 7 and 22) and Marcel (1992. _____________ ________ p.94) Kcnya 7.5 S0 non-fiancial public enterprises + general govt (all-econ. Heller & Tail (19S3, p-40) and FAO production yearbook 19S7 active pop.) Madagascar 3.1 80 non-financial public enterprises + general govt (all econ. Helkr & Tait (1983, p.40) and FAO productionyearbook 1987 ctive pop.) Morocco 5.0 87 general govt (ll-econ. active pop) Morocco statistical yearbook 19S9. pp- 23. 367 Nigeria 3.3 77-84 federal, state local govt + SOEs (all=econ. active pop.) Bienen & Diejonioh (1981. p. 107). FAO production yearbook 1979 for SOEt-UNDP and World Bank. African Developmert Indicators, 1990. p.262 Senegal 3.4 76 non-financial public enterprisec + gencral govt (l U-econ. Helkr & Tait (1983, p.40) and FAO production yearbooks active pop.) Siem Lone 1.3 79 SOEs (all-econactive pop.) Milanovic (1989, p. 7) South Africa 13.2 S5 SA transpoe + central govt + ptovincial and local authorities South Africa yearbook 1987-88 (p.752) (all-econ. active pop.) Swaziland 7.5 S2 non-financial public cterprisecs + general govt (all-econ. Heller & Tait (1983, p. 40) and FAO production yeadmooks active pop.) Tanzania 6.0 78 non-financial public entcrprisea + general govt (all-econ. Heller & Tait (19S33 p. 40) and FAO production yearooks active pop.) Zambia 13.2 so non-financial public enterprisecs + general govt. (all=ccon. Hclkr & Tait (1933, p. 40) and FAO production ycaeoooka active pop.) Zinbabwe 15.2 24 govt + SOEs (all-labor force) Zimbabwe saistical yearbook 1987 (pp.50 80) Asia An.~~~~~~~~~~~~~~~~~ Bangladesh 4.2 831/4 govt + nationalizcd centepriwes (all=ecron. active pop.) Bangladesh yearbook 1986 (pp. 210, 229, 234) and FAO production yearbook 1984 China 20.4 87 sate + urban coop cmployces (all -ccaa. active pop.) China yearbook 19S8 (p. 153) and FAO production ycarbooL 1981 4? COUNTRY STATE YEAR COMPONENTS SOURCES SECTOR state sector employment (all employed) EMPLOY. 12.2 90 pubik administration + publik ervices (all=employed) Cyprua economnic and ocial indicators (1991, p.25) Cyprus _ _ _ _ _ _ _ _ Hongkong 7.9 90 civil service + public project employees (all= labor force incl. Hongkong annual digcst 1991 (p.34) and Hongkong Semi-annual rport (1,°91 p.67) unemnployed) lrel 27.1 37 public nd commercial etcvkes cxcl. public enterpriacs; Irael sttistical abstrct 1938 (pp. 332. 340) (all-civilian labor fowe) Wnia 6.0 77 non-finatncial public enterprisea + general govt. (all fecon. Heller & Tait (1983, p. 40) and FAO production yea.books active pop.) Indonesia 5.1 90 civil ervice exci public enterprises; (all=age 10+) Indonesia aatistical yeaebook 1991 (p. 61. 66) Iran 26.9 36 public aector (govi+SOEa) (all=econ. active pop.) Iran statistical yearbook 1989/90 Jordan 22.2 86 public sector (govt + SOEs) (all =dotnmeic labor fome) Jordan tatiatical yeatbook 1987. pp. 57, 69 Korea S. 9.3 81 non-financial public enterprises + general govt (all =econ. Heller & Tait (19S3, p. 40) and FAO production yearbooks active pop.) Malaysia 8.4 85 govt. employed (all -labor fotce) World BDnk Malaysia report No. 8667-MA, p39; World Bank Malaysia report No 10758-MA. p.30 Pakistan 238 74n5 SOEs (all-econ. active popul.) Milanovic (1989. p.17) Phillippines 11.3 79 non-financial pubik enterprises + general govt (all econ. Heller & Tait (1933, p. 40) and FAO ptoduction yearbooks activc pop.) Singapore 10.4 30 govt. + nmjor public companies (all= econ. active pop.) Singapore atistical yearbook 1933 (p. 64) and 1980-SI (p.45) and Pilbai (1983, tabk __________ ~~~~~~~~~~~~~~~~~~VI) Sri Lanka 23.3 s0 non-financial public enterpriees + general govt. (all-econ. Hcllkr & Tait (1933. p. 40) and FAO production yearbooks active pop.) Taiwan 9.9 75 govt. emtployccs (all =econ. active pop.) Taiwan yearbook of labor sitics 1937 (p.33) and 19T7 (p.1I) Thiland 6.2 33 govt. employees (all =econ. active pop.) World Bank Tlhiland repot No. 9627-TH, p.69 Lodi Ameeria Argentina 15.2 SI non-financial public enterprises + generl govt. (all-econ. Heller & Tail (1983, p. 40) and FAO pgoduction ytctootks active pop.) Bahalma 18.6 7S non-financial public enterprises + general govt. Heller & Tait (1983, p.40) and ILO yearoook at labor statitcs 1985 ( D-cgmployed) Boivia 18.3 39 public sector (all -employed) Psacharopouloet al. (199, anmx 14) Brzil 11.7 a govt. -+ federal and pfovincial public enterprises (ail-econ. Berg & Shirky (1987, p.21); Paul Singer (1939. p.il) and FAO ptoduction yeaebook active pop.) 191 (p.26) Chik 9 2 89spublic wtor (all -employed) Pscharopouloet al. (1992. ama 14) I COUNTRY STATE YEAR COMPONENTS SOURCES SECTOR sate ector empiloymer (all employed) EMPLOY. Coloabbia 10.7 89 public sector (all-employed) Pacharopoulohei dl. (1992, annex 14) Coma Rica 16.9 S9 public ecuor (all- employed) P scharopoulo ct al. (1992, annex 14) Cuba 82.4 13 nate aector incl. agriculture (al -employed) Rudolph (198S, p. 299) Ecuador 23.7 82 govt. + comrnunity arvicke (all-labor fore) Hanratty (1991. p. 256) Guaternal 5.8 al non-financial public eat. + genemrl govt (all econactive Heller & Tait (1983. p.40) and FAO production yeadbooka _ _ _ _ _ _ _ _ _ ~~~~~~~~pop.) Hondurms 9.6 89 public sector (all-employed) Psacharopoula el al. (1992, annex 14) Janaics 11.0 91 goveunmea employees (all-labor force incl. wif-employed) Pxacharopouloaei al. (1992, anncx 14) Mexko 21.4 85 public actor incl. public enerpriae (ad-employed inc. self- Glade (1990, p.41) cmployed) Panama 17.3 79 non-finsncial public ererpsries + general govt (all-econ. Helkr & Tai. (1983. p.40) and FAO production yearbook 1979 active pop.) Pern 14.8 89 public ector (all -employed) Pracharopouloset al. (1992, annex 14) Unauay 21 4 89 public secor (all -employed) Parcheropoulo et al. (1992. anex 14) Venezuela 19.3 S9 public sector (all-eniployed) Pacharapoukoset al. (1992. annex 14) REFERENCES Appktoa. Simon, Paul Collier and P. Horuhell (1990), 'Gcnder, Education, and Employment in Cote d'lvoire', Washington. D C.: World Bank, SDA Working paper No. S. Berg, Elliot and Mary M. Shirley (1987), 'Divestiture in Developing Countrics', World Bank Diacuraion Paper No I , Washington, D.C.. The World Bank. Bienen, Henry and V.P. Diejomanoh (19S1), nke Poriacl Econc y ofllcome Disrbibdon in Mgeroa. New York: Holmes and Meier Gabon Direction Genrnia de I'Economie (1990), 'La conjoncture gbonaise i mi-S9', Ecomie ct Finances, vol. 10, January, p.14. Glade, William E. ed. (1990), Privaozadon ofPubic Enterprises in Lawn Amenca, International Center for Economic Growth, Inaimut of the Americas. Cenger for U.S.-Mexican Siudie- Hanrauy, Dennis M., (1991), Euador: A Country Study, 3rd edition, Washington, D.C.: Foreign Area Studies, The Anerican Univeraity. Heller, Peter S. and Ala A. Tait (1983), Gowvnment Enploynen: and Pay: Some lnernaanaonal Comparisons. Washington, D.C.: Interntional Monetary Fund, October. Esping-Aadcrsen, Goga, he Zhrec WorUd of Wey4arc Socialism, Princeton Pndccton University Premr, 1990. Marcel, KB. (1992), 'Reinatoursation et Evolution de l'enploi dana k ecteur public ct pam-public cn CO&e d'lvoir', Afilque e# Develoapmen: AJWca Dewlkamert, No.l, p.93. Milanovic, Bmnko (1989), LiberUnzadon and Enserprrneunhip: Dynamis ofRerjma hi Socdalism and Capiralimn, Amwonk, N.Y.: M.E. Shpe. Pillhi, Philip Nalliab (19S3), Slr Enirpree in Singapor: Legal amporsance & Devlom,ent. Singapore: Singapore University Paes. (1992), sudies pmgrom, Repo No.27, Decearber. Rndoipb, Jamne D.(ed) (1985), Cuba: A Co.ay Study, 3rd edition, Washington, D.C.: Foreign Ama Studics, The American Univerity. Rutkowska, lmbebs (1991), 'Public Transfcrs in Socialist and Market Econonmics', Research Project Social Expcnditures and their Disitibutionsl Impact in Eastcrn Europe. Pcper No 7, Washington, D.C.: Socialist Econoties Refom Unit, World Bank. Singer Paul (1939), 'LA class obrera trenle a l cnisis inflacionaria y a denmocrstizncion en Brasil', Economla de America Lain, 15-19:17. 45 Anne Table 2. Social transfen kas and in-kind) - peentage of CDr COUNTRY SOCLL YEAR COMPONENTS SOURCES TRANSFERS OECD Australia 17.1 av. 190a Includes heahh + educalion + cducalion + pcnsiona + family Rutkowska (1991) allowances + sicknacanumtmily aiowances + unemployment benefita + welfarce. unics otheriwie indicated Austria 27.9 &v.19U0h Ruikowska (1991) Belgium 30.3 v. 19S0a RuikowAk (1991) Canada 215 19SS OECD, Social Expenditures 1960-1990 (1985. Table I p. 21) Dennurk 33.3 19SN OECD, Social Expenditurcs 1960-1990(19SS. Table 1. p.21) Finland 22.0 av. 1980c Rulkowska (1991) Francc 31.0 av. 19S0 Ruikowska (1991) W. Ccrmany 25.7 *v. 1950e Rutkowska (1991) Grece 16.7 v. 19S0 Rutk'wska (1991) Ireland 25.1 av. 19S0a Rulkowska (1991) Italy 244 av. 1980 Rutkowska (1991) Japan 17.5 19S1 OECD, Social Expcndiurma 1960-1990 (1985. Table I, p.21) Netherlands 31.1 av. 19S0s Rulkowska (1991) New Zealand 19.6 19S1 OECD, Social Expenditures 1960-1990(19SS. Table 1, p.21) Norway 27.1 1981 OECD, Social Expenditures 1960-1990(1985, Table 1. p.21) Protugcl 17.1 av. 190a Rutkowska (1991) Spain 11.1 av. 1 90S Rulkowsks (1991) Sweden 32.2 v. 1950s Ruikowsk (1991) Switzerland 14.9 1979 OECD, Social Expenditure 1960-1990 (195, Tabkc 1, p .21) Turkey 7.3 av.19S0s Rutkowska (1991) United Kingdom 19. av. 19S0a Rutkowaka (1991) United Statcs 17. 7v. 19S0c Rutkowsks (1991) Eine ____p__ __ __ _ COUNTRY SOCIAL YEAR COMPONE.NTS SOURCES TRANSFERS Bulprie 17.9 av.1980 World Bank Country Study. Bulgana: Crisis and Transition to a Market Economy, 1991, vol 2 (Tabkl 6.6, 9.3 and Appendix Tablc 15). GDP from ibid, vol I, p :36. Czechoskovkit 21.3 wv.19S0c Ruikowska (1991) Hungary 9.9 _ v. 1980c Rutkowska (I991) Poland 17.5 v. 1980e Ruskowaka (1991) Romanis 11.7 av. 1980 pcntions, family allowcnces, sicknesa and maternity beneCfits, World Bank Country Study. Romainia: Human Rsources and the Transition to Market health and education. Economy, 1992 (Tabkl 3.1, 5.21. 4.25). Forntr Yugoslavia I.- av.1980s Rulkowaka (1991) Former USSR 15.7 1985 Statistical Offices of Austria, Poland and the USSR (1989, pp.32-3) E. Gernuny 20.2 1985 cash benefits, health and education Statittcal pocketbook (Or die GDR 1938 (pp. 25. 108) Aue Table 2 (coal.) COUNTRY SOCIAL YEAR COMPONENTS SOURCE TRANSFERS Africa AlgeFia 8.6 S6 heatkh, pension, ind.injury, (anily and holiday allowances Algeria astistical yeateook 1990 (p.l 13) and IMF Intemcational linancial Statistics 2991 (P.191) Egypt 7.7 tS social insurance, family benefits, health, public assistace, ILO, The coat of social ecurity 19W86 (tabk 3); education - World Bank WDR education 1937 Gabon 2.3 a5 idem; cxcl. education ILO, The cost of social wecufity 19446 (tabbk 3) Ghana 4.7 t' social aecurity, health, education World Bank Ghana report No. 9475-GH, p. 02 Coke d'lvoire 7.3 35 idem; education (2984) ILO, The coat of social aecnity 191446 (tabbk 3); educatio - World Bank Word Resoufeee 1992-93 (p.240) Kenya 5.6 85 idem MO. The cost of ocial curity 2984-56 (tabke 3); education - World Bank WDR Madagascar 3.2 85 idem LO, The cost of social security 19-6 (table 3); and World Bank Madagascar mepott No. 9101-MAO Morocco 6.3 36 education and beahh Moniswo (1991, p. 1637) Nigerc 1.02 35 Wen; education (1975) MAW, The cost of social scrity 1916 (tabie 3); education - Bienn A Diejnomsh (19S1, p.463; IMP _ucmliial Fuuwcial Statistcs 1991 (p.570) 47 COUNTRY SOCLAL YEAR COMPONENTS SOURCE TRANSFERS Senegal 6.1 a5 idem; educatior (I 914) ILO, The coa of social accurity 1914-6 (table 3); education Wotid Bank World Reources 1992-93 (p.240) Sierr Leon S .8 S5 social welfamr, education, health UN, National accouns statitiks: Main Aggtcptes tnd Detailed Tables, 1990. pp. 1666- 67 South Af6ica 3.9 I6 social ecurity, education, hcalth Moll (1991, p.79) Swaziand 5.9 15 idem; education (1987) ILO, The cod of social scurity 1984-6 (tabk 3); edt-srion = Lketat du mondc, edition 1991Paris (p. 301). Tanania 1.9 S5 idem; education (1986) ILO, The cod of social security 1914-S6 (table 3); education = World Bank WDR 1988 Zambia 6.9 85 idem; education (19t6) fLO, The cog of social scurity 1984-S6 (tablc 3); education = World Bank WDR Zimubbwe 12.2 84-5 education, health and social welfae Zimbabwe satisical ycerbook 19t7 Asia Bangladesb 1.1 85 idem; education (19t6) ILO, The cost of ocial security 1984-66; education = World Bank WDR 1tt8 China 12-0 BB cash social welfare, cash subsidies, education and health China statistical yearbook 1992 (pp. 31. 223, 799. 107) and 1989 (p. 151) Cyprus S1 86 idem; education (1919) ILO, The cost of social ecurity 19U4-S6 (table 3); education = Cyprus econornic and social indicators (1991) Hongkonr 2.9 11 cash & non-cash social welfare (excl.pcmsiona;), health, Chow (1915, p-73); }ongkong annual digcst of sttistics 1990 (pp. 111, 122) education lael 22.1 S5 iderm LO, The cost of social wccutisy 19U-86 (tabte 3); education = World Bank WDR 19t7 India 1.8 S 5 idem ILO, The cost of cocial security 194-86 (table 3): education = World Bank WDR 1987 bdonesis 2.4 S5 idem ILO. The cost o social security 19U4-86 (table 3): education = World Bank WDR 1987 Irn 7.9 15 health, scial security, cducation IMF Govemrnseit Fincncial Statistics, yearbook 1991 (pp. 321-2) Jordar. 5.4 85-S6 health, social security, education Musallan (1990. pp. 132-33; also Annet A, Table IOAI); education = Wodd Bank WDR 1917 Kora S. 2.9 91 social scurity, social assistance (budgel), health and edt cation World Bank Korea Report No. 10733-KO (p. 16) Malaysia 3.0 S5 idem; education (1912) MO, TIe cost of social security 1984 6 (table 3); education - World Bank WDR 1985 Pakitan 1.6 I5 ide ULO, The cot of social security 191446 (table 3); education - World Bank tWDR 1917 Phillippinec 2.3 S5 idem LO, The cost of socil secuAy 191446 (table 3); education - Word Bank WDR 19S7 48 SOCIAL YEAR COMPONENTS SOURCE TRANSFERS Singapore 18.3 85 idem LO. Tlhe coat of scial scuriy 1984-6 (table 3); education = Wordd Bank WDR 1987 Sri Lanka 4.6 85 idem 11.0, The cost of ocial ecurty 19U4-86 (table 3); education = World Bank WDR 1937 Taiwan 8.d 85 social sccurity, cducation, science, cultumr Taiwan gatiuiical databook 1992 (pp. 25, 157) Thailand 4.3 as idcm 1lW Thc cost of social secunty 1984-86 (table 3); educaoon Worid Bank WDR 1987 [Mi. Amneca Aignhina 7.6 85 ide ILO, The cost of social scurity 1984-86 (table 3); education = World Bank WDR 1987 Babamas 1 2 a5 idem; excl. education ILO, The cot of social sccurily 194-86 (table 3) Bolioin 6 J 35 idem 1.0. Thc cost of scial security 191416 (tabk 3)); education e World Bank WDR 1987 Brazil - 5.5 85 idem ILO. The coot of social sccurity 1984-86 (table 3); education World Bank WDR 1987 Chile 19.1 83 social ecurity, education C. Mesa-Lago (1991. p. 19); educalion World Bank WDR 19U6 Coombina 2.0 35 idem; excl. education ILO, Tbc cos of social scurity 1984-86 (table 3) Cosu Rica 12.2 S5 idem 1O1. Thec cost of social security 19U4-U6 (table 3); education - Wodd Bank WDR l587 Cuba 19.4 85 ide,m (O. The cost of scial aecurity 194-6 (tabk 3); education - Cuba uatistical ycarbook 1988 (p.195) Ecuador 68 85 s deid1m 0 The cost of social security 19"46 (tabk 3); education - World Bnk WDAR 1987 Guatmrala 3.3 a5 idem; education (1990) IU0, Tbhc cogt of scial wcarily 1964486 (tabk 3); education - Wodd Bank WDR 1992 Honduras 3.1 as5 iem; cxci. education IO, Tbe cog of social ecurity 194466 (tab!e 3) 1amaics 5.5 55 denm IL0, Te cod of scial secutry 19U4-86 ~tble 3); Boyd (1938 pp- 6. 111); IMF (1991. p. 458-9) Mcxico 5.6 55 idem MD, The coo of mid tecuit) 19U446 (table 3); education - Wod Bank WDR 1987 Panma 13.2 Ss idem; education (1986) RD. Tbe costd of oial socwily 1984-86 (tabk 3); education - Wodd Bank WDRt 1988 Pctu 3.4 35 ideag educatio (1983) IW. Th coat of social sany 194U6 (tl 3); edscatioa - Wodd Bank WADR 1986 49 COUNTRY SOCIAL YEAR COMPONENIS SOURCE TRNFERS UrupY 10.45 i5 idem[ 1W. TMe coat of mocial ecurity 191U46 (tudc 3); eduation - Wodd Bank WDR 1937 Veceual 5.6 S5 idem RD, Tbc cost of social secty 19U46 (table 3); educt to - World Bank WDR Mem denotes tbe une caapms - ia the ctcdy under Egyp Th ue componenl ae social insuece, faily benefits, lhea cam, social i_ e be accaes for public ecor employees (if acpsute). public aseance and education expedture AU items except educatio me obtained from ILO, The Corl of Social Security. Education expeadiures ae obiaid sepamtely mor oftcn fr,m the Word Bank World Deverpsav Nsp.e (WR). If educatio, data do bc mfcis, to sam yea ath eat of duh dat, dim is foowd by edarealon 6ara. REFERENCES 11_m HNesy and V.P. Diejoe_b (1911), The Poll EAwanry of Icom Ddbuion in Mgedh, New York: Holmes iW Meier. foyd. Dck (I 9M), Ecowkmc Management. Jncme Dlalba ad Powrgy in Jamaia. New York: Praepr. Clow, Nelao, 'Hong Kong' in Duion, Jo and Hyung SMik Kim (1985), S&1 WeLfarr in Ada, L1ndon, Sydney: Cnrun Helm. p. 63-92. Ay!rLago, Carce. (1991), SociaI Security and Ptmpeeu for Equity in Latin Anerica', World Sank: World Bfnk Diauaion Papen, No. 140. Moll, Peter G,. (1991), The Gre DrEconk Orbare, Johannesburg: Skotavilc Publidiern. Motis, Chrisi (1991) ' itt , I.omeand Poverty in Moocco', WorU D0evopmcnr, vol. 19, November, p.1633-51. Musallm, Ghaaaa(1990), tcuritySchemandIncomnDidtibutioa'inKsmilAbu Jber,Matthes ubbeandMahaundSiadi(eda).Incom,lDinrbdoninJordan,Boulder,SanFnciao:Westview Special Studies on th MidM Rutkowca, Izabela (199!). 'Public Tranfcr in Socialist and Market Economies', Resarcb Project Social Expenditures and their Distuibution[l Impact in Eastcrm Europc, Paper No. 7, Wathington, D.C.: Socialist Economics Refogn Unit, World Bak. Statitical Offices af the USSR, Austria and Poad (1989), Conparusive Analyas ofSodak Expendinures ad its Famnce, Moscow, Vw.t, WNraw. 50 Annet Table 3. Within-ountry rtgienal heterogeurSty (ratio of incomnet betwev uost devdoped and leAst derdoped regieo COUNTRY RATIO IN YEAR COMPONENTS SOURCES INCOMES OECCD Australia 1.24 19S7 income of wage eorners per capita (all mates) Australia statistical yearbook 1989 (p. 734) Austria 1.22 1990 median gros incotme of employees and the elf-enmployed (9 Austtris satisical yeattbook 1991 (p. 144) prov:nCes) Belgium 1.38 1979 hoscehold incone (II regions) van Weeren and van Paag (1984. pp- 239-270. Table 2) Canada 1.53 1986 family incomne (all provinces) Canada statistical yearbook 1990 (Table 5.62, pg.S-34) Dennark 1.44 1979 houehold income (23 regions) an Weeren and van Poaag (1984, pp. 239-270, Table 2) Finland 1.49 19t6 individuas' incomfe (all provinces) Finland statistical yearbook 1991 (Table 276, p.303) France 1.39 1979 household incone (9 megions) van Weeren and van Pruag (1984, pp. 239-270, Table 2) W. Gertmany 1-62 1979 household incone (10 lander and West Berlin) van Weeren and van Psag (1984. pp. 239-270. Table 2) Greece 2.88 19S1/89 domiestic use of electncal energy per capita (10 regions) Greece satistical yearbook 1988 (Table 11:6, p.17 and Table XI:I l, p.310) Ireland 1.31 1981 vehicles per capita (by county) Ireand satistical abstrct, September 1986 (pp.27 and 327) Italy 1.68 1979 household inconme ('0 regions) van Weeren and van Prhag (1984, pp. 239-270, Table 2) Japan 3.32 1978 incomne per capita (all prcfectures) Japan satistical yearbook 1981-82 (Table 2, p.78) Netherlands 1.31 1979 household incone (I I prmvinces) vsn Weeren and van Phng (1984, pp. 239-270, Table 2) New Zealand 1.66 1980 average salary (by distnict) New Zealand, lncomnes& lnconmeTqx 1979-80(Tsble 21, p.32) Norway :.2S 1980 car per capita (by county) Norway satistical abstract 19S8 (Table 234. p. 17) Portugal 3.14 1986 wsge bill per 1000 persons (by district anrd autonomnous region) Portugal iststuscal yearbook 1989 (Table 13.1.3, p. l8) Spain 1.39 1985 incone per capita (all regions) Spain statistical yearbook 1990 (Table 1.4. p.8S6 and Table 1.5. p-8S7) Sweden 1.35 1988 income per capita (by Lounty) Sweden sttistical yearbook 1991 (Table 225, p.213) Switzerand 2.20 19tt incone per capita (all cantons) Switzerland statistical yearbook 1992 (Table 4.2) Turkey 4.06 19t6 GDP per capita (by region) World Bank Turkey data bas (Tables 1 and Table 2) United Kingons 1.16 1979 boushold income (10 regionsa van Weeren and van Psag (1984, pp. 239-270, Table 2) Urited States 2.07 1938 income per capita (all autes) US statistical abstract 1990 (Table 706, p.437) 51 COUNTRY RATIO IN YEAR COMPONENlS SOURCES INCOMES Bulgria 1.35 193 tv per 1000 persona (all counties) Bulgaria salistical yearbook 1989 (Table XIlw, p. 502) Czechoslovakit 1.09 1988 income per capita (2 republics) Czechoslovakia Federal Statistical Ofrce (1990) Hungary 1.24 1939 tv per 1000 persons (all counthie) Hungary satistical yearbook 19S9-90 (Table 32.19, p-420) Poland 1.47 1990 telephones per 1000 persons (all voivodships) Poland statistical yearbook 1991 (Table m, pp. LVI -LVl) Romania 2.56 1935 iv per capita by counties Romfniar4 atistical yearbook I986 (Tsble I1,p. 13 andTable 221, p337) Formner Yugoslavis 7.3 U 1981/9 income per capita (S republics or autonomous pmvinces) Yugoslavia statistical yearbook 1991 (Table 203-5 p.445 and Table 205-2, p.476) Former USSR 3.00 1390 income per capita (IS republics) Braithwaite (1990, p-34) E. Gersany 1.14 19S7 retail trade per capita (14 regions; excl. ELast Berlin) E. Germany sttistical yearbook 19SS (pp. 1 and 651T) Annex Table 3 (cowt. COUNTR>- RATIO YEAR VARIABLE (break-down by regi(.ns) SOURCE Africa Algeria 1.43 79/10 per capita expcnditure (5 zones) Algeria satistical yearbook 1990, No. 14. p 239 Egypt 1.31 s0 household income (2 tegions) Mohie-Eldin (1982, table 3.20) Gabon 6.9 77 average income (urban vs. rural) ILO (1933, p.23) Ghana 3 72 70 living standards indicator (3 regions) Boateng, Ewusi, Ksnbur and McKay (1990, p.29) Cote d'lvoire 3.4 75 per capita nril income (asl regions) IL) (1982, p.47) Kenya 23.2 76 per capita income (S regions) Bigaten (197S, p-40S) Madagascar 2.7 S0 household income (12 regions excluding large cities) Domosh, Bcrnier Sarris (1990, p.4?) Nigeria 6.1 777n3 per capita income (urban vs. rural) Jamal (19 lp. lS) Momcco 4.2 3o ownership of care per capita (7 areas) Morocco sutistical yearbook 1939. pp. 15, 219 Senegal 1.9 s0 per capita rural income (all regions) [LO (1932, p.47) Sierrn Leone 3.3 75r76 average inconme (urban vs. rural) ILO (1983, p.23) South Africa 4.2 79 per capit income (whites/blacks) Devereux (19S3, p.3S) Swaziland 6.9 74 average incone (urban vs. rural) ILO (1933, p.23) Tanzania 1.2 78 'verage income (non agricultural vs. farmer) 11 (19S2, p.49) 52 COUNTRY RATIO YEAR VARIABLE (break-down by region.) SOURCE Zanmbia 2.S 76 average income (urban vs. rural) ILO (1983, p.23) Zimbabwe 3.9 S3 taxable inconme per capita (4 regions) Zimbabwe abtistical yearbook 19S7 (Table 2.12. 7.14) Asi Bangladebh 3.19 79-80 per capita income (urban v. rural) Bangladesh Bureau of Statistics, Socio-econonmic indicators 19a 1, p. 1 13) China PRC 7.7 S7 per capita income (3 metropolitan tman and 26 provinces) China atistical yearbook 198S (p.55) Cyprus 1.0 regional difference non existant Hongkng I .0 regional difference non-existant Israel 1.0 mrgional difference non-existant India 1 69 74 consumer expenditure per capita (24 states and territories) India Depaitment of Statistics (197S, pp.7-19, 70-S3) Indonesia 6.94 S3 GDP per capita exciuding oil-producing regions (23 regions) Hill and Weidenann (19S9, Table 1. 1) Iran 6.35 76 percentage of households with electricity (urban vs. rural) UN compendium of social development indicators in ESCAP (Economic and Social Commission for Asia and the Pacific) 19S9, p.69 Jordan 2.47 86 household income (10 regions) Sha ban (1990, p 67) Korea S. 1.19 80 household income (urban vs. rural) Sang-Mok Suh (19S5, p-10) Malaysia 3.06 8O per cpita income (14 regions) World Bank Malaysia report No. 8667-MA Pakistan 4.83 SO percentage of households with eleciricity (urban vs. ural) UN comnpendium of social development indicator in ESCAP, p 70 Plhillippines 3 26 71 average family income (10 regions) Pemia (1977, p.7g) Singapome 1.0 regional difereoce non-existant Sri Lanka 5.74 Al percentage of households with electricity (urban vs. rural) Hil UN compendium of social development indicatorm in ESCAP, p. 71 Taiwan 1.85 aS household income (2 merrmpolitan areas and 21 counties) Republic of China (Taiwan) statistical yearbook 1989 (pp. I 4-7) Thailand 6.11 72 household income (5 rural regions and 5 urban regions) Chiswick (1981, pA6) Latin Amsesrca Argentina 6.21 S5 GDP per capita (22 regions) World Bink county study, Argentina: Provincial Government Finances (1990, p.l42) 8ahans 1 .9 90 household income (5 islands) Bahamas statistical abstract 1992 Bolivia 24.5 89 tax payments per capita (9 regions) Bolivia statistical yearbook 1989 Brazil 12.29 70 GNP per capita (26 states) UN ECLA, Distribucion Regional del Producto lntemo B8ito-Sectorial en los Paises de America Latina (1981, p.26) 53 COVNTRY RATIO YEAR VARLABlE (btak-down by regions) SOURCE Chsilb 4.92 76 GDP per capita (24 regions) UN ECLA, Diuribucioa Regional del Psducto lbsrno Rnro-Sectoril en lot Poises de Amnerica LetitA (19t1, p.32) Colombia 6.75 75 GDP per capita (24 departntens) UN ECLA. Distribucioti Regional del Psoducto Ituerno Bnto-Sectorial en los Paises de America LatisA (1911, p.30) Cota Rie 2.95 73 percentage of urban population (7 regions) lantzi (1976, p.28) Cuba 1.14 S8 averfge wage (IS regions) Cuba sttistical yearbook 1988 Ecuador 3.12 I5 GDP per capita (16 provinces) UN ECLA, Distribucion Regioral del Producto Interno Bruto-SectonIal en loe Paises de America Latina (1981, p.33) Guatenala 4.74 86/87 non-poor households as percentage of 1ll households (S Pinto et al (1992, p.80) regions) Honduras 15.9 79 cars per capita (18 departments) llonduras satistical yearbeok 1979 (pp. 4. 131) lantaica 3.39 91 percentage of population in receipt of por trlief (13 regions) lamaics economnic and social survey 1991 (pp. 15.3, 23.4) Mexico 6.92 70 GOP per capita (32 states) UN ECLA, Distribucion Regional del Producto Intemo Bruto-Sectorial en los Pai es de America IAtins (1981, p.36) Panama 3.76 68 GDP per capita (9 provinces) UN ECLAA Diatribucior Regional del Producto ptro Bnjto-Sectorisl en los Paiwe de Peru 16.42 77 GDP per capita (23 departments) UN ECLA, Disitribucion Regional del Producto Intemo Bruto-Sectorial en los Paise de America Laitina (1981, p.3t9) Uruguay 1.0 regional difTerence non-existant Venenels 2.92 19 population with access to sewage (19 regions) Venezuea rstatiisical yearbook 1989 (pp. 179. 648) REFERENCES Bigsten, Arne (1978), Regional Ineqsumaiy and Developnten: A Case Stady ofKenya. Nacionalekonomiska Institutionen, Gnteborg University. Braithwaite, Ieanine (1990), 'Povecty Differentils in the USSR: Implications for Social subility', mimeo, July. Chiawick, Canmel Ullman (1981), 'Income Distribution and Ernployment Progratnne', World Etnloytetr Programnte Rsearch Working Paper No. 97, February. Czechoslovakia Federal Statistical Office (1990), Mibu-estsis 1989:1.dil, Prague: Federil Statitiical Offec. Devereux, Steven (19t3), 'South African Income Distribution, 1900-90', SALDRU Working Paper No. 51, University of Cape Town, August, p. 73. Dorosh, Paul A., Rene E. Bernier and Alexander H. Sarris (1990). Macreco'nomic Adjusienm an d Ae Poor: The Case of Madaga5car, Coriell Food and Nutrition Policy Prognm Monograph No.9 Boateng, f. Oti, Kodwo Ewusi. Ravi Kmnbur and Andrew McKay (1990), 'A Poverty Profile for Ghana, 19t7-88'. Social Ditensions of Adjustment in Sub-Saharan Africa, Working paper No. 5, Waasington. D.C.. World Bank. 54 Hill, Hal and AnnaWeidenun(1919). 'Regional Developnuncin Imndonesia: Panens and lsues' in Hal Hill (ed), Unt and Divnsy: Regional Eec kDevelopmensiulmd as4ne 1970, oxford: Oxford Univemity Pm, pp. 3-54. ILO (1982). Rmol-w*mm Cdp ad lncme Dis rbudowSynaexis Repor of Sew,ece AJHw. C,.mhies, Addis Ababa: NL. MD (19UM). Api e ofDedarado, rv'pies and gw of Actd oldie World E;playment Conferene, Sixth aftcan regional conference, Tunis, October. India Depaumus of Statitkt, Tbleas on Conume Ezpenditurs', National Sanple Survey Oruaniation, No.240, New Dehi. J anl, Vail (1911). Rutl-UfbanC Gap and Inequality in Nigeri', ILO, Addis Ababs, Working Papera, September. Jamizi, Vemnon Eugene (1976), 'Sttutral Detenrnirmatt of bhe Lcation of Rural Development', Latin American Studies Prgram: Diuenation SeFies, Cornel Usiversity. January. Mohie-Eldia, Amw (1982), The Deveopaneh of the Share of Agricutural Wage Labor in bhe National Inomne of Eypt' in Gouda Abdel-Calkk and Robedt rw (eds). h Poditdcal EBonrm.y of Icome l Lsiribia o hI Ejg7p, New York, London. Holmes ad Meier. Pinto, Ileans dmi. (1992), 'Evoluio ofthe Mcroeconoi Policy and ha Effectacn the Pfoduction nd Markeing: Ptoceces of Non-Trdional Producta'. Gaudermlm:IF (laenational Food Policy ReAeab lnstiute) Special Pr*ct 2363-OW), Apil. Petnmx. Emnsto del Mar (1977). U,*.uzatim, PopeJon Grn,sh. and EBenIc Dewopmen in die Philippines, Wesport, Com.:Gretnwood Prem. Sha'b.,, Radvan AJi (1990), 'Econic Inequality in Jordan, 1973-19U' in Kamel Abu laber, Mathems Bubbe and Mahanunad Stndi (ede), Ihnc Distribution in Jolom, Bouder. San Francisco: Wcetvew Specil Studies on dhe Middle Eect. Song-Mok Sub (1915), 'Econornc Growth and Chne in lncxme Distibution: The Korean Case', Korea Developmnent Institute, Working Paper U15, Sepnmber. van Weeren, Hans and BDenud van Psaag (19S4).,-e Ineuratdy of Actual Incomer and Earning Capacities between Houwehod in Europe', E iwpem Ecmioni ArWew. 24:239-256. 55 As Table 4. The Gid ieadims Ci i Year Components Soumaes COUje coeffienA OECD Ausmfali 31.6 S1-12 D(pfYp) Bidhop, Fonrby, Smith (1991, Table 3 & 4). US dats. Autria 24.9 39 DONIYp,e);workers houaehoda Calculated fiom Austria Statistical "eadaok 1990 (p 161) Bdelgium 27.4 t3 D(bfY%b) Veleduc (1917, p.97) Canada 320 51 D(pY*p) Bishop, Formby, Smnith (1991, Tables 3 & 4). US data. Dem_rk 23.0 Estinated fiom Trakoglou (1992. p.27) Finland 202 S5 D(pIYp*,c) Ringen(1991) France 30 7 Sl D(p/Yp¶e) Mitche (1991, Table C3). US data. WGernany 27.S II D(p/Y-p) Bishop, Fomiby, Smith (1991. Tabice 3 & 4). US data. Groece 39.9 86 D(hIYh); txabe popul. only Livada(1°,°,Tabl 1) Ireand 34.6 t7 D(htY-h) Calculted from Imeand Central StLtistical Office (199). kaly 31.3 90 D(h/Y) Btndolini (1992, Table B12) Japan 35.0 a5 D(hfYh) Oshima (1991, Figures I & 2) Netheranda 32.1 13 D(pfY*p) Bishop, Fornby, Smith (1991. Tabke 3 & 4). LIS data. New Zealand 30.0 S5-S6 D(p/Yp,e) Saunden, Stott. Hobbes (1991, Table 5, p175). US data. Norway 26.9 79 D(p-Y'p) Bidh"p, Formby, Snuth (1991, Tables 3 & 4). US data. Portugal 38.1 73-74 D(h/Yh) Calculated from Portugal lntituto Necional de Esatistica (1977, p 16) Spain 31.5 aS D(hMYh) Calculated from Spain biuituto Nacionsl de Etatistica (1919, p.3t0) Sweden 22.9 Sl D(pIYp) Bishop, Formby, Smith (1I99, Tables 3 & 4). US data. Switzerand 35.5 32 D(pNYp) Bishop, Formby, Smith (1991, Tables 3 & 4). US data. Turkey 43.3 37 D(hY*b) Cakulated from Turkey tatistical yearbook 1990 (pp. 206-7) frnt houaehold budget survey 1987. United KinSdom 2k.1 79 D(p/Yp) Bishop, Formby, Smith, (1,91, Tables 3 & 4). US data. United States 34.4 79 D(pNYp) Bishop, Formby, Smith, (1991,Tabks 3 &4). LIS data. Eastem E pe . 56 Gini Year Coaponents Sources COadrMy coefficiert Idkgapt 21.7 9 ID(pY*p) Cakulated from houwehold budget survey 1919. Fotmtr 19.5 aS D(pYIp) Calculated fram Czechoslovakia Flderal Statistic.l Office (1989); houwehold budgets. Czecheblovakia Hunpgry 23.1 39 D(pNYp) Calculated from household survey 1938. Poabnd 26.0 19 D(p/Y p) Calculated trom Poland Central Statistical Office (1990); houwehold budgets. Roamnis 25.7 91 D(p/Y-p) ole (1992, p.25) Former 37.9 S9 D(pfY*p) Calculated fran Yugoalevis Federal Statiatical Ofrice (1990); household budgets. Yugoslavi Former USSR 2S.3 90 D(pNYp) Cakulated from household survey 1990. EGermny 19.3 89 D(pfY*p,e) Hauter, Mueller, Wagner (no date, p.9) Algeria \ 39.9 S9 Ahimd (1992) Egypt 43.0 74-75 D()/YT) Hanen (1992, p.221) Gabon 63.0 77 D(pyp) RLo (1992) Ghana 36.7 9849 D(p/F4) Chen, Datt, Ravalliott (1993) Cage d'lvoire 54.0 35 D(pNYp.e) Kozel (1990) Kenya 57.3 1-U3 D(h/Yh) Cben, Datn, Revallion (1993) Madagascar 41.9 11 0 DOM) Pryor (1990, p.26) Morocco 53.3 So D(pfYp) Bourguignon, Morritson (1989, p. 167) Nigeria 60.0 73-74 Dfh/h) Jamal (1911) Senegl S .3 70 DQsYh) Leceillon, Paulkert, Morrison, Gcrnidis (1984) Sierra Leone 49.0 75-76 D(Ply) 1LO (1992) South Aftica 57.0 s0 Devereux (1983, p.73) Swaziland S7.0 74 D(plyp) ILO (1992) Tanzre 59.0 as DQIEr) Lins Ferreire (personal communication) Zanbia 57.0 72-73 DQVyh) Cakluated frim Fry (1979, p.92) 2inihabus 50.1 70 D(h/yb) MD (1992) 57 | Gini Year Components Sources Cc"*y coedficient Axis Bangladeim 35.0 U3 D(hfYh) Oshinu (1991) Chino 33.2 So D(pIY'p) Rcnci (1992) Cyr-" 35.7 4-5 D(hWYb) Caculated from Chimtodoulou (1992, p.225) Hoe o 4S.5 Sl D(hb) Orbina (1991) bisal 33.3 79 D(p/Yp.c) O'Higins, Schnuus, Stephcnson (1989) India 40.0 75-76 Dowling (1984. p.l5) Indonoa 51.0 77 D(w/Yw) Rao (1939, p-59) isa 42.9 84 Dth?F") Bebdad (1929. p.327) Jordan 39.7 *6 DO&lYb) Shsaa (1990. p.67) S. Ko^M>" 35.7 S2 DWl') Cboo(1991.p.S) Malaysia 43.4 b9 D(pYp) Chaen, Dan, RavalCioa (1993) Pakimun 38.3 S4 D( bM) Ahmd and Lulow (19U3, p.23) Philippines 45.5 87 D/Yh) Otmina (1991; Figures I & 2) Singapon 41.0 87-8 D(bM) Rao (1990. p.147) S ilanka 43.0 I5 D(bNh) Odhina (1991; Figures I A 2) Taiwarn 32.5 37 DQhIYa) Orhima (1991; Figures I & 2) Thailand 47.8 SS-t9 D(hYh) Bh ngmkapat (1990, p 166) Argentins 47.6 89 D(p/Yp) Pascharopou!s et dI. (1992, annex 3) Bahamas 42.5 S9 D(hlEh); mral only Calculated from the Commonwealth of the Bahama (1992. p.102). Bolivia 52.5 39 D(p/Yp) Pscharnpouloaset *I. (1992, annex 3) Brazil 63.3 X9 D(pfYp) Pesacb apoulaet al*. (1992, annex 3) Chile 4S.2 S7 D(pYp) Caculaed fn lbindl, Budinich. Irnnzaval (1939, pp. 47-9) Cdonba1S.6 S7 DQWa) Aamir (19S4, p. 266) Co"e Ric 46.0 89 D(pfYp) Psachainpouloe al . (I992, annex 3) 58 Gini Year Comnponents Source Country coefficient Cuba 26.0 7S D(pNYp) Rodrigues (1989, p.218) Ecuador 445 S7 D(p/w) Psacharopouloset al. (1992, anriex 3) Guatenla 59.5 89 D(pNYp) Puacharmpouio et al. (1992, annex 3) Hondur s 59.1 S9 D(pNYp) Psachsropouloset al. (1992, annex 3) Janaice 44.5 75 D(h/Yh) Boyd (198S, p. I00) Mexico 50.6 U4 D(pNYp) Pacharopouloset al. (1992, annex 3) Panama 56.5 S9 D(p/Yp) Psacharopoulos et al. (1992, annex 3) Penu 57.0 81 D(hIYh) BerFy (1989, p.200) Uruguay 42 4 89 D(p/Yp) Psacharpouloat ami. (1992, annex 3) Venezuea US44. 89 D(pNYp) Psacharopouloses al. (1992, annex 3) Defin ion ofcoumponea.. Diabristion of (recipientsAype of income per recipien) where recipentas are p-peronsor h-houweholds. and incomne is Ygroa incofe. Y-=dispouble income and e denote equivslized income. Thus D(p/Yp) indicates dat the Gini coelliciert is cakulated from the distribution of perona ranked by their per capita gfios incone; or D(hIYh) denotes distribution ofhouseholds according to total honawhold dipoamble income. For Indonesia, the recipients are only wage-rnenr and incorme i wage income (w) only; for Ecuador, incone is approximated by labor income (w), for the Bahamas, Ghana, Tanzania and Iran, incomne is approxigted by expenditurs (E). REFERENCES Abnad, EJutisnam (1992), 'Poveny, Iequality and Public Poliky in Tranaitioo Econonics', Public Finance, Supplement: Public Finance r at World of Trnsdioe, 47196 Ahrnad, Ehtisham nd S. Ludlow (19U8), 'On Change& in Inequality in Patiean: 1979-84', The Developenrt ResearCh Program, London Scbool of Economica, June, p. 23 Ahantir, Ocr (I94), 'Poverty, Incom Distribution and Chi.d Welfare in Latin Amerka: A Comparison of Pre- and Po t-Recestaon Daw', Wor D,ewlopnen4 12:3. The Co_nIonwealth of the Bahansa, Ssasdcal Absn'c Repou, Departnedt of Statisicx, 1992, p. 102. Bebdad, Sobrab (1989), 'Winners and Lwens of the kanian Revolution: A Study in Income Distibution, IntenadoroJounWJ of MidALe Eat Sidisa, August. Berry, A. (19S9), 'Incoe Disributrion and Foreign Trde: The Cae of Perm', Externl Trade and Income Distribution, OECD, Paris, p. 200. Bhognukapat Tecan (1990) 'lrAmne Distributio in a Rapidly Growing Economy: Th Case of Thailand', 9ngapor Eeo _onk ReWne, April, p. 166. BiW p, lde A., John Formby ad W. Jones Smith (1991), 'hierational Comparison of loea Inequality: Teas for Lorenz Dominance acrts Nie Cauntrie', Ecos'ossa, 5S, pp. 461-7. Bourguignon, Franoia and Chrtian Morriaao (19S9), 'Incone Distribution and Forcign Trade: The Cas of Moncco', in P. Bourguigon ad Ch. Motisno (eb), riental Trade and Incoe Dkndhde, Paris: OECD. 59 Boyd, Detek (19U), E&ono.c Mmigemen, Income Ddstibuuon mwd Poway in Jamuaca, New York:Prweger, p. 100. Branlni. Andita (1992), 'A Description and an Aauasnet of the Ssmple Surveys on the Personal Distibution of Incones in buy'. drimeo, August. Cben, Shahnus, OGurv D3 and Mtis RavalLo (1993O , 'Ir Poventy Inceasing in the Developins Wodd?', tnimeo, April. Cboo, Hankchug (1993), 'A Comarimonof Incoan Disribution in Japan, Koea and Taiwan in T. Moguchi (ed.), Making Ee rmos More Eidena mad Morc Eqadrbk: Fcotn Dcernedning Income Diaubuiu, Econnie reseach sefies No. 29, Tch btitue of Econonaic Rcarch, Hitotsbaysabi University, KunitAcha, Tokyo and Oxfxd: Kinokuniya Conpapny Lid. and Oxford Univerity PrM. Chaidodoak D., (1992), Ide A. C)pr Mk,ue. The Labovs of.n, Embil MWi-Ec,nomI. University of Miaoa, Minnso Mediteunmn and Eas European Studies, Vol. 2, p.225. Czacbolwakis Fedenl Ssieal Offic (1990), M*bvcena 1991 .dll, Paue: Fedeal Statisil Office. D nevex, S_ag (3913), -SqIl Ahriean lcome Diuibuon, 190040-, SALDRU Woring Paper No. 51, University of Cape Town, Agusd, p. 73. Dowling, EM. (9194). Dietsih.sto. and Poer iy Sdected A"n Coustrie', The Aun Dvedopne Bank, Manila, Noveuber, p. 15. Fr, ham (I7, Nw&aywu ded Nome Distri n, hI At A)11cm Economy, Crooen Hedm, Londan, p.92. 11mi Eer I., Em Dadhice mod _ w.e nam vl (19), GiC Soidl Fje,cve. Saio de Chile: Ofeina de Plaificaeaon Nacioal and Univeraidad de Chile, Facuad de Cimeia Economicas y Ai-dy_. Haes., D"d (1992), Erp mW T-rky: Mm P.Ua Ecnmy qfEEqdry mEd GamA, Wadkingo D.C. and Oxford UK: Word Bank and Oxfd Univ y PNa. Hasse, Richard and KIU Mul1 (no de), 'Ie in East and West Geany o h Eve of Union: Some Rchs Bad ond he Gma Socioeco c Pael'. n Iw (192), hIuUrkme GA stud f; ca D bhRep ofSeenarenfarm C4mr,ies. Addis Ababs: DO. balktd Cerd Statidl O0In (1969), BemeA _ Audgeu Smwy 1967, vol. 1, Ccrl Statilea Office: Dublin. Ja)a, Vai (I13), 'Rural-Usbam Gap _td haqnaal is Nieri', RD. Addia Ahaba, Working Papas, Selsenber. Kind, VWe (31990), -M Cn yoabion ud Duitrlnidoof 1cm in Came divoie, Wanbing, D.C.: WoU Dank, LSMS Woking pa No. 68, Mamc. Lcaillon Jaee, Fdeli hlkat, Chri_m Miun. Disai Gertidi (1934), Ircm.e Diu,1budde mad Econmic Deelopmea. GenevawLO. Livbd,W Alemxan (3993), 'heem nequality Gme: A Statistical and Econoaeti Analyasa', Qr,r4d Dultedn ofFEcwdoelcz med SaaLUICS, 53:1. pp. 69-82. Machell, Deborah (99'3), Iwrw Trmifv hI Ten Weav SW"s, Aldersdt, Hongkong. Avebury. O'Higgim, Micbael, Guter Sclmua and Geoffly Stepbmon (1989), 'Income Diribution nd Redistribution: A Microdata Aalysis for Seven Counrie?, h.Iv ojf lNCt dmd WellA, p. 107. Oaima, Hanry (1991), 'Kuoets' Cve and Aaa Income Ditrlution' in T. Mioguebi (ed). Makng Econanal More deI mod Moem Egd:abk Feona Desenrankng Income Ddsibuzdon, Econic rmearcb senmae No. 29, The Inst of Econovc Reanreb, Hitoubabu University, Kunitachi, Tokyo, Kikuniya Con_a y Ltd. and Oxford United Pta. Pdland Central Statistical Office (1990). BCUd Gospodov DomowycA w 1989 RbOA, Warsaw: Ceoul Statiicl Office. Pryor, Frderic (1990), 'Chenges in Ince Di dribution in Poor Agiculural Nation: Mlawi and Malagaacar', Ewanc Devlopm old Oanad Chmge, October, p 26. 60 Peacharopouloa ct al. (1992). Powny and Income Diswibudon in LIadn Amedca: The Srory of the 19&b, Wassington, DC.: World Bank Latin American and the CaribbeanTechnmcal Dqsfametv, Rcgiontal sadics pnoam tcport No.27, Deccmber. Rao, tUonoji (19S9) 'Income lne4uality and Pvetly in East Asia', Indian Economuc Journal, Oct-Dec., p.59. Rao, Bhoaoji (1990). 'Inome Disribution in Singaporm Trends and Issucs', 3in*aporr E&aoosk Rei*w, April. p. 147. Renwci, Zbco (1992), 'Thtec Features of thc Distribution of Incomc During Trnnition to Reform', miurco. Ringen, Stcin (1991). 'Houwlch, Standard of Living, nd Inequality', ReWew of income amd Weakl, 37:1, Mamh. Rodrigucs, Jose Laui (1919), 'LA Eradication de la Pubeza y Is Djisrbution dcl Ingrso en Cuba', EcocnM4 de America Lamna, 1-19:21 1. Root, Richard (1992), 'ldeitifying Needs for Social Prtcwtion in Ronania: A Contrined Enirical Anclyai,', minnno, April. Saunders, Pctcr, Hekn Stoet and Garry Hobbcs (1991), 'Incorn Inequality in Australia and Ncw Zealnd', Reew of Income and Wealth, 37 1, March, pp. 63-79. Sha'ban. Radva Ali (1990), 'Economic Inequality in Jordan, 1973-19S6' in Katnel Abu Jabcr, Maahcs Buhbe and Mahutmnad Smadi (cdi), bome Dribudiout it Jordan, Boulder, San Francisco: Wrttviev Special Studier nn the Middlc Eatt. Spain Initio Nacionnl de Etatistica (199), Ecesasta C4n w ide Pesupuesios Familiares. Ano 19S8: Meiodologla y Resr uadoa, Madrid. Tsebou, Pama (1992), 'Ineqaity and Welfare in EEC Countries', RusdLen of Economic Researrh, No. I Vakenduc, Christian (19S7), 'Les revcnus disponibles dca rnlnages: dcazriptio ct analyse de cedain types d'inigalites', Bladke de Doc_wevatio, Brucsbls: Minisrm des Fm nces, Yugosavia Federal Office of Statistice (1990). Ankera o pouvInji domainsrava i 199: iRspolo5va i i porrbljena sredirva: Pdoed po aanu domalinw, Statistical Bulletin No.17SS, Belgrade: Federal Office of Statistics. 61 Anex Tabl e Iiwome data Purchuming power parity (PPP) aI international prices GDP per capita for 1988 (sonmetimes for 1987) arm obtained from Summtce and Hcaton (1991). The exceptions are the data for Bulgaria, Czcchoslovakia,Romania, the Soviet Union, East Gcrrrany and Cuba which atc obtained from Marer ct &l (1992). For all the cmuntriesexcept Cuba, the data reter to 1987 Fr Cuba, to 19#8 Sirsc theme sources are widely available, the data are not reproduced herc. Summerm, Robert and Alan Hcdton (1991), `Tbe Penn World Table (Mark 5): An Expanded Set of International Comparisoa', Quanrey Jouna f Economics, vol. 106, No. 2, May. p 327 Marer, Paul, larbo Arvay. John O'Connor, Mortin Schrenl and Daniel Swanson (1992), HLitorically PlanneJ Economics: A Guide so the Data, Washington, DC.: Woild Bank 62 Policy Research Working Paper Series Contact Title Author Date for paper WPS1221 Does Research and Development Narcy Birdsall November 1993 S. Rajan Contribute to Economic Growlh Changyong Rhee 33747 in Developing Countries? WPS1222 Trade Reform in Ten Sub-Saharan Faezeh Foroutan November 1993 S. Fallon Countries: Achievements and Failures 38009 WPS1223 How Robust Is a Poverty Profile? Marlin Ravallion November 1993 P. Cook Benu Bidani 33902 WPS1224 Devaluation in Low-Inflation Miguel A. Kiguel November 1993 R. Luz Economies Nita Ghei 39059 WPS1225 Intra-Sub-Saharan African Trade: Faezeh Foroutan November 1993 S. Fallon Is It Too Little'e Lant Pritchett 38009 WPS1226 Forecasting Volatility in Commodity Kenneth F. Kroner November 1993 F. Hatab Markets Devin P. Kneafsey 35835 Stiin Claessens WPS1227 Designing Water Institutions: Marie Leigh Livingston December 1993 C. Spooner Market Failures and Institutional 30464 Response WPS1228 Competition, Competition Policy, Bernard M. Hoekman December 1993 L. O'Connor and the GATT Petros C. Mavroidis 37009 WPS1229 The Structure, Regulation, and E. P. Davis December 1993 P. Infante Performance of Pension Funds in 37642 Nine Industrial Countries WPS1230 Unemployment in Mexico: Its Ana Revenga December 1993 R. Stephen Characteristics and Determinants Michelle Riboud 37040 WPS1231 Making a Market: Mass Privatization Nemat Shafik December 1993 A. Correa in the Czech and Slovak Republics 38549 WPS1232 Will GATT Enforcement Control J. Michael Fincer December 1993 N. Artis Antidumping? K. C. Fung 37947 WPS1233 Hedging Cotton Price Risk in Sudhakar Satyanarayan December 1993 D. Gustafson Francophone African Countries Elton Thigpen 33714 Panos Varangis WPS1234 Price Formation, Nominal Anchors, Andres Solimano December 1993 S. Florez and Stabilization Policies in Hungary: David E. Yuravlivker 39075 An Empirical Analysis Policy Research Working Paper Series Contact Title Author Date for paper WPS1235 Eastern Europe's Experience with Allredo Thorne December 1993 N. Jose Banking Reform: Is There a Role for 33688 Banks in the Transition? WPS1236 The Impact of Two-Tier Producer Maurice Schiff December 1993 S. Fallon and Consumer Food Pricing in India 38009 WPS1237 Bank Performance and the Impact Yavuz Boray December 1993 C. Lim of Financial Restructuring in a Hector Sierra 30864 Macroeconomic Framework: A New Application WPS1238 Kenya: Structural Adjustment in the Gurushri Swamy January 1994 V. Saldanha 1980s 35742 WPS1239 Principles of Regulatory Policy David E. M. Sappington January 1994 WOR Design 31393 WPS1240 Financing the Storm: Macroeconomic William Easterly January 1994 R. Martin Crisis in Russia, 1992-93 Paulo Vieira da Cunha 39026 WPS1241 Regulation, Institutions, and Pablo T. Spiller January 1994 B. Moore Commitment in the British Ingo Vogelsang 35261 Telecommunications Sector WPS1242 Firiancial Policies in Socialist Buris Pieskovic January 1994 M. Jandu Countries in Transition 33103 WPS1243 Are Institutional Investors an Punam Chuhan January 1994 R. Vo Important Source of Portfolio Investment 31047 in Emerging Markets? WPS1244 Difficulties of Transferring Risk- Edward J. Kane January 1994 P. Sintim-Aboagye Based Capital Requirements to 38526 Developing Countries WPS1245 The Adding-Up Problem: Strategies Takamasa Akiyama January 1994 A. Kim for Primary Commodity Exports Donald F. Larson 33715 in Sub-Saharan Africa WPS1246 Determinants of Cross-Country Branko Milanovic January 1994 R. Martin Income Inequality: An "Augmented" 39065 Kuznets' Hypothesis