Income inequality, welfare, and poverty : an illustration using Ukranian data

%.-A i9I( POLICY RESEARCH WORKING PAPER 1411 Income Inequality, Welfare, The standard of liMng in Ukraine increased significanty and Poverty in the 1 980s, and income inequality declined. But in U 1991-92 income inequality An Illustration Using Ukrainian Data and povertinceased again. partly because govemment Nanak Kakwani benefts went more to richer families than to those in need. T'he World Bank Policy Researc Department Transition Economics Division January~~~~~~~~~~~~~~~~~~~~~~~ 199  1ICi.Y(:Y RI-SIFAIitI-I WORKIN(G PA1'+R 1411 Summary findings Ukraine is now faced with economic crisis on an families dependent on government transfers fell by more unprecedented scale. The government has to follow than one-half. rigorous demand management policies, which entail e Income inequality declined in the 1980s, to rise lowering the population's standard of living. To design again in 1991-92. In particular, the family incomes of policies that protect the poorest and most vulnerable state and collective farm workers - rclative to industrial groups in the society, it is important to understand the workers - improved between 1980 and 1991. The nature of poverty and incomc inequality. increase in inequality that occurred in 1991-92 came Kakwani addresses the following questions: What is about, among other reasons, because govemment the extent of income incquality and is it increasing? How benefits tended to be redistributed to richer families, not can observed changes in inequality be cxplained? Is the those in need. burden of income tax evenly distributed across the Poverty in Ukraine declined over the period 1980-91, population? from 38 percent of the population to 9 percent. But in The Ukrainian data base is far from satisfactory, so 1992, 30 percent of the population was poor again, an Kakwani's findings are only tentative. Among them: alarming increase attributable both to a decline in real * The standard of living increased significantly in the per capita income and an increase in income inequality. late 1980s, then fell in the 1990s. Real per capita family Still, income inequality was lower in Ukraine than in income grew by an average 7 percent in 1989-90, then most other former republics of the Soviet Union. fell about 24 percent in 1991-92. Per capita income for This paper - a product of the Transition Economics Division, Policy Research Department - is part of a research project on income distribution and poverty during transition. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Grace Evans, room NI 1-041, extension 85783 (54 pages). January 1995. The Poley Research Working Paper Series disse,niiates the findngs of uwi m progress to cowurage the exchange of ideas about development issu. An objective of the scries is to get the findings out quicy. ven if tbe presentations ar kss than fully polishedS The papers carry tbe names of the auors and should be used and cited accordingly. The findigs, interpretations, and conclusions are the authors' own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its nmeber countrie Produced by the Policy Research Dissemination Center  TRANSmON ECONOMICS DIVISION RESEARCH PRoEC POLICY REsEARCH DEPARTMENT INCOME DISnhEBUTION DURiNG THE TRANSITON WORLD BANK INCOME NEQUALITY, WELFARE AND POVERTY: AN ILLUSTRATION USING UKRAIIAN DATA Nnak Kakwani The University of New South Wales Kensington, New South Wales Austulia This paper is a product of a research project on social expenditus in Eastan Europe undertaken by the Socialist Economs Reform Unit under the direction of Bnkmo Milanovic. These papers carry the name of ffie autos mnd are prminay versions for discussion only. The findings, ineprtations, and conclusions are the author's own. They should not be attributed to the World Bank. its Boad of Directors, its managemt. or any of its member countries. Comments and suggsdons are welcome Phone (202) 4736968, Fax (202) 522-1151  Contents 1. Introduction ....................................... . 1 2. Welfare and nequality Measures Used ......................... 2 3. Weffare by Income Components ............................... 4 4. Methodological Problems ................................... 7 5. Trends in Real Pr Capita Income in Ukraine .................... 12 6. Wages and Salaries in Ulkaine .............................. 18 7. Trends in inequality in Ukraine .......... ................... 21 S. Trends in Welfare inUkraine .......... .................... 30 9. Poverty inUkrine ..................................... 36 10. Comparing Ukraine with other Soviet Republics ........ 41 11. Concluding remarks . .................................... 49 Referces .................. ................... 52 1 would like to express my gmatitude to Dr Ihor Gordijew wihoutwhose help I could not have completed tis pap. Khannm Sabahmyam provided expert compational as_stce. My gratitue to Banko Milanovic who mad m-y substantial dtailed comments which proved extremly useful in revising the paper. I also bemefitted grealy from many usefu discussions I had with Michelie Ribout, Paolo Roberti and Subbauo. i  Tables Table Price Indices in Uklaine ............................... 10 Table 2 Distribution of Average Income by Income and Expenditure Components . 13 Table 3 Real Average Per Capita Income and Components ... .. ..... .. ... 14 Table 4 Annual Growth Rate, of Per Capita Real Income and Components ...... 15 Table 5 Aggregate Income Per Family Member by Types of Workers: At Current Prices - Ukmaine ............. .................... ....... 17 Table 6 Per Capita Income of State and Collective Farm Workers and Pensioner Families as a Percentage of Industrial Worker Families ... .... 17 Table 7 Average and Minimum Wages in Ukrine ..................... 19 Table 8 Average Wage Eaming and It's Inequality by Age and Education for Male and Female Workers and Employees: Ukamine 1989 ............. 20 Table 9 Inequality of Per Capita Aggregate Family Income: Ulaine 1980 to 1992 . 21 Table 10 Income Inequality by it's Components ...................... 24 Table 11 Income Inequality by Expenditure Components ................. 25 Table i2 Explaining Changes in Inequality in Terms of Income Components .... 28 Table 13 Redistributive Effects of Direct Taxes . . . 30 Table 14 Welfare in kllraine 1980-92 .................... ............ 30 Table 15 Welfare by Income and Expenditure Components ............ ... 32 Table 16 Explaining Changes in Welfare in Tenns of Income Components .... 34 Table 17 Explaining Changes in Welfare in Terms of Expenditure Components ... 35 Table 18 Povertyin Ukraine .................................. 37 Table 19 Explaining Changes in Poverty ........................... 40 Table20 InequalityinUSSR Republicsin 1990 ...................... 42 Table 21 Real Per Capita Aggregate Family Income and Welfare inUSSRRepublicsinl98 UkainianPrims ....................... 43 Table 22 Annual Growth Rates of Real Per Capita National Income forUSSR Republics ............ 45 Table23 Poverty inUSSR Republicsin 199 ....................... 46 Table 24 Ratio of Average Salis of State Farmers to Blue and White Collor WorklsbyUSSR'equblics ................................. 47 Table 25 Ratio of Average Salaries of Collctive Farmers to Blue and White Color WorkersbyUSSRRepublics ................................. 48 Table 26 Consumption of Foodstuffs by the Popatuim of Ukmine (Kilos Per Year) 51 ii  1. Introduction Ukraine with a population of 52 million people is the second largest country of the Former Soviet Union. It is weil endowed with fertile land and rich mineral resources. Huge coal and iron deposits have led to considerable industrial development especially in heavy industry. Ukraine was once known as the bread baskeL of the Russian Empire and later of the Soviet Union. With such wealth and a highly educated labour force, Ukraine has the potential to become a rich country. Unfortunatcly, it is now faced with economic crisis, the severity of which is on an unprecedented scale. Output has been filling continuously since 1990. According to a recent World Bank (1993) report, Ukraine's real net material product (NMP) registered declines of 3 percent and 11 percent, respectively in 1990 and 1991 and is esdmated to have contracted by another 15 percent in 1992. The country is faced with hyperinflation. The Lkrainian currency is depreciating at a phenomenal rate. The financial conditions of its enterprises have deteriorated badly. Faced with such crisis, it is hard for any government to secure a smooth tansition from a centrally planned to a market-oriented economy. To provide a stable macroeconomic enviromenti the government of Ukaine will need to follow rigorous demand mement policies. This entails lowering the standard of living of the population. Under these circumstances, it is important that the governent design policies which protect the poorest and most vulnerable groups of society. Ukraine has very extensrve socia protetion programs inheried from Soviet times. The current levels of government benefits seem unsustinable in the long run and the overaU system will undoubtedly have to undergo an overhaul. Spending on social programs, including subsidies and pensions was estimated to be 44 percent of GDP in 1992 (World Bank 1993). The shrining tax base due to resion and a much needed contracion in the fiscal deficit would necessitae a substantial reduction in welfare expenditures together with a change in their composition. It is, therefore, important to adopt the most efficient ways of directing expenditures towards various populadon subgroups. Obviously, the most efficient policy will be to target expenditures to the poorest secton of the population. To design such policies, it is important to understand the nature of poverty and income distribution. This study attempts to do exactly this for Ukaine for the first time. The study is cocerned with such questions as: What is the extent of the inequalty of income and in which way direction is it moving? How can the observed changes in inequality be explained? Is the burden of tax evenly distributed across the population? How do the vanous income and expenditure components affect total welfare? What are the effects of currnt government programs on total welfare? This study also attempts to measume the aggregate level of poverty in Ukraine and the extent to which it is affected by econonic growth and income redistribution. 1  Quality of data. however, is a serious Umitation that must be kep in mind. Ukoaine is still using the sampling procedure inherited from the old Soviet system. It has a number of flaws, the most important of which, for our study are: the concentration on the "average" types of housegolds with a consequent underrepresentation of the poorer and richer segments of the population; bias toward the employed vs. pensioners, students and other non-active members; and underreporting of "unofficial" sources of income. The paper is organized as follows: Section 2 discusses briefly the welfare and inequality measures used in the study. Section 3 develops techniques intended to analyze total welfare (and inequality) in terms of individual income and expenditure components. The welfare elasticities derived in this section provide a *link between total welfare and income or expenditure components. Section 4 offers a brief discussion of numerous methodological problems that arise in the measurement of economic welfare and inequality in Ukraine. Trends in real per capita income in Ukraine are discussed in Section 5 while Section 6 deals with trends in wages and salaries in Ukraine. Trends in inequality and welfare are analyzed in Sections 7 and 8, respectively. Section 9 is focussed on poverty in Ukraine and Section 10 provides a comparison of welfare and poverty in Ukraine with other Soviet republics. Finally, Section Il is devoted to some concluding remarks. 2. Welfare and Inequality Measures Used Among several measures of economic welfare available to economists, per capita national income is widely used. Although providing a comprehensive picture of the country's productive capacity the aggregate income measure has been subject to much criticism. National income as conventionally measured excludes many factors that contnbute to economic welfare while incorporating other factors which have an adverse effect on welfare. Since our concem is always with the welfare levels of individuals, it is reasonable to derive aggregate welfare from individual incomes. One such measure is the average per capita income of all individuals in a particular society. The main dmawback of this measure is that it does not take into account the inequality of income which always exists among individuals. This measure needs to be modified to make it sensitive to inequality in the distribution of income. Several inequality measures have been proposed in the literature. Among them, the Lorenz curve is widely used. It is defined as the relationship between the cumulative proporton of individuals and cumulative pmportion of income received when individuals art armanged in ascending order of their income. The curve is represented by a function L(p) which is interpreted as the fraction of total income received by the lowest pth fraction of income units. If the Lorenz curve for one distribution X lies everywhere above that for another distibu ion Y, then the distribution X may be said to be more equal than the distribution Y. However, the ranking provided by the curve is only partial - when two Lonz curves interest, neither distribution can be said to be more equal than the other. 2  Despite the fact that the Lorenz curve provides only a partial ranking of the distributions, it is a powerful device for judging the distributions from the welfare point of view. Since the Lorenz curve ranking assumes that the distributions have the samnt mean income, it can only be used to compare inequality in the distiibutions. If the distributions have different means which is usually the case, the Lorenz curve may fail to provide a welfare ranking of the distributions. Working independently on extensions of the Lorenz ordering Shorrocks (1983) and Kakwani (1984) arrived at a criterion which would rank any two distributions with different mean incomes. The new criterion given by L(,u,p) may be called the generalized Lorenz curve and is the product of the mean income A and the Lorenz curve L(p). This criterion of ranking has been justified from the welfare point of view in terms of seveml alternative classes of social welfare functions. Thus, it can be said that if the genemalized Lorenz curve for distribution X lies everywhere above the generalized Lorenz curve for another distribution Y, then distribution X is welfare superior to distribution Y. The Lorenz curve and the generalized Lorenz curve will be extensively used in the present study to analyze the distribution of income in Ukraine. Like the Lorenz curve, the generalized Lorenz curve also provides a partial ranking of distributions. When the two generalized Lorenz curves interest, then neither distribution could be said to be welfare superior to the other. To arrive at a complete welfare ranking of distributions, we must use a single meaure of welfe. Such a measure can be derived by giving different weights to individuals with different incomes. Suppose in a society there are n individuals who are aranged in ascending order of their incmes: xl s x2 S ... s x;, then a welfare measure may be defmed as a unique function of xl, x2, ..., xn. Sen (1974) considered the following welfare funcdon n W = E x,v, i-1 where v; is the weight given to the person with income xi. It is obvious that if vi = for n all i, then W is equal to average income of individuals. To make W sensitive to ineqal in the distribution, we must give higher (lower) weight to individuals with lower (higher) incomes. Sen (1974) proposed that v; (the weight given to the income of the ith person) should be proportional to the number of persons who are at least as well off as i.' From this poposition, Sen arrived at the welfare function: W = A, - G) t() where I is the mean income of the society and G is the Gini index which is a well-known 1 Kakwai (1980) has proposed an altemrtive weighting scheme in which v, is proportional to the total incom of persons who are at Lst as well off as individual i. 3  measure of income inequality. The Gini index is equal to one minus twice the area under the Lorenz curve. Although there exist several altemative welfare measures-, this study is confined to using W as a basis for analyzing welfare in Ukaine. We give two reasons for this choice. First, we regard W to be a reasonable welfare measure which takes into account both the size and distribution of income. Second, since income distribution data are available only in grouped form, W is the only welfare function which can be estimated most accurately from such data. 3. Welfare by Income Components Since the individual income is the sum of several income components, it will be useful to analyze total welfare (and inequality) in tenns of individual income components. The methodology for disaggregation of Gini index by income components was first developed by Rao (1969) and subsequently refined by Kakwani (1977, 1980), Field and Fei (1974), Fei, Ranis and Kao (1978), Fields (1979) and Pyatt, Chen and Fei (1980). In this section we extend this methodology to analyze the contribution of each income component to total welfare. Suppose there are k income components and g4 is the mean of the jth component. Then it is obvious that k -1 and the disaggregation of Gini index in terms of income components is written as (Kakwani 1980): G -E 'Lici ~~~~~~~~~~~~~~~~~~(3) where Cj is called the concention index of the jth income component. The concentration index C3 is smar to the Gini index except tth e ranking of individuals is by the total income and not the jth income component. As a result, the index can be negafive. The concenton index of an income component measures how evenly or unevenly that income component is distributed over the total individual income. If C; is greater (smaller) than G, it implies that the jth income component is distributed over the total income in favc- of richer (poorer) individuals. Combining (l) with (2) and (3) gives 2 See for instance Atkinson (1970) who derived a clas of welfare measurs based on a homothetic utility fi_n. 4  A W E - F>(1- C,) (4) Jul which shows how the total welfare can be decomposed in terms of individual income components; #^(l - Cj) being the contribution of the jth income component to the total welfare. To see how the change in the jth income component affects the total welfare, we compute the elasticity of W with respect to Sj p1(l Ft-C, (5) AI1-G) which implies that if 1j increases by 1 per cent; then the total welfare increases by rj per cent. It is instructive to write (5) as _(G-C,) (6) Is A,l-G) in which the first term may be called the income effect and the second term, the inquality effect. The inequality effect measures the gain or loss in welie as a result of inome isdistribution. If the increase in the jth income component favors the poor more than the rich, the inequality component will be positive, odterwise it will be negative. Ibis leads us to define a new progressivity index of the ith income component as the mtio of the inequality component to the income component:3 pi=(G-C) 7 (l (-G) A positive value of Pj implies the jth income component to be progressive and the negative value implies the jth component to be regressive. Thus, the magniude of Pj indicaes whether the icrease in the jth income component favors the poor or the rich. If the jth income component is distibuted in proportion to total income, Cj will be equal to G which gives Pj to be equal to zero. In this case, the effect of an increase in the jth income component favors neither the poor nor the rich. Pj can be used to devise an optimum tax - expenditure policy. It provides a quantitative basis for maxiiig the country's total welfare with minimum cost. Next, we wish to evaluate the effect of price changes on the total welfare. So assume 3 Note that i% in (5) is the infinite elasticity; it is derived under the assumption that income source j i inflitsimaily across al income recipients, or in other words, it is asmed that cses in income sources do not chng the nng of recipients. This elasticity is similar to the elasticities of the Gini index with espect to income compoet derived by Lemmmn and and Yitzhaki (19BS). 5  that there are k expenditure items whose prices are pi, P.,---,Pk and if q1, q2,...qk are the quantities consumed of the k expenditure items. then the total money income of an individual can be written as k k X E Spq E v, Vy(8) Jal Jp1 where vj = piqi is the consumption expenditure on the jth commodity. Let x = e(p, u) is the cost or expenditure function which is the minimum expenditure required by an individual to reach the utility level u at the price vector p.4 Suppose the prices have changed to p'. If the individual remained on the same level of welfare as before, the new expenditure of the individual will have to be x = e(p, u) (9) Thus, the individual should be given an additional income of (x'- x) in order that he remained at the same level of welfare as before. Suppose in the initial period the individual enjoyed a standard of living equivalent to x but because of changed in prices, the new standard of living of the individual will be given by y = x - (x - x)= 2x - x (10) Applying Taylor's theorem on (9) gives Akdpj e(p ,u) = e(p, u) + -vj(x) (11) i-I PJ whem d7J is the percentage change in the price of the jth commodity and use has been made Pi of the result: Be(p,u) = q which the well-known Shephard's Lemma in demand theory 8p (Deaton and Muellbauer (1980)).5 Substituting (11) into (10) gives 4 Note that the expenditure function e(p,u) will be equal to the actual expenditure x if it is assumed that evety individual maximizes his or her utility in every period which is the usual assumption made in the consumer theory. 5Note that the tenns of higher order of smallness in (11) have been ignored. 6  Y dX- p,VJ(X) (12) Ju1 Pi Applying Theorem 8.5 of Iakwani (1980) on (12) gives A dpi W'uX W- ' d -mI(I - C) (13) Jul Pij where W = lt(l - G) is the social welfare before the price change and W' is the social welfare after the price change, nm is the mean expenditure on the jth item and Cj is the conen index of the jth item. Equation (13) immediately gives the price elasticity of the aggregate welfare as mE- -c) (14) which indicates that if the price of the jth commodity increases by 1 percent, then the aggmegt welfae changes by ej percent. e will always be negative. Thus, the magnitude of E3j can be used to evaluate the efects of price changes on the aggregate welfare. 4. Mathodological Problem This section provides a brief discussion of many methodological problems which arise in the measurement of economic welfare and inequality in Uhmine. Our study is based on the data obtained from fte Family Budget Surveys (CBS), which had been carried out regularly in the Soviet Union since the 1950s. These surveys are the only source that provides mcome distibution data over time. But they have been subjected to ever criticisms (McAuley 1979, Shenfield 1983 and Ainson and Micklewright 1992); the main among them is that the surveys are not representative of the population. The sample covered the teritory of the former Soviet Union incompletely and unevenly (Cazes and Cacheux 1992). It is not known what proporDi of families were sampled in Utkrine. The families in the sample were mainly selected on the basis of their industrial affiiation of wage earnms. Consequently, social groups (such as students) which did not live in households with wUOing members were excluded from the survey. The old age pensioners (orginaly excluded) have been incluied in the survey since 1977. The prObabi;Ly of a household being selected in the survey increased monotonical1y with the number of wage earners in the households. This wVIl clearly bias the results on economic welfare and inequality. 7  In this paper we have used per capita household income as a measure of household economic welfare.' The income concept used is the personal income which includes [McAuley (1979. pp.9-12)] thr, following: Money earnings from employment (a) in the state sector (excluding holiday pay) (b) in collective fanrs. Transfer payments (a) pensions (b) other social security payments (c) stipends. Receipts from the rmancial system (a) interest from saving deposits (b) lottery prizes Other receipts (a) the value of agricultural production for own consumption (b) receipts in kind from collective farms (c) other income from rnvate econonic activity. The consumption of home-grown products is evaluated at state prices. In the case of many products, especially meat and frit, this may understate the true value of the product to the household. The imputed rent of owner-occupied housing is not included. Since rents charged by the state are very low, the distortionary effect of this omission should not be too great. When the index of household welfare is constructed, the next step is to detrmine the welfare of the individuals in the households. In this paper individual welfare was derived by assing every individual in a household a value equal to the per capita income for that household (Kakwani 1986). If there are severe intrahousehold inequities in the distibuin of food and non-food items, poverty and inequality will both be undeimatd. Ihis problem could not be corrected because of non-availabilty of information concerning the intra-household distribution of resources. But intrahousehold inequality may not be bad in Ukraine. Ukrainians 6A better measure of household welfare will of course be the per equivalent adult income which corrects for the differing needs of adults and children. But this measure could not be employed because the FBS data were available only in grouped form (the groups formed on the basis of per capita household expenditure). We could have remedied this only by assuming that the rankdng of households by per capita income is the same as that by income per equivalent adult. This assumption which is unlikely to hold will result in more serious estimation errors. 8  are more family orientated than West-Europeans so they tend to look after tfieir children and aged parents better than in Anglo-Saxon societies. However, because of male dominated society, women tend to perform more domestic work than in Western societies which may contribute to inequality in welfare within households. This aspect of welfare is not the focus of the present paper. Taxes, fees and fines paid by individuals are not subtracted from personal income. This may lead to an over-estimation of economic welfare. Since data on the amount of taxes, fees and fines paid by individuals are avaiable, an attempt is made to quantify their effect on the total individual welfare in Section 6. To compare welfare and poverty across different time periods, one needs to adjust the distributions given in current pnces for pnce changes over time. We used the official price indices for Ukraine as presented in Table 1. The only official price index available until 1990 is the retail price index. This index has severe limitations, the main among them being that it is constrcted on the basis of official list prices in the stae retail outlets. It ignores the prices charged in colectve farm markets and other secondary markets. A new price indexcalled consumer price index (CPI) was introduced in January 1991. This is a superior index based on a wide range of prices actually charged. As can be seen from Table 1, prices in Ukaine have been increasing at a phenomenal rate since January 1992. It seems the country is on the brink of hyperinflation. The government has lost control over the budget. The currency has been depreciating at an accelerating rate. This clearly has severe implications for the standard of living. If wages lag behind price inflation, the standard of living for a large majority of people will certainly decline. Measurement of economic welfare is problematic under the unstable situation of hypernflation. Relative prices will be changing constantly rsulting in severe disotions. Since in the Family Budget Surveys, households are not interviewed at the same time, so if there is a high rate of inflation, the distribution of nominal household income which these surveys provide will differ from that of real household income. Needless to say, it is the real income of households that is relevant for the purpose of analyzing the inequality of income. Kakwani (1987) has demonstrated that the inequality of nominal income from these surveys tends to overestimate the inequality of real household income (income adjusted for inflation occurrng dung the survey period). The FBS data are available in grouped form giving only the percentage of people in each group. This may be supplemented by information on overal mean income. The mean income of each income range were not available. Esdmates of inequality from such data will clearly be biased because of loss of infonnation due to grouping. Clearly, if the number of groups is small, the bias in estimates can be quite large. To estimate inequality from such data, one needs to employ some interpolation device. A commonly used procedure is to fit a density function to the entire income range and then compute inequality mcasures from the parameters of the fitted function. The difficulty with this approach is that there exists no single function which fits the entre income range. 9  Table I Price Indices In Ukraine Year or Month Overall | Fo od Clothing and Rent, Water, HouIehld | Medical Caf Tranedt RRooeation | Personal Cue Index Foowar Fuel and Good and Camm Education and Effoct Power _ It:u Ratin Pie* Index frfm 19W0 to 19O and Cuneumer Mae Index 1991 and 1992 1980 100 - - - r 1985 104 - - . - - r 1986 106 - - - 1987 107 - - | 1988 107 Ig89 109 . . . 1990 114 . . . 1991 213.6 - . 1992 3232.5 ' . - Coneumer Mae Indices OctoberI 1990=100 Doe 1990 100 100 | 100 100 100 100 100 100 100 Jan 91 103.7 101.9 105.7 100. 104.5 100.2 100.9 109.1 106.1 Feb 109.6 103.3 115.2 101.6 119.4 100 101.7 119.9 113.4 March 117.7 109 121.3 123.7 157.8 99.8 103.4 133 131.8 April 195.9 199 225.4 150.9 229.8 100.3 173.4 155.9 186.8 May 198.3 194.2 236.2 153.5 249.3 99.1 176.5 162 203.6 Juno 199.9 191.9 241.4 161.6 259.8 100.3 177.2 167.2 206.2 July 200.9 190 243.8 162.4 267.6 101.3 180.2 171.5 207.6 Aug 201.5 186.2 250.9 166.3 275.4 102.3 180.4 174.1 198.9  Table 1 (continued) Sep 206.5 188.4 259.2 167.3 284.5 108 184 178.1 213.4 Oat 219.5 197.1 283 170.6 305.8 115.7 186.4 192.9 239.9 Nov 235.7 209.1 310.5 175.2 341 132.2 188.3 209.3 244.7 Doc 91 280.9 228.1 354 188.3 373.4 132.3 192.1 243 262.8 Jan 92 919.4 927.5 966.4 773.7 1411.1 369.5 594 895.7 646.8 Fab 1129 994.3 1256.3 1755.5 1920.5 401.2 696.2 1223.5 730.9 Mafoh 1246.4 1086.8 1398.3 1896.9 2225.9 483 914.1 1290.8 896.1 April 1417.2 1211.9 1542.3 2057.1 2337.2 487.8 1038.4 1774.9 1187.3 May 1740.3 1379 1648.7 2275.2 2887.8 533.7 1694.7 3187.7 1269.2 ';ane 2001.3 1795.S 1813.6 2341.2 2870.8 868.1 2060.8 3583 1387.2 July 2551.7 2449.1 2111 3834.9 3269.6 780.7 2388.5 4249.4 1789.5 Aug 2947.2 2838.5 2377 4291.3 3890.8 773.7 2744.4 6056.8 1970.2 Sep 3386.3 3233.1 2728.8 4980.7 4264.3 864.2 3356.4 5936.7 2179 Oct 4246.4 4041.4 3574.7 5119.4 5262.1 1292 5843.5 6756 2834.9 Nov 5206.1 4954.8 4557.7 5308.8 6525 1357.9 8397.1 7681.8 2922.8 Deoc 92 7246.9 6428.4 6193.9 6503.3 9924.5 1447.5 14812.5 10001.4 3866.9 Jan 1993 14298.1 15693.3 10009.3 18540.9 16851.8 3219.2 24b29.5 11781.8 8511 Feb 18344.5 18706.4 14093.1 24066.1 23457.7 4548.7 32894.1 16553.1 12630.3 Maroh 22105.1 20951.2 18039.2 27363.2 32254.3 6172.6 41808.4 21370.1 17076.2 Aprl 93 26238.8 24324.3 21611 31495 36898.9 7505.9 53389.3 26114.3 18903.4 ~~. _ _. m  In the present paper we have used a general interpolation device proposed by Kakwani (1980). This method utilizes, within each income range, a separate continuously differentiable function which exactly fits to the data points. The inequality and poverty measures are then computed by liniing this function. Kakwani's approach requires information on the mean incomes in each income range which is not available in the present case. A modification of the approach is used in this paper. S. Trends in Real Per Capita Income in Ukraine Tables 2, 3 and 4 present information on the real per capita income in Ukraine based on data from the family budget surveys. The nominal incomes were converted to the real incomes by means of the price indices given in Table 1. The real per capita incomes given in the tables are at 1980 prices. Table 4 giving the growth rates of per capita real inoome shows that the average standard of living in Ukraine increased quite significantly in the late 80s. The 1989-90 period registered a growth rate of 7.4 per cent, but this was eroded in the subsequent periods from 1990 to 1992. In 1990-91, the real value of salary income declined by 7.54 percent and that of collective farm income went down by a massive 30.67 percent. It is intersting to note that government cash benefits to families increased by 45.79 percent. Since government cash benefits are generally concentrated among poorer families, it is to be expected that the distribution of per capita family income would have become more equal dunng the 1990-91 period. Per capita family income declined by 23.56 percent during the 1991-92 period. This represents a drastic reduction in the average standard of living of the Ukrainian people. Salary income declined by a massive 31.44 percent. Families dependent largely on government transfers suffered the greatest decline in their incomes.7 Their per capita income fell by 58.33 percent. The substantial income gain made by transfer recipients in the 1990-91 period was completely eroded in the subsequent period. The collective farm income increased, however, by 49.56 percent. 'Note that many families receive wages, smalholding income etc. in addition to government transfers and, therefore, their economic situation may have changed somewhat differently than what is shown by changes in govemment transfers. 12  Table 2 Distribution of Average Income by Income and Expenditure Components Components 11980 | 1985 1989 11990 11991 1992 Income Components Salary Income 59.6 55 56.9 57.3 51.8 46.4 Collective Farm Income 9.9 10.7 10.4 10.2 6.9 13.5 Govt Cash Benefits 13.6 14.9 12.8 13.4 19.1 10.4 Smaliholding Income 9.8 11.1 9.9 10.2 11.5 19.6 Other sources 7.1 8.3 10 8.9 10.6 10.2 Total Income 100 100 100 100 100 100 Expenditure Components l Food 39.9 36.6 32.7 32.8 38 38.3 Non food 29.9 29.6 29.6 31.4 32.3 31.9 Alcohol 2.8 2.6 2.5 2.7 2.1 2.1 Social Expenditure 8.8 8.3 8.8 8 6.2 3.7 Taxes 6.9 6.8 7.3 7.2 5.1 5.2 Other Expendtures 6.5 7.8 8.5 6.5 7.1 7.5 Savings 5.2 8.3 10.7 11.4 9.3 11.2 Total Expenditure 100 100 100 100 100.1 100 13  Table 3 Real Average Per Capita Income and Components Components 1980 1985 | 1989 I 1990 1991 1992 Income Components Salary Income 73.3 74.6 84.6 91.5 84.6 58 Collective Farm Income 12.2 14.5 15.5 16.3 11.3 16.9 Govt Cash Benefits 16.7 20.2 19 21.4 31.2 13 Smaliholding Income 12.1 15.1 14.7 16.3 18.8 24.5 Other Sources 8.7 11.3 14.9 14.2 17.3 12.7 Total Income 123 135.6 148.6 159.6 163.4 124.9 Expenditure Components Food 49.1 49.6 48.6 52.3 62.1 47.8 Non Food 36.8 40.1 44 50.1 52.8 39.8 Alcohol 3.4 3.5 3.7 4.3 3.4 2.6 Social Expenditure 10.8 11.3 13.1 12.8 10.1 4.6 Taxes 8.5 9.2 10.8 11.5 8.3 6.5 Other Expenditures 8 10.6 12.6 10.4 11.6 9.4 Savings 6.4 11.3 15.9 18.2 15.2 14 Total Expenditure 123 135.6 148.6 159.6 163.4 124.9 14  Table 4 Annual Growth Rate of Per Capita Real Income and Components 1980to 1 1985to 1989to 1 1990to 1991tol Components 1985 1989 1990 1991 1992 Income Components l Salary Income 0.35 3.19 8.16 -7.54 -31.44 Collective Farm Income 3.51 1.68 5.16 -30.67 49.56 Govt Cash Benefits 3.88 -1.52 12.63 45.79 -58.33 Srrnllholding Income 4.53 -0.67 10.88 15.34 30.32 Other sources 5.37 7.16 -4.7 21.83 -26.59 Total Income 1.97 2.32 7.4 2.38 -23.56 ______________________ ___________ Expenditure Components Food 0.2 -0.51 7.61 18.74 -23.03 Non food 1.73 2.35 13.86 5.39 -24.62 Alcohol 0.58 1.4 16.22 -20.93 -23.53 Social Expenditure 0.91 3.76 -2.29 -21.09 -54.46 Taxes 1.6 4.09 6.48 -27.83 -21.69 Other Expenditures 5.79 4.42 -17.46 11.54 -18.97 Savings 12.04 8.91 14.47 -16.48 -7.89 Total Expenditure 1.97 2.32 7.4 2.38 -23.56 It is clear from the evidence that the income structure by source has changed quite substantially in the late 80s and the early 90s. The share of salary income has declined quite substantially from a level of 59.6 percent in 1980 to 46.4 percent in 1992 and that of farmers increased from 9.9 percent in 1980 to 13.5 percent in 1992. It is interesting to note that income from personal garden plots (small holding income) increased more or less monotonically between 1980 and 1992. The real value of this income doubled during this period. The share of this income stands at 19.4 percent in 1992 - which is the second highest after the share of wage and salary income. Most of the smallholding income comes from the output of small household plots. Until 1989, this output was valued at state prices which were lower than private market prices. Since the government introduced the new consumer price index in 1990, prices in collective farm markets or secondary markets have been used to evaluate the output of small household plots. This change in the valuation of output could be the main reason for a large increase in the share of smalUholding income. 15  Next, we look at the structure of expenditure components. Although the share of expenditure on food increased slightly between 1991 and 1992, real per capita expenditure on food declined quite substantially between these years. During a recessionary period, people generally try to maintain their consumption of food by cutting expenditure on other items of consumption. This does not seem to be happening in Ukraine. Food expenditure in real term declined almost as much as real non-food expenditure. This can probably be explained in terms of price changes for food and non-food items. The overall CPI increased at a faster rate than the food CPI during the 1991-92 period. This implies that the decline in the quantity of food might not be as high as indicated by the figures in Table 3. It is interesting to note that the real outlays on wsocial expenditure' has declined by a massive 54.46 percent. This is understandable. Expenditure on social activities is a luxury which people would not be able to maintain. Further, it can be seen from Table 1 that the collective price index for recreation and cultural activities was higher than that of the overall CPI for the entire period of 1992 which means that the decline in the real expenditure on cultural activities would have been even larger than indicated by the figures in Table 4. Table 2 shows that in 1992 Ukrainian families paid 5.2 percent of their income in direct taxes (including fees and fines) and at the same time received 10.4 percent of their incomes in the form of government transfers. This means that the Ukmainian government relies heavily on indirect taxes. The value-added tax (VAI) accounted for about 45 percent of general government revenues in 1992 whereas for the personal income tax this figure was only 10 percent. Excise and trade taxes and non-tax revenues accounted for 16 percent of the total tax revenues. Since indirect taxes are generally regressive, it seems that the overall tax system in Ukraine favors the rich rather than the poor. As noted earlier the share of collective farm income increased during the period between 1980 and 1992. This suggests that the relationship between the average eaniings of blue-white- collar workers and collective farmers may have changed over this period. Table 6 shows that this has indeed been the case. The ratio of average incomes of families of collective farm workers to that of industrial workers increased more or less monotonically over the period from 1980 to 1991. Since collective farm families have had much lower per capita income than the blue-white - collar families, the overall income distribution in Ukraine should have gradually become more equal between 1980 and 1991. Further, it is intesting to note that pensioner families of collective farmers also improved their reladve position quite substntally dunng the 1980-91 period. This will have a further equalizing effect on the overall income distribution. We shall return to this issue in the next section. 16  Table 5 Aggregate Income Per Family Member by Types of Workers: At Current Prices - Ukraine Family typos 1980 1985 1986 1987 1988 1989 1990 1991 ~~l~ - - I NNNNNN All families 123 141 145 148 149 | 162 182 349 Worker Families Industry 132 151 152 154 152 174 196 377 State farm 103 120 125 127 134 143 163 318 Collective farm 96 125 129 131 136 151 171 323 Pensioner Families Blue & White Collor 74 | _97 | 891 92] | |___ 106 I 116 [ 227 Collective farm 78 | 101 106 | 105 11S 132 J 155 33 Table 6 Per Capita Income of State and Collective Farm Workers and Pensioner Families as a Percentage of Industrial Worker Families Warker Families | Industry 100 100 100 100 100 100 100 State farm 78 79.5 82.2 82.5 82.7 82.2 83.2 84.4 Collective farm 72.7 82.8 84.9 85.1 84 86.8 87.2 85.7 Pensioner Families Blue & White Collor 56.1 64.2 58.6 59.7 | 60.9 59.2 60.2 Collective farm 59.1 66.9 [ 69.7 6.2 71 75.9 79.1 88.3  6. Wages and Salaries in Ukraine As noted in the previous section, real per capita family income fell substantially in 1992. There is evidence that the Ukrainian economy deteriorated further in 1993. Family Budget Survey results are not available to assess the standard of living since 1992. However, data on average and minimum wages for the beginning of 1993 were provided to us by the Ukrainian statistical department. This information is summarized in Table 7. It can be observed that the total average wage and the average wage of industrial workers fell by about 10 percent between 1991 and 1992. The deterioration in wages is quite striidng since 1992. In early 1993, real wages for all workers were about sixty percent of what they were in early 1992. The decline in wages among the industrial workers has been even greater. The drastic fall in real wages is due to the deliberate wage and income policy pursued by the Ukrainian govemnment. In an effort to control inflation, the government has suspended wage indexation and introduced tax-based wage controls on January 1, 1993. This led to a substantial reduction in real wages. Average nominal monthly wages increased by 16 percent in January and by only 9 percent in February 1993, whereas inflation rates during these months were 97.3 and 26.3 percent, respectively. These results tend to suggest that the standard of living in Ukraine is falling at a phenomenal rate. Will the govemment continue squeezing real wages? How much more do real wages have to fall before there is a slowing down of rie inflation? While it is not possible to answer these questions, it is useful to consider what is ikely to happen to the distribution of income and poverty in ULkraine. This is attempted in Sections 7, 8 and 9. In the mean time we consider the issue of male-female differentials in earnings. The distributions of earnings among men and women are summarized in Table 8. It is evident that women on average earn considerably less than men. This conclusion holds for all levels of education and age groups. The ratio of female-male earnings varies quite substantially with respect to education and age. The ratio is highest among workers with higher education. The disparity of earnings between male and female workers exists in Ukaine despite the fact that the Soviet Union had in its constitution a provision for sexual equality. Several reasons may be given for this disparity, but the most important ones are the levels of educafion, age, occupational segregation and hours of labor supply. Since the results presented in Table 8 tabe account of education and age, the remaining two reasons, viz, occupational segregation and hours of labor supply are most likely. These two factors are likely to be less important among workers with higher education. It is, therefore, to be expected that the ratio of female-male average earnings will be highest among these workers. 'Our finding that women are least discriminated against at high education level is an interesting one because the finding in other countries is often the opposite. In the West, difference between women and men is greater at higher levels of education than that for low-sldlled jobs. 18  Table 7 Average and Minimum Wages in Ukraine At Current Prices At Constant Prices Date Minimum Average Average Wage Minimum Average Average Wage Wage Wage Total Industry Wage Wage Total Industry 1990 247 - 0 216.7 - 1991 474 548.4 - 221.9 256.7 1992 - 6423.2 7885 - 198.7 243.9 Jan 91 80 - - 67.7 - April 91 145 - 64.9 - Oct 91 185 - - 74 - Dec 91 400 - - 134.5 - Jan 92 400 1655.9 2280.8 38.2 158 217.7 Feb 400 1847.4 2622.9 31.1 143.6 203.9 March 400 2068.6 3031.7 28.2 145.6 213.4 April 400 3068.1 4529.1 24.8 190 280.4 May 900 3572.4 4508.4 45.4 180.1 227.3 June 900 5464.4 5696.7 39.5 239.6 249.8 July 900 5379.7 6715.6 30.9 185 230.9 Aug 900 5705.9 7096.5 26.8 169.9 211.3 Sep 900 7689.8 8789.1 23.3 199.3 227.7 Oct 900 8715.9 10414 18.6 180.1 215.2 Nov 2300 11314.3 14436.3 38.8 190.7 243.3 Dec 92 2300 20072.5 25385.3 27.8 243 307.4 Jan 93 4600 15802 19180 28.2 97 117.7 Feb 4600 20192 23060 22 96.6 110.3 March 4600 24857 28491 18.3 98.7 113.1 June 6900 38169 - 19  Table 8 Average Wage Earning and It's Inequality by Age and Education for Male and Female Workers and Employees: Ukraine 1989 Male Employes Femdae Employees Ratio of Falde-Male Employees Chwractedstics Aveage Gini Avae Ginl Averag income Wage Index Wage Index Wge tnequdity Levels of Education Higher Education 236.124542 19.0183462 184.3897645 18.17798 78.1 95.6 Secondary Special 216.259565 20.9938104 144.6056877 19.41531 66.9 92.5 Secondary General 215.895647 22.5010022 143.6918695 22.29261 66 6 99.1 Incomplete Secondary 204.12568 23.9308589 140.6166197 24.30334 68.9 101.6 Primary Education 177.860896 23.4948659 129.3665318 24.12899 72.7 102.7 Age of Employees 16-24 176.715836 21.9304696 125.755102 20.0741 71.2 91.5 25-29 210.367413 21.08665 139.0529129 19.33895 66.1 91.7 30-39 229.616129 21.4136811 154.2772768 20.57662 67.2 96.1 40-49 230.484869 20.5815562 161.2882208 21.58169 70 104.9 > 50 205.076292 22.45246 149.5554106 23.52911 72.9 104.8  From the evidence given in Table 8, it is not possible to conclude that earnings disparity is less for women relative to that for men. This conclusion holds for workers up to the age of 39. The opposite conclusion emerges for the workers aged 40 or more. Similarly, the lower disparity for women is observed only among the workers with education levels equal to or higher than the secondary school. An interesting question that arises is whether earning differences between men and women are widening or narrowing during the recession. This aspect will not be pursued here due to lack of data. 7. Trends in inequality in Ukraine Table 9 presents the inequality measures for Ulraine calculated from the family budget surveys covering the period from 1980 to 1992. The table gives the quintile shares, Gini index and decile distribution ratio (DDR). The DDR is defined as the share of the bottom 40 percent in relation to the share of the top 20 percent. An increase in this measure will imply a reduction in income inequality. All these measures have been computed by assigning eadh individual in a household an income level equal to the per capita income of that household. Thus, these measures indicate inequality in individual income distribution. Table 9 Inequality of Per Capita Aggregate Family Income: Ukraine 1980 to 1992 Ouintles Yea_ 1980 1985 1989 1890 1991 1992 I1St 8.13 7.09 9.41 10.35 11.01 10.56 2nd 13.04 14.83 13.99 14.31 14.72 15.53 3rd 16.04 16.37 17.37 17.51 17.82 16.07 4th 19.83 20.51 21.33 22.83 23.99 23.69 5th 42.96 41.2 37.9 35 32.46 34.15 Gini Index 34.64 32.16 26.5 25.13 21.8 23.4 Decile Destribudion Ratio 49.28 53.2 61.74 70.46 79.27 76.4 21  The results are quite strikdng. Inequality declined monotonically between 1980 to 1991, but rose between 1991 to 1992. This is indicated by the Gini index as well as the DDR. The share of the first quintile increased monotonically between 1985 to 1991, then fell in 1992. The share of the top quintile declined monotonically between 1980 to 1991 and then rose in 1992. The Lorenz curves for the years 1985, 1989, 1990 and 1991 shift gradually upwards at all points. This implies that income distribution in Ukraine became gradually more equal between 1985 to 1991. Since the Lorenz curves for the years 1991 and 1992 intersect, we cannot unambiguously state that income inequality rose between 1991 and 1992 although this conclusion holds on the basis of the Gini index and the DDR. The question that arises is: why has inequality declined consistently during the 1980-1991 period? The evidence given in Table 6 provides a parta answer. The families of state and collective farm workers were generally poorer than those of industrial workers. The results in Table 6 suggest that the income gap between these families narrowed quite substandally between 1980 and 1990 which would have made the income distribution more equal. Further, it is interesting to note that the pensioner families of both blue and white collar workers and collective farmers also improved their relative incomes (compard to industrial workers) between 1980 and 1991. For instance, in 1980, the average per capita income of pensioner families of collective farm worlers was about 59 percent of that of industrial wrkers, but this percentage ineased to 88.3 percent by 1991. This would clearly have an eq-Jimzing effect on the distribution of income. We may also attempt to explaun changes in income inequality by means n observed changes in income components. The shares and distribution of income components air reinted in Table 10. The information in Table 10 could be compiled only for the years 1989, 1991 and 1992 because of non-availability of the appropriate data for other years. Results in the table show that the income structure has changed quite substandally in the years between 1989 to 1991 and 1991 to 1992. The percentage share of salay income deeased monotonically during the three year period. This may be due to the effect of the recession which occurzed during this period. It is interesting to note that the share of income from small holdings such as family plots incased quite substntially during this pedod. This is understandable. When people lose theirjobs due to a recession, they divert their efforts to small holdings to maintain their standard of living. The concentration index of an income component measures how evenly or unevenly that mcome component is distributed over the per capita total family income. It can be seen that the concentration index of wage and salary income is 32.8 in 1989, which is considerably higher thn the Gini index of total per capita fanmly income. This implies that income from wages and salary is unevenly distributed over the total income in favor of richer families. From quintile shares, it can be seen that the first quintle gets only 6.1 percent of total wage income whereas the top quintile gets more than 40 percent of total wage and salary income. The coective farm income is also unevenly distibuted in favor of richer households but its regmssivity is less than that of wage and salary income in the non-farm sector. 22  The concentration index of income from government transfas is -0.7 which implies that this income component is more or less distributed on a per capita basis with no focus o the poor.9 Although the overall effect of govemment transfers is to reduce the inequality of total income, the magnitude of the effect is small. The progressivity of government rnsfers can easily be increased by focusing them on the poor rather than distributing them on a per capita basis. It is unfortunate that the progressivity of government benefits in fact fell considerably in 1992. The value of the conentration index for government transfers changed from -0.5 in 1991 to 14.2 in 1992. From the quintile shares we note that the share of the goverment transfers going to the bottom 20% of individuals reduced dramatically in 1992 while the share of the top 20 percent of people increased from 20.87 percent in 1991 to 29.16 percent in 1992. During recession the government should target its scarce resources but in Ukraine exacty the opposite happened. The greater proportion of government transfers went to the richer fa_m in 1992. Since we do not have the breakdown of government transfers, it is not possible to explain why the govemment transfers became less progressive in 1992. The last column in Table 10 gives the % contribution of each income component to the total inequality. In 1989, the contribution of salary income of blue and white collar worker was 70.7 percent which fel substantally to only 62.1 percent in 1992. This shows that the changes in wage and salary income strucure led to a consderable reduction in total income inequality. In 1992, the share of smallholding income was 19.8 percent but it contributes only 16.6 percent to total inequality. It would seem then that any policy that increases the smaiholding income has an equalizing effect on the size distribution of total income. Table 11 presents inequality of total income by its expenditure components. The personal income tax paid by the families is shown as an item of expenditure. It can be seen that the share of xs to total family income declined from 7.3 percent in 1989 to only 5.1 perent in 1991. Smce the concentration index of taxes is considerably larger han the Gini index for towal income, taes can be said to be highly progressive. So the decline in the share of aes woud have the effect of incrsing the inequality of total income in the 1989 and 1991 period. In the subsequent period (from 1991 to 1992) the share of taxes remained more or less the same whereas their concentration increased from 32.1 percent in 1991 to 38.7 percent in 1992. This would have an effect of decreasing the inequality of total income in the 1991-92 period. 'The absence of targetting of govemment trnsfers is a common finding for sociaist counties, see for instance Mlanovic (1993). 23  Table 10 Income Inequality by It's Components lQuintile Shwars J Concen- % coantib. to Income -_ I %9 Shere tratlin Index total Components lit 2nd 3rd 4th 5th Inequality Year 1989 Salary Income 6.1 12.31 17.77 23.05 40.77 56.9 32.8 70.7 Collective Farm Income 10.07 13.16 15.53 20.02 41.22 10.4 28.5 11.2 Govt Cash Benefits 21.8 20.92 17.94 16.37 22.97 12.8 -0.7 -0.3 Smaliholding Income 10.64 14.91 17.54 20.04 36.87 9.9 23.8 8.9 Other sources 10.48 14.54 16.17 20.47 38.34 10 25.4 9.6 Total Income 9.41 13.99 17.37 21.33 37.9 100 26.4 100.1 Year 1991 Salary Income 7.51 13.43 18.11 25.39 35.56 51.8 28.5 67.7 Collective Farm Income 12.7 15.79 16.98 22.77 31.76 6.9 18.8 6 Govt Cash Benefits 21.65 19.13 18.62 19.73 20.87 19.1 .0.5 .0.4 Smaliholding Income 8.99 13.31 16.28 25.47 35.95 11.5 27.5 14.5 Other sources 10.05 13.83 17.19 24.03 34.9 10.6 25 12.2 Total Income 11.01 14.72 17.82 23.99 32.46 99.9 21.8 100 ____ ____ ____ ____ ___ ___ _ _______Year 1992 Salary Income 7.97 13.83 14.46 23.86 39.88 46.4 31.3 62.1 Collective Farm Income 13.69 18.31 18.65 23.15 26.2 13.5 12.4 7.2 GovtCashBenefits 14.4 17.37 17.6 21.47 29.16 10.4 14.2 6.3 Smallholding Income 11.41 16.36 16.93 24.35 30.95 19.6 19.8 16.6 Other sources 12.62 16.16 18.8 24.65 29.77 10.2 18 7.8 Total Income 10.56 | 15.53 r 16.07 | 23.69 r 34.15 1 100.1 | 23.4 100  Table 11 Income Inequality by Expenditure Components Clubitle shoes ~~~~~Cancan- I% contrib. Expenditure S 1 | 2nd | 3ft | 4* | C 1 % 8hare tiation to total can4wente lst 2nd 3rd 4th 5th k________ index Inequality . ~~~~~~~~~~~~~~~Year 1989 Food 1333_ _ 16.44 18.05 20.45 31.73 32.7 16.8 20.7 Non food 8.66 13.8 17.7 22.23 37.61 29.6 27.3 30.5 Alcohol 8.58 13.93 1 6.88 21.4 39.21 2.5 28.2 2.7 Social Expenditure 9.83 16.57 18.79 22.31 33.5 8.8 22.2 7.4 Taxes 5.48 11.41 17.13 22.92 43.06 7.3 35.5 9.8 Other Expenditures 7.09 11.61 15.24 19.6 46.46 8.5 35.7 11.5 Savings 3.89 9.3 15.24 20.95 50.62 10.7 43.3 17.5 Total Expenditure 9.41 13.99 17.37 21.33 37.9 100.1 26.5 100.1 Year 1991 Food 14.85 16.75 18.5 22.33 27.57 38 12.9 22.5 Non food 10.11 14.44 17.89 24.44 33.12 32.3 23.4 34.7 Alcohol 10.76 14.32 17.34 24.41 33.17 2.1 22.9 2.2 Social Expenditure 11.4 15.29 19.44 23.33 30.54 6.2 19.4 5.5 Taxes 6.5 12.37 17.15 25.73 38.25 5.1 32.1 7.5 Other Expenditures 9.57 12.23 15.08 24.67 38.45 7.1 29.2 9.5 Savings 1.83 10.24 16.29 28.04 43.6 9.3 42.4 18.1 Total Expenditure 11.01 14.72 17.82 23.99 32.46 100.1 21.8 100  Table 11 (contlnued) Year 1992 Food 14.85 16.75 18.5 22.33 27.57 38.3 14.1 23.2 Non food 10.11 14.44 17.89 24.44 33.12 31.9 24.2 33.1 Alcohol 10.76 14.32 17.34 24.41 33.17 2.1 22.5 2 Social Expenditure 11.4 15.29 19.44 23.33 30.54 3.7 20.7 3.3 Taxes 6.5 12.37 17.15 25.73 38.25 5.2 38.7 8.6 Other Expenditures 9.57 12.23 15.08 24.67 38.45 7.5 34.1 1 1 Savings 1.83 10.24 16.29 28.04 43.6 11.2 39.2 18.8 Total Expenditure 11.01 14.72 17.82 23.99 32.46 99.9 23.3 1 00  It may be enlightening to explain the change in the inequality of total income in terms of its income or expenditure components. This may easily be done by means of equation (3) which can also be written as k G = S w,,CI (15) J-1 where G is the Gini index of the total income, C; is the concentration index of the jth income or expenditure component and wj = gfIt is the share of the jth income or expenditure component. Suppose that over a period, income or expenditure shares change from wj to w; which may be accompanied by changes in their concentration indices from C3 to C;. These changes would change the inequality of total income fiom G to G, which from (15) can be written as A G-~~~~ - J-y wiC) (16) 1 which shows that (wjCj - wj will be the contribution of the jth income or expenditure component to the change in the inequality of total income. This contribution may be further decomposed into two components; one, due to a change in the share of the jth income compone and second, due to a change in the distiuton of the jth income component These individual components may be wntten as (w,C - W, -C)= w,(Wi W)(C1 C} ) . -(C1 C)(w, + w1) (17) in which tie first term on the right side measures the effect of a change in the share of the jth income component and te seoond term measures the effect of a change in the distrbution of the jth incoe amponent. Subsfituting (17) into (16) gives G- - G