Download Impact of Income Inequality on Growth and Poverty in SE Asia and Latin America and more Papers Comparative Law and Politics in PDF only on Docsity!
Chapter 1
MAGNITUDE OF INEQUALITIES
Latin America and the Caribbean have the greatest dis-
parities in income distribution in the world. A quarter of
all national income is received by a mere 5 percent of the
population, and the top 10 percent receive 40 percent. Such
proportions are comparable only to those found in some
African countries, whose per capita income levels are half
those of Latin America, and they are considerably higher
than those of any other group of countries (Figure 1.1). In
Southeast Asian countries, the wealthiest 5 percent receive
16 percent of all national income on average, while in the
developed countries they receive 13 percent.^1
The counterpart to the great concentration of income
in the hands of the wealthy is found at the other end of the
income scale in Latin America: the poorest 30 percent of
the population receive only 7.5 percent of total income,
less than anywhere else in the world, where it is over 10
percent. (Figure 1.2). This income concentration applies
more to Latin America—which is the focus of this study
due to the availability of information—than to the English-
speaking Caribbean, where disparities are more moderate.
The indicator most commonly used to measure in-
come inequality is the Gini index, which draws together
information about the breakdown of income among all
12
14
16
18
20
22
24
26
Income of the 5% wealthiest
$2,000 $4,000 $6,000 $8,000 $10,000 $12,000 $14, GDP per capita
Figure 1.1. Income Received by the
Wealthiest 5 Percent of the Population
(Percent of total income)
Source: IDB calculations based on Deininger and Squire (1996).
Africa
Latin America
Rest of Asia
Southeast Asia
Developed 7
8
9
10
11
12
13
Income of the poorest 30%
$2,000 $4,000 $6,000 $8,000 $10,000 $12,000 $14, GDP per capita
Figure 1.2. Income Received by the Poorest 30 Percent of the Population (Percent of total income)
Source: IDB calculations based on Deininger and Squire (1996).
Africa
Latin America
Rest of Asia Southeast Asia
Developed
(^1) The source for these comparisons is the database in Deininger and Squire (1996), which has information on income distribution based on reliable house- hold surveys in 108 countries. Southeast Asia includes only Hong Kong, Ko- rea, Singapore and Taiwan. (^2) Henceforth, the IDB’s own calculations based on the most recent house- hold surveys are used, except for Colombia, Guatemala, Jamaica and the Do- minican Republic, for which the most recent findings according Deininger and Squire (1996a) are used. Appendix 1.2 describes the characteristics of the sur- veys and the main socioeconomic indicators by income deciles and country.
population groups (Box 1.1). The Gini average for all the
countries in the world for which there is reliable informa-
tion on income distribution is 0.4. A perfectly equal dis-
tribution would produce an index of zero, but in fact the
best instances of distribution such as Spain, Finland and
some other European countries show Gini indices of be-
tween 0.25 and 0.3 (Figure 1.3). At the opposite extreme,
the indices of greatest income inequality are around 0.6,
which are found almost solely in Latin America and the
Caribbean. With the exception of Jamaica, whose inequal-
ity index of 0.38 is closer to European than to Latin Ameri-
can patterns, the other countries in the region for which
there is reliable information for the 1990s show inequal-
ity levels higher than the world average, and 11 of them
have indices higher than 0.5.^2
12 CHAPTER 1
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Figure 1. Typical Lorenz Curve
in Latin America
(In percent)
A
B
Gini coefficient
5 10 15 20 25 30 Income of 5th quintile/1st quintile
Figure 2. Gini Coefficients and
Income Gaps
*Countries with urban data only. Source: IDB calculations based on recent household surveys, and Deininger and Squire (1996a).
Mexico Panama
Peru Costa Rica
Bolivia*
Uruguay*
Paraguay
El Salvador
Ecuador
Argentina*
Honduras
Brazil
Venezuela
Chile
Bahamas
Trinidad and Tobago
If incomes were distributed in a fully equitable manner, each person would receive the same share of income. The Gini index measures how far real distribution is from such a hypothetical reference point. Real income distribution (or that of any other vari- able) can be represented as a cumulative curve showing what percentage of total income is received by each per- centage of the population, lined up with its income level (which is called a Lorenz Curve). Consider points A and B in Figure 1. Point A indicates that the poorest 20 percent of the population receive 4 percent of income and that 90 percent receive 60 percent, which is actually more or less what happens in Latin American economies. Inasmuch as the diagonal represents the case of perfect distribution, the Gini coefficient is simply the area separating the Lorenz curve from the diagonal, divided by the area under the di- agonal. In theory, the Gini coefficient can vary between zero—perfect distribution—and one—complete concen- tration in a single person. In practice, Gini coefficients of per capita income vary between 0.25 and 0.60. Of the countries for which comparable information is available, only five have Gini coefficients outside of this range. In- equality indices in Latin America are on average 0.52, with a minimum of 0.43 for Uruguay and a maximum of 0.59 for Brazil. There are other measurements of income inequal- ity, such as the Theil index or the logarithmic income vari- ance. Each inequality measure assigns a different weight to observations by income level, which can be interpreted as a way of aggregating individuals in order to obtain an overall measurement of social welfare. The empirical evi- dence shows that all the usual measurements of inequal- ity produce highly correlated results, and hence for pur- poses of comparative analysis between countries, either one is adequate. Instead of the Gini or one of the other measure- ments of concentration, some economists prefer to refer simply to the income gaps between the groups at either end, such as between the wealthiest 20 percent and the poorest 20 percent of the population (called the first and fifth quintiles, respectively), or between the wealthiest and poorest 10 percent (tenth and first deciles). This measure- ment is easy to understand, and it relates quite closely to the Gini coefficient (Figure 1). It is, however, a rather crude measurement, for it is based simply on comparing two points in the distribution curve, and that can be decep- tive. For example, Ecuador and Panama have income gaps greater than Brazil’s, but their Gini coefficients are better because the distribution among their middle groups is better.
Box 1.1. Gini: A Simple Way to Measure Income Inequality
14 CHAPTER 1
Note that inequality of total income is greater than
the average of the urban and rural inequalities taken sepa-
rately, except in Chile, where the total and the urban Gini
are the same (Figure 1.5). If the average income levels in
the countryside and the city were similar, the inequality
of total income would be located at some point between
the separate concentrations of one region and the other.
But urban incomes are substantially higher than rural
ones, which means that, taken together, income inequal-
ity will be greater than in each region taken separately.
The largest gaps are found in Brazil, where per capita in-
come in urban areas is three times what it is in rural ar-
eas, and the smallest in Costa Rica and Venezuela, where
urban incomes are 75 percent higher than rural ones (Fig-
ure 1.6). Due to these great differences in per capita in-
come, total income inequality is greater than the sepa-
rate concentration in the countryside and in the city. On
average, the total Gini indices are 3.3 points higher than
separate Ginis for rural and urban areas. It is important
to keep this difference in mind, inasmuch as for three of
the countries analyzed in this study—Argentina, Bolivia
and Uruguay—information is not available on rural in-
come distribution (and hence they do not appear in these
comparisons). For example, the Ginis reported for coun-
tries that cover only urban areas (and only Greater Buenos
Aires in the case of Argentina) tend to be underestimated.
In Figure 1.5, we have assumed that the underestimate is
by these 3.3 points. With this correction, Uruguay dis-
plays the same inequality index as Costa Rica, Argentina
comes close to El Salvador, and Bolivia remains at in-
equality levels similar to those of Chile or Panama.
AS BAD AS BEFORE
Regardless of the measurement used, Latin America
stands out among all regions for its high inequality. In-
come distribution has not improved during the 1990s,
and according to the limited information available, it is
as high today as it was two decades ago (Figure 1.7).
The period of rapid growth in the region, which be-
gan in the 1960s and lasted until the outbreak of the debt
crisis in 1982, led to a notable improvement in income
distribution. Between 1970 and 1982 the region’s Gini
coefficient^4 fell by 5 points (that is, by 10 percent), and
the income ratio gap between the wealthiest 20 percent
of the population and the poorest 20 percent fell from 23
to 18 during that same period. While low-income groups
apparently improved their income share by around 10
percent, the highest groups stood still or fell, especially
between 1980 and 1982 (Figure 1.8). The wealthiest 10
percent saw their share of income fall by 6 percent dur-
ing this period and middle groups gained significantly.
But these improvements in distribution were short-lived.
During the 1980s, the decile with the highest incomes
increased its share by over 10 percent at the cost of all
other income deciles. The poorest 10 percent in Latin
Gini coefficient of total income
0.42 0.44 0.46 0.48 0.50 0.52 0.54 0.56 0. Gini coefficient of urban income
Argentina*
Bolivia* Brazil Chile
Costa Rica
Ecuador
Honduras
Mexico
Panama
Peru
Paraguay
Uruguay* El Salvador
Venezuela
Figure 1.5. Total and Urban Income Concentration
*Actual data cover urban incomes only. Gini for total incomes are estimates. Source: IDB calculations based on recent household surveys.
Urban/rural income
Brazil Mexico
Ecuador Paraguay
Panama El Salvador
Peru Honduras
Chile Costa Rica
Venezuela
Figure 1.6. Urban-Rural Income Gap
Source: IDB calculations based on recent household surveys.
(^4) Approximations based on available information for the population of 13 countries that together make up 83 percent of the Latin American population (Londoño and Székely, 1997). Note that the Gini inequality index for the entire region is not the same as the average of Ginis by countries, since it is based on the incomes of all individuals combined. The Gini for the region is higher due to differences of average income between countries. Since these estimates are based on partial evidence of the evolution of distribution in each country, they should not be considered an exact description of the behavior of this variable over time.
Magnitude of Inequalities 15
America suffered a 15 percent drop in their share of in-
come. Thus the gaps widened again and the improve-
ments in distribution from before the debt crisis were
wiped out.
The economies of the region have undergone great
changes in the 1990s. High inflation rates have been
halted, deep economic reforms have been adopted to fa-
cilitate market operations, and productivity and economic
growth have been restored. All these changes have brought
about shifts of wealth and income. Concentration, how-
ever, has remained practically unchanged: the region’s Gini
has stood at around 0.58. What explains this apparent con-
tradiction is that the changes have affected some groups
differently from others. The poorest 10 percent in the re-
gion saw a 15 percent loss of their share in income be-
tween 1990 and 1995, and the next 10 percent a loss of 4
percent. The richest 10 percent also suffered a relative set-
back, while those who gained were the remaining groups
in the middle. Hence, although the indicators of average
concentration for the region have changed very little dur-
ing the 1990s, distribution has by no means stood still.
Moreover, movement has not been homogeneous
between countries. In Brazil, Chile and Mexico, income
inequality worsened in the 1980s, but that was halted in
the 1990s. In Colombia and Costa Rica, distribution pat-
terns have remained quite stable, and indices of concen-
tration in the 1990s have stood at levels similar to what
they were a decade ago. In Honduras and Jamaica, in-
come distribution worsened in the early 1990s, but in re-
cent years it has been better than in the 1980s. In Ve-
nezuela, there have been periods of sharp decline, but
they have been transitory.^5
WHERE INEQUALITY IS FOUND
One of the most striking features of the poor income dis-
tribution in Latin America is the huge gap between the
families in the highest income decile and everyone else.
In relatively egalitarian societies such as Sweden or
Canada, an individual who belongs to the wealthiest
decile of the population earns on average 20 to 30 per-
cent more than someone in the next decile. The succeed-
ing differences in the next deciles are also lower, and
hence there are no sharp gaps between social strata. In
Latin America, the gaps within the middle-income groups
are not so pronounced, but between the wealthiest decile
and the next one there is an abyss: in the Dominican Re-
public or Chile, to cite the two most critical cases, the
income of someone who belongs to the tenth decile is
three times as great as in the previous decile and more
than 30 times greater than that of the poorest decile (Fig-
ure 1.9a). These differences are possibly even greater,
because income from capital, which is more important in
the richest decile of the population, is most certainly
underreported in the income surveys from which these
calculations are taken (Box 1.2).
At the other end of the income scale there are also
major gaps: because the incomes of the poorest 10 per-
cent of the population are really quite low, the next 10
percent receive twice as much income in most countries,
and in Ecuador and Panama close to triple the income of
Gini coefficient
1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994
Figure 1.7. Income Concentration in Latin America, 1970-
Source: Londoño and Székely (1997).
1970 1975 1980 1985 1990 1995
Poorest 20% Middle quintiles Wealthiest 20%
Figure 1.8. Participation of Each Income Group in Total Income, 1970- (Percent of normalized income)
Source: Londoño and Székely (1997).
(^1982 )
(^5) The countries mentioned are those for which there are at least five original observations since 1980. The interpolations and calculations used in the fig- ures are by Londoño and Székely (1997).
Magnitude of Inequalities 17
Who Are the Rich?
Among the many aspects that set the heads of house-
holds of the richest 10 percent of the population apart,
four features in particular stand out: their level of educa-
tion, the features of their work, where they live, and the
number of children they have. These characteristics are
valuable keys for determining the factors that cause and
perpetuate poor income distribution.
Education is the main productive resource on which
most people rely. This is even valid for the wealthiest 10
percent of the population. On average, in the 14 coun-
tries considered, the heads of household of the highest
income decile have 11.3 years of education. Although this
level amounts to slightly less than finishing secondary
school (12 years in most countries), it is 2.7 years higher
than the education level of the heads of household of
the next decile and almost seven years higher than the
heads of household of the poorest 30 percent of the popu-
lation. The most pronounced education gaps between the
two wealthiest deciles are found in Brazil, Mexico and
Box 1.2. Household Surveys: A Basic Source of Information on Inequality
Information on income distribution comes mainly from the sur- veys made regularly by governments of representative samples of households on a national level. Besides information on the income of each household, such surveys provide information on education, age, occupation and other characteristics of each household member. They are therefore a rich source for study- ing the relationship between income distribution, work activity, participation in the education system, and the size and make- up of households. Hence, they are currently the most effective instrument for designing and evaluating all kinds of social pro- grams. Even though this tool is now used around the world, sur- veys are still not carried out in some Latin American countries (Guatemala, Guyana and Haiti), while others perform surveys that leave out rural or nonmetropolitan areas (Argentina, Bo- livia and Uruguay). For the sake of studying income distribution, the most common shortcoming in some Latin American surveys is the lack of information on nonlabor sources of income (such as rent- ing out of housing, noncash incomes, etc.) and on complemen- tary variables that help establish the socioeconomic level of fami- lies (access to services, for example). Because nonlabor incomes are often reported at less than their true value, especially among upper-income families, sur- veys tend to underestimate the level of real inequality in coun- tries. If adjustments are made to correct this shortcoming, the results are subject to the arbitrary character of the method used. For Mexico in 1994, for example, Gini coefficients ranged from 0.477 to 0.60, depending on the adjustment method used.
The household surveys used in this study are of high qual- ity and in keeping with international standards (Deininger and Squire, 1996a). Nevertheless, they do not all apply the same methodology for data gathering and sampling, and in a few cases with regard to the observation unit.^1 Since it is never possible to attain perfect comparability between surveys of different coun- tries, differences in methodology do not necessarily invalidate the making of comparisons (Atkinson, 1995). In this study we have used reasonably comparable surveys, avoided any kind of subjective adjustment to the primary information, and applied identical methods of processing and statistical analysis to all surveys used. (See the Appendix to this chapter for the main features of the surveys used.) In order to deal with the shortcomings in coverage and quality of the surveys and make them more relevant for policy design, the Inter-American Development Bank has implemented the Improvement of Surveys and Measurement of Living Stan- dards (Mecovi) Program, in coordination with governments and other international agencies. The program provides financing and technical assistance for carrying out household surveys.^2
(^1) See Berry, et al (1983), Atkinson and Micklewright (1992), and Gottschalk and Smeeding (1997). (^2) The program has provided funding to carry out surveys in Argentina, El Salvador, Paraguay and Peru and will expand its operations to Bolivia and Nicaragua in the near future.
Honduras, where they are over three years, and only in
Peru are they less than two years (Figure 1.12). Between
the richest decile and the 30 percent at the bottom of the
income scale, the average schooling gaps are over nine
years in Mexico and between eight and nine years in Bra-
zil, Panama and El Salvador, all countries with high in-
come inequality. The lowest education gaps between rich
and poor are found in Uruguay, Venezuela and Peru, coun-
tries whose income inequality is moderate when com-
pared with patterns in the region. Thus, education is a
factor differentiating the rich, not so much because their
education level is so high, but rather because most oth-
ers have not spent much time in school.
A second distinctive feature of the wealthy is the
kind of work they do. A quarter of the heads of household
of the highest income decile work directly as profession-
als, technical personnel or senior executives of compa-
nies. The portions range from 18 percent in Honduras and
Paraguay to over 35 percent In Bolivia and Panama (Fig-
ure 1.13). In the next income decile, these proportions
drop off noticeably in most countries, and between the
18 CHAPTER 1
three lowest deciles only the tiniest number of heads of
household reach leadership positions or have technical
responsibilities. Most successful is Bolivia, where 6 per-
cent of the poorest heads of households occupy such a
position. Moreover, in most countries, between 10 per-
cent and 20 percent of the population of the highest in-
come decile are employers (Figure 1.14). Of the 14 coun-
tries for which such information is available, only in
Panama is this proportion under 10 percent, while in Ec-
uador it represents around 25 percent of the wealthiest
decile. In the next decile far fewer have the chance to
employ other workers. Differences with the wealthiest
decile are especially marked in the countries where in-
equality is greatest, such as Brazil, Paraguay, Ecuador and
Chile. The proportion of individuals who can employ oth-
ers is considerably less in lower income groups, even
though it is still somewhat significant in the lesser devel-
oped countries, such as El Salvador, Honduras, Bolivia
and Ecuador.
A third feature differentiating the rich is where they
live. In Latin America, poor households are found prima-
rily in the countryside, while most of the wealthy live in
cities. Among the countries in Figure 1.15, only in Brazil,
Chile and Venezuela are over half of the households of
the three lowest deciles in urban areas. By contrast, nine
out of ten of the households of the two highest income
declines are urban, except in Honduras, where the pro-
portion falls below eight.
Number of children is another characteristic that
varies between rich and poor households. In all the coun-
tries studied, the number of children in the highest in-
come decile is lower than in any other decile. In Hondu-
ras, where the families of the wealthy are the largest relative
to the wealthy in other countries, they do not even have
on average two children under 18. In Argentina and Uru-
guay, only one out of two families of the highest decile has
a minor child. The contrast with the poorest households is
striking: the average number of children in the three low-
est deciles is over two in all countries, reaches 3.5 in Mexico,
Peru and Venezuela, and is four children per household in
Paraguay (Figure 1.16). Hence, per capita income in each
home is higher in the top decile not only because heads of
household earn more, but because the number of persons
among whom it must be distributed is smaller.
Decile 10 Decile 9 Deciles 1-
Figure 1.13. Percentage of Professionals and Executives in Rich and Poor Families
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
Uruguay* Costa Rica
Peru
Venezuela
Argentina
El Salvador
Honduras Bolivia*
Mexico
Panama
Chile
Ecuador
Paraguay
Brazil
0
10
20
30
40
Figure 1.14. Percentage of Employers in Rich and Poor Families
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
Uruguay* Costa Rica
Peru
Venezuela
Argentina
El Salvador
Honduras Bolivia*
Mexico
Panama
Chile
Ecuador
Paraguay
Brazil
Decile 10 Decile 9 Deciles 1-
0
5
10
15
20
25
Decile 10 Decile 9^ Deciles 1-
Figure 1.12. Years of Schooling in Rich and Poor Families
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
Uruguay* Costa Rica
Peru
Venezuela
Argentina
El Salvador
Honduras Bolivia*
Mexico
Panama
Chile
Ecuador
Paraguay
Brazil
2
4
6
8
10
12
14
20 CHAPTER 1
index of 0.52 that is the inequality of per capita house-
hold income. This similarity can also be seen in each
country by itself, which indicates that it is not simply the
result of aggregating individual cases that could be very
dissimilar among themselves (Figure 1.20). The greatest
differences between one coefficient and another are found
in Bolivia and Panama, but even in these cases they only
reach five points. Similar differences between household
per capita income distribution and labor income per
worker are also found in developed countries.
The connection between labor income inequality
and how per capita income is distributed should not lead
to the conclusion that the impact of income from capital
(or from other sources) is secondary in income distribu-
tion. What it suggests is rather that such factors are not
independent of one another. As will be seen in Chapters
2 and 4, the differences in income between some workers
and others, or more precisely between workers who have
greater or lesser education, can be explained by the rela-
tive demands for different types of work, which depend
on the relative abundance of other factors and how they
interact among themselves and with the labor factor.
WHY SHOULD WE CARE?
Social Justice
Income distribution should be a cause for concern for
reasons of ethics and social justice. If income distribu-
tion reflected solely personal preferences about work,
effort and savings, there would be no reason for it to con-
stitute an ethical problem from the standpoint of distribu-
Percent of employers, deciles 9-
0.40 0.45 0.50 0.55 0. Gini coefficient of per capita income
Uruguay*
Costa Rica
Venezuela
Peru (^) Argentina*
El Salvador Bolivia*
Mexico
Honduras
Chile
Ecuador
Paraguay
Panama
Brazil
Figure 1.18. Income Concentration and Gap in the Proportion of Employers among the Rich
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
0
Children per family, deciles 9-
0.40 0.45 0.50 0.55 0. Gini coefficient of per capita income
Uruguay* (^) Costa Rica
Venezuela
Peru
Argentina*
El Salvador
Bolivia*
Mexico
Honduras
Chile
Ecuador Paraguay
Panama
Brazil
Figure 1.19. Income Concentration and Number of Children among the Rich
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
tive justice. Inequality and poverty become an ethical
matter calling for outside intervention when it is acknowl-
edged that the conditions generating them are not the
choice of individuals but rather a result of circumstances
beyond their control or a legacy of past problems.
Once it is acknowledged that effort and attitudes
toward education, work, risk and savings are not inde-
pendent of each individual’s starting conditions, the way
is opened for other concerns. The point is not simply to
assure “equality of opportunity.” If equality of opportu-
nity is understood as equality of access (free basic edu-
cation), it will not be enough to assure equality of capa-
bility of use (attendance at school), let alone equality of
results (academic achievement). According to the social
Figure 1.20. Concentration of Total and Labor Income
Gini coefficient of labor income
*Countries with urban data only. Source: IDB calculations based on recent household surveys.
Gini coefficient of household income
0.2 0.3 0.4 0.5 0.
0.2 Germany
UK
USA
Uruguay*
Argentina
Peru
Costa Rica
Venezuela
El Salvador
Panama
Honduras
Bolivia* Mexico
Paraguay
Ecuador
Chile Brazil
Magnitude of Inequalities 21
justice objectives pursued by society in each area, policy
actions ought to be aimed at changing the distribution
of capabilities of use (school subsidies, for example) or
the distribution of results (leveling programs and other
kinds of support).
Equality and Economic Development
After having been neglected for years in the academic
debate among economists, the topic of income distribu-
tion has been regaining interest in recent years. The cur-
rent debate centers precisely on determining whether,
when governments try to improve distribution, they pro-
duce adverse consequences for the welfare of the popu-
lation. Until recently, most theories on the subject pre-
sumed there was a conflict between equity and growth
(Box 1.3). The usual arguments were either that greater
concentration would allow for the generation of more
savings, thereby facilitating investment and growth, or
that concentration was the other side of the coin of effort
and productivity. But empirical international evidence
Since Adam Smith, most economists have considered growth and equality to be largely incompatible. Only recently has a new school of theoreticians come to argue what seems plain to the eye: that high levels of economic development are found in rela- tively egalitarian societies where accumulating physical, human and social capital is attractive. For classical economists, the capital accumulation needed for growth could only be obtained from the savings of capitalists, because workers would always consume their entire earnings. Equity and growth were incompatible. A similar con- flict emerged within the pure neoclassical vision, but it was rooted in microeconomics: any redistribution would ultimately be at the expense of a productive factor, thereby lowering eco- nomic efficiency. Even a decade ago, these two visions had a great deal of influence on the position of economists on the issue of redistribution. Both implied that a shortage of produc- tive resources would become sharper if a portion were set aside for redistribution purposes, thereby constituting an expenditure, not an investment. Because redistribution policies would have nothing to offer for economic growth or development according to these traditional approaches, income distribution was for a long time absent from the central agenda of economic theory and policy (Atkinson, 1997). Over the past decade, the relationship between growth and equity has come back to the center of the academic debate, thanks to theories of endogenous growth and political economy. The central argument is that poor income distribution can weaken the pace of the accumulation of physical and human capital or affect productivity growth, which are the sources of economic growth.^1 Education can be the channel through which poor income distribution lowers the possibilities of growth: families with lim- ited resources are not in a position to put aside money for edu- cation, even though such an effort could be socially and eco- nomically profitable. Moreover, families with little education and few possibilities for future education for their children will pre- fer to have more children than those in the opposite situation, thereby reinforcing the vicious circle of inequality and poverty (see Banerjee and Newman, 1991; Galor and Zeira, 1993; Aghion and Bolton, 1997; Piketty, 1997; and Dahan and Tsiddon, 1998).
Box 1.3. Equality and Growth: Partners or Adversaries?
Restricted access to capital markets is another channel that perpetuates poor income distribution. Because access to credit requires being able to provide guarantees, those who ini- tially have a higher level of wealth have more opportunity to invest in physical and human capital. Hence, in societies where wealth is very concentrated, many investments that could be profitable at the individual and social level cannot be made, thereby impeding growth (Aghion and Bolton, 1992; Galor and Zeira, 1993; and De Gregorio and Kim, 1994). The connection between inequality and growth can also take place through various political and economic channels. The first is political participation, through which voters express their preferences for policies of economic redistribution. In a society where wealth and income are highly concentrated, most people will support redistribution policies financed with taxes on capi- tal or similar measures that will discourage investment and pro- ductivity. The “median voter” is less inclined toward such poli- cies to the extent that wealth is better distributed (Alesina and Rodrik, 1992 and 1994; Persson and Tabellini, 1992 and 1994; Alessina and Perotti, 1993; and Perotti, 1994). A second connec- tion is that poor income distribution causes distributional and social tensions that lead to political instability and uncertain- ties that hinder investment. A third possibility is that the power groups that arise in unequal societies can erode genuinely dis- tributive policies by seizing government institutions and other income-producing activities, perpetuating inequality and low growth (Benhabib and Rustichini, 1996; and Birdsall and Londoño, 1998). Education, restricted access to capital markets, and po- litical and economic mechanisms conditioning government poli- cies are thus different channels by which income distribution affects growth.
(^1) Alesina and Perotti (1994), Alesina and Rodrik (1994) and Solimano (1998) contain excellent reviews of this literature.
Magnitude of Inequalities 23
The relationship between income distribution and economic growth has been a recurring topic of discussion among econo- mists. In theory, arguments can be found to justify both a direct and an inverse relationship between the two variables. What does the empirical evidence say? The prevailing conclusion in recent empirical studies is that poor income distribution harms economic growth and that therefore, instead of the conflict between equality and develop- ment that used to be regarded as a serious constraint on distri- bution policies, the relationship is mutually reinforcing. Studies by Clarke (1992), Alessina and Perotti (1993 and 1994), Persson and Tabellini (1994), Birdsall, Ross and Sabot (1995), Deininger and Squire (1996b) and Perotti (1996) conclude that poor income distribution reduces prospects for economic growth. This conclusion remains valid even after taking into ac- count the influence on development of other factors, such as initial per capita income, education levels or other economic, political or demographic variables. Such empirical studies are no doubt open to question not only because of shortcomings in the quality of information, but also because of the possibility that the relationship between growth and equality over time by countries is different from what can be seen in these cross-sec-
Box 1.4. Inequality Hinders Growth: Summary of the Empirical Evidence
tional comparisons between countries (Forbes, 1997). Neverthe- less, the empirical studies that have focused on the channels connecting distribution with growth also support the conclu- sion that these variables tend to be mutually reinforcing rather than in conflict. Econometric models that simply include income distri- bution as an extra variable explaining growth do not shed light on the channels that can determine this relationship. But sev- eral studies have tried to fill this vacuum. Perotti (1996) shows that joint decisions made by families regarding education and number of children may perpetuate a society’s high inequality and low growth. Another possible channel is political instabil- ity: poor income distribution often goes hand in hand with po- litical unrest that affects investment and growth (Alesina and Perotti, 1993 and 1994; Perotti, 1996). On a worldwide scale there is little evidence for the thesis of the “median voter,” namely that poor distribution leads to higher tax levels that hinder growth. Hence, even though the empirical evidence is not that of complete consensus, it does tend to favor those hypotheses according to which good income distribution tends to stimulate economic growth.
(^8) These surveys have been carried out annually in 17 countries since 1996 by Latinobarómetro, a private, nonpartisan and independent agency. Presented here are results for those countries that have comparable income distribution statistics.
Recent public opinion surveys in Latin America in-
dicate that income inequality can in fact change the way
democratic institutions operate.^8 Income inequality can
weaken acceptance of democratic institutions and prin-
ciples. Where income inequality is less pronounced, as
in Uruguay or Costa Rica, a high proportion of the popu-
lation believes that “democracy is preferable to any other
kind of government” (Figure 1.22). In the more unequal
countries there is a greater tendency to accept authori-
tarian governments, and more people think that “it makes
no difference whether a regime is democratic or non-
democratic.” There also tends to be less trust in these
countries in the institutions proper to a democracy, such
as government, civil service, political parties, the legisla-
ture, large companies and business associations.
The lack of confidence in democracy and its insti-
tutions associated with poor income distribution in Latin
America could have implications for national political sys-
tems. In fragmented societies where institutional confi-
dence is low, the process of aggregating individual pref-
erences is more complex and inexact, and conflicts over
distributing public resources are more intense. Moreover,
the social and economic integration of different groups
is difficult and the state apparatus is more susceptible to
the influence of interest groups, corruption and ineffi-
ciencies, all of which feed inequality.
24 CHAPTER 1
Appendix 1.1. Main Features of Household Surveys Used in this Study
Sample size
Income
Reference
Property
Capital
Non-
Imputed
Country
Year
Name of the survey
Coverage
month
Households
Individuals
Labor
rent
rent
Transfers
monetary
Rent
1
Argentina
80
Encuesta Permanente de Hogares
Greater Buenos Aires
October
3,
11,
X^
aX
aX
X^
na
na
96
Encuesta Permanente de Hogares
Greater Buenos Aires
April and May
3,
11,
X^
bX
bX
bX
na
na
2
Bolivia
86
Encuesta Permanente de Hogares
Urban
1986
2,
12,
X^
X^
na
na
na
na
95
Encuesta Integrada de Hogares
Urban
June
5,
25,
X^
X^
X^
X^
na
na
3
Brazil
81
Pesquisa Nacional por Amostra
National
September
103,
482,
X^
X^
X^
X^
na
na
de Domicilios
95
Pesquisa Nacional por Amostra
National
September
85,
334,
X^
X^
X^
X^
na
na
de Domicilios
4
Chile
94
Encuesta de Caracterización
National
November
45,
178,
X^
X^
X^
X^
X^
X
Socioeconómica Nacional
and December
5
Costa Rica
81
Encuesta Nacional de Hogares
National
July
6,
22,
X^
na
na
na
na
na
95
Encuesta de Hogares
National
July
9,
40,
X^
bX
bX
bX
na
na
de Propósitos Múltiples
6
Ecuador
95
Encuesta de Condiciones de Vida
Nacional
August
5,
26,
X^
X^
X^
X^
X^
na
to November
7
El Salvador
95
Encuesta de Hogares
National
1995
8,
40,
X^
bX
bX
bX
X^
X
de Propósitos Múltiples
8
Honduras
89
Encuesta Permanente de Hogares
National
September
8,
46,
X^
na
na
na
na
na
de Propósitos Múltiples
96
Encuesta Permanente de Hogares
National
September
6,
33,
X^
na
na
na
na
na
de Propósitos Múltiples
9
Mexico
84
Encuesta Nacional de Ingreso Gasto
National
Third quarter
4,
23,
X^
X^
X^
X^
X^
X
de los Hogares
94
Encuesta Nacional de Ingreso Gasto
National
Third quarter
12,
60,
X^
X^
X^
X^
X^
X
de los Hogares
10
Panama
95
Encuesta Continua de Hogares
National
August
9,
40,
X^
aX
aX
X^
na
na
11
Paraguay
95
Encuesta de Hogares - Mano de Obra
National
August
4,
21,
X^
X^
X^
X^
na
na
to November
12
Peru
85-
Encuesta Nacional de Hogares
National
July 1985
4,
26,
X^
X^
X^
X^
X^
na
sobre Medición de Niveles de Vida
to July 1986
96
Encuesta Nacional de Hogares
National
Fourth quarter
16,
88,
X^
aX
aX
X^
X^
na
sobre Niveles de Vida y Pobreza
13
Uruguay
81
Encuesta Nacional de Hogares
Urban
Second Semester
9,
32,
X^
X^
X^
X^
X^
X
95
Encuesta Continua de Hogares
Urban
1995
20,
64,
X^
X^
X^
X^
X^
X
14
Venezuela
81
Encuesta de Hogares por Muestra
National
Second Semester
45,
239,
X^
na
na
na
na
na
95
Encuesta de Hogares por Muestreo
National
Second Semester
16,
92,
X^
bX
bX
bX
na
na
a^ Cannot separate between property and capital rent.b^ Cannot separate between property rent, capital rent, and transfers.
Appendix Table 1.2.II. Demographics
A. Average size of household by income level
Deciles
(^1) The surveys for Argentina include only Greater Buenos Aires.
Appendix Table 1.2.III. Education
A. Average years of education for 25 year olds by income level
Deciles
B. Primary Completion Rates for 20-25 year olds by income level
C. Secondary completion rates for 20-25 year olds by income level
(^1) The surveys for Argentina include only Greater Buenos Aires.
Magnitude of Inequalities 29
REFERENCES
Aghion, P., and P. Bolton. 1992. Distribution and Growth in Models of Imperfect Capital Markets. European Economic Review 36. __________. 1997. A Theory of Trickle-down Growth and Develop- ment. Review of Economic Studies 64(2) April: 151-72. Alesina, A., and R. Perotti. 1993. The Political Economy of Growth: A Critical Survey of Recent Literature and Some Results. World Bank. Mimeo. Alesina, A., and D. Rodrik. 1992. Income Distribution and Economic Growth: A Simple Theory and Empirical Evidence. In A. Cukierman, S. Hercovitz, and L. Leiderman, eds., The Political Economy of Business Cycles and Growth. Cambridge, MA: MIT Press. __________. 1994. Distributive Politics and Economic Growth: A Criti- cal Survey of the Recent Literature. The World Bank Economic Review 8(3). Atkinson, A. B. 1997. Bringing Income Distribution in From the Cold. The Economic Journal (March): 297-321. Atkinson, A. B., and J. Micklewright. 1992. Economic Transformation in Eastern Europe and the Distribution of Income. Cambridge: Cam- bridge University Press. Banerjee, A., and A. Newman. 1991. Risk Bearing and the Theory of Income Distribution. Review of Economic Studies 58: 211-35. Benabou, R. 1996. Inequality and Growth. Paper presented at Elev- enth Annual Macroeconomics Conference, Cambridge, MA. Benhabib, J., and A. Rustichini. 1996. Social Conflict, Growth and Income Distribution. Journal of Economic Growth 1(1): 125-42. Birdsall, N., and J.L. Londoño.1997. Asset Inequality Does Matter: Lessons from Latin America. American Economic Review 87(2) May. __________. 1998. No Tradeoff: Efficient Growth Via More Equal Human Capital Accumulation in Latin America. In Beyond Tradeoffs: Market Reforms and Equitable Growth in Latin America, Nancy Birdsall, Carol Graham and Richard Sabot, eds. Wash- ington, D.C.: The Brookings Institution and the Inter-Ameri- can Development Bank. Birdsall, N., D. Ross, and R. Sabot. 1995. Inequality and Growth Re- considered. World Bank Economic Review 9(3) September. Clarke, G. 1992. More Evidence on Income Distribution and Growth. World Bank Working Paper 1064. Dahan, M., and D. Tsiddon. 1998. Demographic Transition, Income Distribution and Economic Growth. Journal of Economic Growth. Forthcoming. De Gregorio, J., and S. Kim. 1994. Credit Markets with Differences in Abili- ties: Education, Distribution and Growth. IMF Working Paper 94/97. Deininger, K., and L. Squire. 1996a. New Ways of Looking at Old Is- sues: Inequality and Growth. World Bank. Unpublished.
__________. 1996b. A New Data Set Measuring Income Inequality. World Bank Economic Review 10(3), September: 565-91. Forbes J., K. 1997. A Reassessment of the Effect of Inequality on Growth. Massachusetts Institute of Technology. October. Un- published. Galor, O., and J. Zeira. 1993. Income Distribution and Macroeconom- ics. Review of Economic Studies 60. Gottschalk, P., and T. Smeeding. 1997. Cross-National Comparisons of Earnings and Income Inequality. Journal of Economic Litera- ture 35 (June): 633-87. Londoño, J. L., and M. Székely. 1997. Persistent Poverty and Excess In- equality: Latin America 1970-1995. Office of the Chief Econo- mist Working Paper Series No. 357, Inter-American Develop- ment Bank. Lustig, N. , and R. Deutch. 1998. The Inter-American Development Bank and Poverty Reduction: An Overview. Sustainable De- velopment Department, Inter-American Development Bank. Mimeo. Morley, Samuel. 1998. Poverty During Recovery and Reform in Latin America: 1985-1995. Inter-American Development Bank. Un- published. Perotti, R. 1993. Political Equilibrium, Income Distribution and Growth. Review of Economic Studies 60: 755-76. __________. 1996. Growth, Income Distribution and Democracy. Jour- nal of Economic Growth 1 (June): 149-87. Persson, T., and G. Tabellini. 1991. Is Inequality Harmful for Growth? Theory and Evidence. NBER Working Paper No. 3599. National Bureau of Economic Research, Cambridge, MA. __________. 1992. Growth, Distribution and Politics. European Eco- nomic Review 36. __________. 1994. Is Inequality Harmful for Growth? American Eco- nomic Review 84(3). Piketty, T. 1997. The Dynamics of the Wealth Distribution and the Interest Rate with Credit Rationing. Review of Economic Studies 64(2) April: 173-89. Solimano, Andrés. 1998. The End of Hard Choices? Revisiting the Relationship Between Income Distribution and Growth. In Solimano, A., ed., Social Inequality Values, Growth and the State. University of Michigan Press. Forthcoming.
