Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Income inequality: some dimensions of the problem - Excerpts from a chapter of the new World Bank book that explores policies to improve income distribution.

Author(s):
International Monetary Fund. External Relations Dept.
Published Date:
September 1974
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Montek S. Ahluwalia

Recent discussions of economic development reflect an increasing concern with widespread poverty in underdeveloped countries. The fact of poverty is not new: it was always self-evident to those familiar with economic realities. What is new is the suspicion that economic growth by itself may not solve or even alleviate the problem within any “reasonable” time period. Indeed it is often argued that the mechanisms which promote economic growth also promote economic concentration, and a worsening of the relative and perhaps even absolute position of the lower-income groups. This pessimistic view has led to some questioning of growth-oriented development strategies which assume that the poverty problem would be solved without much difficulty if growth could be accelerated.

The empirical evidence underlying the new pessimism is limited but persuasive. Detailed studies of the nature and extent of poverty in particular countries show that the problem is of truly gigantic proportions. A study of poverty in India estimated that, in 1960, about 38 per cent of the rural population and 50 per cent of the urban population lived below a poverty level defined by consumption yielding 2,250 calories. A recent study of Brazil showed that, also in 1960, about 30 per cent of the total population lived below a poverty level defined by the minimum wage in northeast Brazil (the poorest region). More importantly, both studies argued that the situation had worsened over the 1960s, at least in terms of relative equality….

These studies raise important questions relevant to policy formulation. What is the extent of relative and absolute poverty in underdeveloped countries and does it vary systematically with the level of development? What evidence is there on the relationship between growth and inequality and how far can this relationship be affected by policy? What are the economic characteristics of the poor and what do they imply for distributional strategies? …

Relative inequality

The conventional approach to income inequality is to define the problem in purely relative terms. A familiar technique for this purpose is to measure inequality by the extent to which the income share of groups of individuals or households differs from their population share. In this section, we will examine the problem in terms of income shares of the lowest 40 per cent, the middle 40 per cent, and the top 20 per cent of households ordinally ranked by income….

Table 1 presents income share data for sixty-six countries cross-classified according to different levels of overall inequality and per capita income levels. The table distinguishes between three inequality levels defined as high, moderate, and low (according to specified ranges of the share of the lowest 40 per cent) and three income groupings defined as high, middle, and low (according to specified ranges of per capita GNP). The extent of inequality varies widely among countries but the following broad patterns can be identified.

Table 1.Cross-classification of countries by income level and equality 1
HIGH INEQUALITY

Share of lowest 40% less than 12%
MODERATE INEQUALITY

Share of lowest 40% between 12% and 17%
LOW INEQUALITY

Share of lowest 40%, 17% and above
PerPerPer
capitaLowestMiddleTopcapitaLowestMiddleTopcapitaLowestMiddleTop
Country (year)GNP US$40%40%20%Country (year)GNP US$40%40%20%Country (year)GNP US$40%40%20%
Kenya (1969)13610.022.068.0Burma (1958)8216.538.744.8Chad (1958)7818.039.043.0
Sierra LeoneDahomey (1959)8715.534.550.0Sri Lanka (1969)9517.037.046.0
(1968)1599.622.468.0Tanzania (1967)8913.026.061.0Niger (1960)9718.040.042.0
Iraq (1956)2006.825.268.0India (1964)9916.032.052.0Pakistan (1964)10017.537.545.0
PhilippinesMadagascar (1960)12013.525.561.0Uganda (1970)12617.135.847.1
(1971)23911.634.653.8Zambia (1959)23014.528.557.0Thailand (1970)18017.037.545.5
Senegal (1960)24510.026.064.0Korea (1970)23518.037.045.0
Ivory CoastTaiwan (1964)24120.439.540.1
(1970)24710.832.157.1
Rhodesia (1968)2528.222.869.0
Tunisia (1970)25511.433.655.0
Honduras (1968)2656.528.565.0
Ecuador (1970)2776.520.073.5
El Salvador
(1969)29511.236.452.4
Turkey (1968)2829.329.960.8
Malaysia (1970)33011.632.456.0Dominican RepublicSurinam (1962)39421.735.742.6
Colombia (1970)3589.030.061.0(1969)32312.230.357.5Greece (1957)50021.029.549.5
Brazll (1970)39010.028.461.5Iran (1968)33212.533.054.5Yugoslavia (1968)52918.540.041.5
Peru (1971)4806.533.560.0Guyana (1956)55014.040.345.7Bulgaria (1962)53026.840.033.2
Gabon (1968)4978.823.767.5Lebanon (1960)50813.026.061.0Spain (1965)75017.636.745.7
Jamaica (1958)5108.230.361.5Uruguay (1968)61816.535.548.0
Costa RicaChile (1968)74413.030.256.8
(1971)52111.530.058.5
Mexico (1969)64510.525.564.0
South Africa
(1965)6696.235.858.0
Panama (1969)6929.431.259.4
Venezuela (1970)10047.927.165.0Argentina (1970)107916.536.147.4Poland (1964)85023.440.636.0
Finland (1962)159911.139.649.3Puerto Rico (1968)110013.735.750.6Japan (1963)95020.739.340.0
France (1962)19139.536.853.7Netherlands (1967)199013.637.948.5United Kingdom
Norway (1968)201016.642.940.5(1968)201518.842.239.0
Germany,Hungary (1969)114024.042.533.5
Fed. Rep. (1964)214415.431.752.9Czechoslovakia
Denmark (1968)256313.638.847.6(1964)115027.641.431.0
New ZealandAustralia (1968)250920.041.238.8
(1969)285915.542.542.0Canada (1965)292020.039.840.2
Sweden (1963)294914.042.044.0United States
(1970)485019.741.538.8
Income up to US$300Income US$300-$750Income Above US$750Note: The income shares of each percentile group were read off a free-hand Lorenz curve fitted to observed points in the cumulative distribution. The distributions are for pretax income. Per capita GNP figures are taken from the World Bank data files and refer to GNP at factor cost for the year indicated in constant 1971 U.S. dollars.

The data used in the tables in this article are largely taken from Jain, S. and Tiemann, A. 1973, Size Distribution of Income: A Compilation of Data, Development Research Center Discussion Paper No. 4, World Bank, Washington, D.C.

Income up to US$300Income US$300-$750Income Above US$750Note: The income shares of each percentile group were read off a free-hand Lorenz curve fitted to observed points in the cumulative distribution. The distributions are for pretax income. Per capita GNP figures are taken from the World Bank data files and refer to GNP at factor cost for the year indicated in constant 1971 U.S. dollars.

The data used in the tables in this article are largely taken from Jain, S. and Tiemann, A. 1973, Size Distribution of Income: A Compilation of Data, Development Research Center Discussion Paper No. 4, World Bank, Washington, D.C.

The socialist countries have the highest degree of overall equality in the distribution of income. This is as we would expect, since income from the ownership of capital does not accrue as income to individuals. The observed inequality in these countries is due mainly to inequality in wages between sectors and skill classes. Since the structural factors operating toward equality are the strongest in these countries, their average income share of the lowest 40 per cent—amounting to about 25 per cent of total income—may be taken as an upper limit for the target income share to which policymakers in underdeveloped countries can aspire.

The developed countries are evenly distributed between the categories of low and moderate inequality. The average income share of the bottom 40 per cent amounts to about 16 per cent, which is lower than the average for socialist countries but better than most of the underdeveloped countries…. Most of the underdeveloped countries show markedly greater relative inequality than the developed countries. About half of the underdeveloped countries fall in the high inequality range with another third displaying moderate inequality…. Those of the underdeveloped countries classified in the low inequality category have income shares for the lowest 40 per cent averaging 18 per cent, as is the case with the most egalitarian of the developed countries. Against this, however, half the underdeveloped countries show income shares of the lowest 40 per cent, averaging only 9 per cent….

Absolute poverty

The extent of relative inequality in underdeveloped countries is an important dimension of the problem of income distribution, but it tells us little about the extent of absolute poverty. Yet much of the current interest in income distribution is not simply due to a concern with relative inequality. It is more often a concern with absolute standards of living in terms of calorie intake and nutrition levels, clothing, sanitation, health, education, and so on…. The incidence of poverty in underdeveloped countries defined in absolute terms has powerful appeal for dramatizing the need for policy action in both domestic and international spheres. Estimates of this type have been attempted for some countries using arbitrary poverty lines for each country to measure population below these levels. Similar estimates can be derived for the underdeveloped countries in Table 1 by combining income share data with total income estimates obtained from the national accounts. For each country we have estimated (see Table 2) the population living below two arbitrary “poverty lines” of annual per capita incomes of US$50 and US$75 (in 1971 prices)….

Table 2.Estimates of population below poverty line in 1969
1969 GNP

per capita
1969

population

(millions)
Population below US$50Population below US$75
CountryMillions% of total

population
Millions% of total

population
LATIN
AMERICA
Ecuador2645.92.237.03.558.5
Honduras2652.5.728.01.038.0
El Salvador2953.4.513.5.618.4
Dominican
Republic3234.2.511.0.715.9
Colombia34720.63.215.45.627.0
Brazil34790.812.714.018.220.0
Jamaica6402.0.210.0.315.4
Guyana390.7.19.0.115.1
Peru48013.12.518.93.325.5
Costa Rica5121.7..2.3.18.5
Mexico64548.93.87.88.717.8
Uruguay6492.9.12.5.25.5
Panama6921.4.13.5.211.0
Chile7519.6........
Venezuela97410.0........
Argentina105424.0........
Puerto Rico16002.8........
Average and Total545244.526.610.842.517.4
ASIA
Burma7227.014.553.619.271.0
Sri Lanka9512.24.033.07.863.5
India100537.0239.044.5359.366.9
Pakistan (E&W)100111.836.332.564.757.9
Thailand17334.79.326.815.444.3
Korea22413.3.75.52.317.0
Philippines23337.24.813.011.230.0
Turkey29034.54.112.08.223.7
Iraq3169.42.324.03.133.3
Taiwan31713.81.510.72.014.3
Malaysia32310.61.211.01.615.5
Iran35027.92.38.54.215.0
Lebanon5702.6..1.0.15.0
Average and Total132872.0320.036.7499.157.2
AFRICA
Chad753.51.543.12.777.5
Dahomey902.61.141.62.390.1
Tanzania9212.87.457.99.372.9
Niger943.91.333.02.359.9
Madagascar1196.73.653.84.769.6
Uganda1288.31.821.34.149.8
Sierra Leone1652.51.143.51.561.5
Senegal2293.8.922.31.335.3
Ivory Coast2374.8.37.01.428.5
Tunisia2414.91.122.51.632.1
Rhodesia2745.1.917.41.937.4
Zambia3404.2.36.3.37.5
Gabon547.5.115.7.123.0
South Africa72920.22.412.03.115.5
Average and Total30383.823.828.436.643.6
Average and
Grand Total2281200.3370.430.9578.248.2
Note:.. negligible.
Note:.. negligible.

The countries included in Table 2 account for about 60 per cent of the total population of the developing countries excluding China. About a third of this population falls below the poverty line defined by US$50 per capita and about half falls below US$75 per capita. Much of this is clearly due to the low levels of per capita income of many countries rather than to highly skewed income distribution patterns. India, Pakistan, Bangladesh, and Sri Lanka with 55 per cent of the total population together account for about 75 per cent of the population living below US$50. These countries are all characterized by low to moderate inequality. More interestingly, the table shows that a high per capita income does not ensure that there is no “absolute poverty” problem. Differences in patterns of income distribution between countries mean that the poverty problem may be equally serious in countries with very different per capita income levels. Both Ecuador and Sri Lanka have about a third of the population below the US$50 poverty line even though Ecuador’s per capital income is three times as high…. These estimates provide some indication of the scale of absolute poverty in underdeveloped countries and its relationship to per capita GNP and the distribution of income. Much of the poverty problem is a direct reflection of low levels of per capita income, but skewed distribution patterns are also important. Observed differences in the degree of inequality are such as to offset per capita incomes which are two or three times higher. It follows that development strategies which succeed in raising the level of per capita income may not have much impact on the poverty problem if they are accompanied by a deterioration in relative income shares.

Growth and the lowest 40 per cent

The above discussion of distributional patterns … has been limited to describing existing conditions. We have not considered whether these conditions are improving or deteriorating over time. Yet it is precisely these questions that are most often raised in evaluating performance and designing policy.

Measuring changes in distributional conditions can be done in terms of either relative income shares or absolute incomes. The limitations of a purely relative approach are self-evident: changes in relative equality tell us little about changes in income levels of the poor unless we also know what has happened to total income. An alternative approach which places greater emphasis on absolute income levels of the poor is to consider whether the levels of living of the poor have improved over time.

Systematic examination of such trends calls for time series data on both the distribution of income and the growth of income. Unfortunately time series data on the distribution of income are not available even for most developed countries. At most, we have a collection of countries for which distribution data are available for two points in time. These data can be combined with national accounts data to give us rough estimates of the income accruing to the lowest 40 per cent at two points in time. Figure 1 plots the estimated annual growth rate of income of the lowest 40 per cent against the rate of growth of GNP for 18 countries…. Countries above the 45 degree line are countries in which the income share of the lowest 40 per cent increased over the period so that the estimated rate of growth of income for this group is higher than for the economy as a whole. Countries below the 45 degree line are countries in which the relative income shares of the lowest 40 per cent declined.

Figure 1.Growth and the lowest 40 per cent

The scatter suggests considerable diversity of country experience in terms of changes in relative equality…. Both Peru and Sri Lanka, for example, experienced the same rate of GNP growth over the respective periods reported, but income of the lowest 40 per cent grew over 8 per cent per annum in Sri Lanka—compared to only 3 per cent in Peru—because of improvements in relative income shares. In other cases, a high rate of growth of GNP offsets a deterioration in relative income shares to produce substantial increases in income of the poor. Mexico and Brazil, for example, experienced an increase in inequality in terms of relative income shares but income of the lowest 40 per cent grew by about 6 per cent per annum in both cases.

Since individual observations are subject to substantial error, it is perhaps more important to look for patterns in the data. The evidence suggests that there is no strong pattern relating changes in the distribution of income to the rate of growth of GNP. In both high-growth and low-growth countries there are some which have experienced improvements and others that have experienced deteriorations in relative equality. The absence of any marked relationship between income growth and changes in income shares is important for policy purposes. It suggests there is little firm empirical basis for the view that higher rates of growth inevitably generate greater inequality. This may have happened in particular cases but an explanation for this must be sought in the circumstances of each particular case and not in terms of a generalized relationship….

Determinants of inequality

Ideally the determination of relative income shares should be analyzed in the context of a fully developed theory of the size distribution of income. Such a theory should take into account not only the economic factors affecting the distribution of income, but also the political and institutional context in which these factors operate. Needless to say, we are far from having such a comprehensive theory. There are, however, several partial hypotheses about particular factors affecting the distribution of income which provide some of the elements of a comprehensive theory. As a first step in the analysis we need to study the empirical validity of these hypotheses using available data.

In the absence of time series data, such tests must rely heavily on cross-country data of the type discussed above. Crosscountry differences in income inequality can be explained in terms of various “explanatory variables” reflecting different influences on distribution patterns. Associative relationships of this type cannot, of course, be presented as proof of causality, but they help to indicate relationships which deserve further study…. In this section we present preliminary results (obtained by the World Bank’s Development Research Center) using multiple regression to estimate equations “explaining” variations in the income share of the top 20 per cent, the middle 40 per cent, and the lowest 40 per cent.

The results of our cross-section analysis may be summarized as follows:

  • The explanatory variables used included both structural variables such as the level of per capita income and the share of agriculture in GDP and other variables which can be influenced by policy such as the rate of growth of the economy, the rates of enrollment in primary and secondary schooling, and the rate of growth of population. The variables explain about half of the observed variation in income shares across countries. The large proportion of variation unexplained is not surprising. We have not considered a number of potential explanatory variables which can be identified a priori. The most important of these is the concentration of wealth (including agricultural land) and mechanisms perpetuating this concentration pattern. Other economic factors that may be relevant are various institutional and market mechanisms that discriminate against low-income groups. The influence of these factors could not be explained due to lack of data and the difficulty in specifying an appropriate explanatory variable.
  • There is some confirmation of the hypothesis that income inequality first increases and then decreases with development….
  • Education is positively related to equality in terms of income shares of the lowest and middle group….
  • The growth of population is positively related to inequality as measured by the income share of the lowest 40 per cent….
  • The cross-section evidence does not support the view that a high rate of economic growth has an adverse effect upon relative equality. Quite the contrary, the rate of growth of GDP in our sample was positively related to the share of the lowest 40 per cent, suggesting that the objectives of growth and equity may not be in conflict.

The policy implications of these results are difficult to evaluate given the limitations of both the data and the methodology. Perhaps the most important finding is that income shares are related not only to structural factors such as per capita income levels but also to variables which can be influenced by policy. The level of education and the rate of growth of population are particularly important in this context since they indicate areas in which government action can improve distribution patterns…. Income share data provide a useful summary picture of the degrees of inequality in a country but they do not tell us about the underlying economic structures causing this inequality. These structures may vary widely between countries and yet produce the same overall degree of inequality. Since the impact of particular factors on the distribution of income will depend upon the nature of these underlying structures, there is no simple relationship between income shares and various determinants of inequality which is valid for all countries. The impact of education on income inequality, for example, cannot be determined independently of whether the structural characteristics of the economy are such as to encourage the absorption of skilled labor into high-wage employment. Broad cross-country comparisons can hardly capture the complex nature of these interactions.

Because of these limitations it may be more useful to adopt a disaggregated approach to analyzing the determinants of income distribution. Rather than trace direct relationships between broad measures of the distribution of income (such as income shares of percentile groups) and various economic factors, we can treat the problem in two stages. The first stage consists of identifying the composition of the low income population—the lowest 40 per cent, for example—in terms of homogeneous socio-economic groups with particular economic characteristics….

Poverty groups

The need for detailed “profiles of poverty” highlighting the economic characteristics of poverty groups has been widely emphasized in recent literature. Indeed, it can be argued that the overwhelming need for data on income distribution is not so much for better data on income shares as for better data on the sectoral distribution of the poor, their occupational characteristics and educational levels, their ownership of productive assets, and their access to key production inputs. These characteristics determine the processes of income generation in poverty groups and the constraints on these processes. In this section we … assume that the poverty groups can be defined in relative terms as the lowest 40 per cent of the population.

Rural-urban balance

The sector in which poverty groups are located is a key element in the profile of poverty, since governments frequently intervene in various ways to influence the sectoral balance of the economy. Such interventions have a direct impact on income distribution and can be designed to achieve distributional objectives.

The basic fact that the poor are disproportionately located in the rural areas and are engaged in agricultural or allied rural occupations is well established in conventional wisdom and easily verified. Table 3 presents data for three countries showing the distribution of income recipients in different percentile groups across broadly defined economic sectors. While the percentage breakdown is different for different countries, in all cases the poorest group corresponds to the lowest 40 per cent to 50 per cent of the population. About two thirds of this group earn their livelihood from agriculture and can be assumed to be small farmers and farm workers…. Given the scale of the problem and the limited capacity of other sectors to expand productive employment, it follows that a viable strategy for raising incomes of the lowest 40 per cent of the population must necessarily focus on the agricultural sector. But it is important to recognize that a mere shift in sectoral emphasis toward promoting agriculture and allocating resources to rural development is not enough. The impact of government policies on the target population will also depend upon the distributional incidence of these policies within the agricultural sector. General support schemes (which may involve significant direct and indirect resource costs) may prove inefficient for our purposes if the incidence of their benefits is sharply skewed in favor of upper-income groups in the rural areas.

Table 3.Sectoral distribution of income groups(Figures in each row are percentage distribution across sectors)
Percentile

groups
AgricultureMining

and

Industry
ConstructionTransport

and

commerce
ServicesOtherTotal
I. Mexico (1963)
Richest619.023.02.01937.0100.0
1730.019.02.01830.01.0100.0
3029.024.05.01625.01.0100.0
Poorest4763.09.06.0814.0100.0
Total10045.016.05.01821.0100.0
Percentile

groups
AgricultureMining

and

Industry
ConstructionTransport

and

commerce
ServicesOtherTotal
II. Malaysia (1970)
Richest59.014.03.02547.02.0100.0
4633.013.04.12424.02.0100.0
Poorest4971.07.02.0127.01.0100.0
Total10050.010.03.01817.02.0100.0
Percentile

groups
AgricultureMining

and

Industry
ConstructionTransport

and

commerce
ServicesOtherTotal
III. Chile (1968)
Richest533.016.09.01819.05.0100.0
1951.013.05.01314.04.0100.0
3757.014.08.086.07.0100.0
Poorest3970.07.05.052.011.0100.0
Total10056.010.06.0108.010.0100.0

This problem is often ignored in policy formulation, but its importance can be appreciated by examining income inequality within the rural sector and inferring from the existing degree of the inequality the distributional incidence of general policies aimed at the agricultural sector. The assumption that increases in sectoral income will be distributed along the same lines as total income is arbitrary but quite persuasive. Income distribution in agriculture is determined largely by structural factors, such as the distribution of land, and it is reasonable to suppose that the distribution of additional income generated will be similarly determined. Table 4 presents data on the degree of inequality in the rural sector for ten countries and compares it with inequality in the urban sector. In most countries the rural sector is more equal than the urban but the degree of inequality in the rural sector is nevertheless considerable. Even if we defined the target beneficiaries as the lowest 80 per cent of the rural population, this group receives only about 50 per cent of total income; the rest is appropriated by the upper 20 per cent, who do not form part of the target population.

Table 4.Rural and urban inequality
Share of top 20%Share of lowest 80%
CountryRuralUrbanRuralUrban
1 Chile (1968)48.350.251.749.8
2 Colombia (1970)50.758.249.341.8
3 Honduras (1968)55.055.845.044.2
4 India (1964)43.057.057.043.9
5 Mexico (1963)54.056.246.043.8
6 Pakistan (E&W) (1964)42.552.057.049.0
7 Panama (1968)46.045.354.054.7
8 Thailand (1970)51.045.549.054.5
9 Tunisia (1961)50.050.050.050.0
10 Venezuela (1962)50.050.050.050.0

Policies of general support to a particular sector, therefore, are likely to involve substantial leakage of benefits beyond the intended beneficiaries. Given the high resource costs of most sector-promotion strategies, there is a clear need to design these policies to ensure that such leakages are minimized. These designs must be based on the specific socio-economic characteristics of poverty groups.

Employment

Along with the sectoral characteristics discussed above, we need better information on the employment status of poverty groups in terms of broad categories such as “employer,” “employee,” “self-employed,” “unemployed,” and so forth. Differentiation along these lines is useful for policy purposes because the determinants of income generation in each group are different and the policy intervention needed to help each group will differ accordingly.

Particularly relevant in designing suitable policies is the relative importance of wage employment as the primary source of income for the poor. Discussion of distributional problems in underdeveloped countries is frequently conducted in terms of factors affecting the level of employment and the share of labor…. But the available data suggest that this classification may not be very illuminating in underdeveloped countries. A substantial proportion of the poor in these countries are not engaged in wage labor, nor can they be described as currently unemployed and searching for employment. They are engaged in production as “independent” workers, i.e., they are self-employed but suffer from very low income levels….

The existence of self-employment in the poverty group has immediate implications for both theoretical analysis and policy formulation. In countries where the poverty group is essentially a part of the labor market, distribution policies must rely heavily upon expanding employment to absorb the unemployed and upgrading the structure of demand for labor to generate high-wage employment for those currently employed at low wages. But if the bulk of the poverty group is engaged in self-employment, this approach may not be sufficient. Expanding employment is undoubtedly one way of absorbing the population engaged in low-income activity, but we also need to consider the alternative of raising production levels in existing occupations. An operational plan along these lines requires a detailed identification of the different types of self-employment in which the poor are engaged, the constraints on production in these occupations, and the extent to which these constraints can be relaxed by policy action.

Ownership of capital

Another characteristic of poverty groups that is relevant for the diagnosis of distribution problems is their lack of capital. Comprehensive data on the distribution of wealth are not available for any underdeveloped country, but there can be little doubt that the distribution of total productive wealth in these economies is even more unequal than the distribution of income. This inequality is an underlying cause of income inequality since concentration of productive assets produces greater concentration in income.

The most important productive asset for our purposes is agricultural land, which is a critical constraint on income of small farmers. Data on the distribution of land by size of holding…indicate severe concentration patterns with the bulk of operational holdings being of very small size and accounting for a small proportion of total crop land. Estimates based on the 1960 world census for agriculture suggest that today there are more than 100 million smallholders in the developing countries, operating farms of less than five hectares. About half of these holdings are less than one hectare. The problem of poverty in this socio-economic group is therefore inseparably linked to the availability of land, or at least to the availability of capital needed to improve the quality of the land. Similar arguments apply to the self-employed urban poor who suffer from constraints in terms of supply of capital.

The skewed distribution of land—or other productive assets—presents a static picture of the problem of lack of capital in poverty groups. Behind this static picture are a number of forces which tend to generate and perpetuate this concentration pattern over time….

First, the demographic characteristics of poverty groups may operate systematically in favor of capital concentration. If population growth in poverty groups is faster than for the rest of the economy, there is a tendency toward greater dilution of owned capital in these groups. In the case of agriculture this would lead to progressively diminishing holdings, or larger families supported by the same holding, and also to a steady migration of landless poor to the cities….

Second, differences in rates of saving by income class perpetuate patterns of capital concentration over time…. A large number of household budget surveys show that average savings rates are much lower for lower income groups…. There are many reasons why the observed savings rates in the low-income groups are likely to be underestimates, so that observed variations in savings rates probably exaggerates the true variation. But even when allowance is made for these factors it is likely that different income classes (and also socio-economic classes) do save different proportions of income and thus have very different ability for generating internally the supply of capital needed to raise incomes.

Finally there is the problem of access to capital. The constraints on production in the poverty groups are not solely due to the lack of internal generation of capital. They are also a reflection of a limited access to capital due to market fragmentation, institutional rigidities, and other forms of nonmarket allocation mechanism. Limitations of access cover a broad range including tenancy rights, access to financial markets, and access to public infrastructure, all of which impose constraints on the ability to raise production in the poorer groups. These limitations are probably as important as the observed concentration in ownership of capital.

These factors perpetuating inequality all relate to the availability of physical capital in poverty groups. Equally important from the point of view of income distribution is the limited availability of human capital. As we would expect, the many economic disadvantages of the poor are also reflected in a lack of schooling. Low levels of education and other labor skills may be an important constraint on the ability to absorb the low-income population into an expanding modern sector.

The economic characteristics discussed above are obviously not a comprehensive list for analyzing problems of income distribution in underdeveloped countries. They serve only to translate the general concern about low-income groups—the lowest 40 per cent—into a concern about specific groups with defined socio-economic characteristics. But this is only the first stage of the analysis of distributional problems. These characteristics together with the characteristics of the rest of the economy interact to determine the distribution of income between groups. The scope for government intervention is then determined by the extent to which we can affect these interactions through policy. (Subsequent chapters in Redistribution with Growth are addressed to these questions.)

*The book is being published by the Oxford University Press (copyright of the World Bank). U.S. price is $16 cloth and $4.50 paperback; U.K. price is £5 and £1.40.

Other Resources Citing This Publication