Bulmer-Thomas V., Coatsworth J., Cortes-Conde R. The Cambridge Economic History of Latin America: Volume 2, The Long Twentieth Century
Подождите немного. Документ загружается.
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
602 Miguel Sz
´
ekely and Andr
´
es Montes
Latin American countries.
13
Arecent paper by Lundberg and Squire argues
that changes in GDP and in income inequality are jointly determined and
should therefore be examined in a system of simultaneous equations in
which the direct relationship between these two variables is of secondary
interest.
14
Therefore, after considerable controversy and empirical testing, the jury
is still out on the relationship between development levels and inequality as
postulated by Kuznets. Many scholars have devoted their efforts to trying to
either prove or disqualify this hypothesis, opening the door for many ques-
tions. Given the scarcity of evidence, the only verifiable – and alarming –
conclusion is that Latin America seems to have remained on the top part
of the curve for many years already, without showing any decline in the
inequality trend, but rather a constant and upward-rising trend toward
more poverty and inequality.
poverty, inequality, and growth
As for poverty, the straightforward explanation is that its evolution depends
mainly on economic growth.
15
To engage in this analysis, first one must
identify who counts as belonging to the group of “poor” and then deter-
mine how to measure the incomes of that group to track changes in their
conditions over time.
One approach widely followed for defining the poor is a relative defini-
tion, such as all people in the lowest quintile or decile. The central question
this approach seeks to address is the nature of the relationship between eco-
nomic growth and growth in the income standard of the poor, and whether
the growth elasticity of this income standard exceeds, equals, or falls below
unity. The earlier papers in this strand of literature – including Adelman
and Morris; Ahluwalia; and Ahluwalia, Carter, and Chenery – were
13
Robert Barro, “Inequality and Growth in a Panel of Countries” (Mimeograph, Harvard University,
1999); Alain De Janvry and Elizabeth Sadoulet, “Growth, Poverty and Inequality in Latin America:
A Causal Analysis, 1970–1994,” Review of Income and Wealth 46, 3 (2000): 267–87;Samuel Morley,
La distribuci
´
on del ingreso en Am
´
erica Latina y el Caribe (Mexico City, 2000).
14
Mattias Lundberg and Lyn Squire, “The Simultaneous Evolution of Growth and Inequality,”
(Mimeograph, Washington, DC, 2000).
15
Many other factors have been considered to be of influence regarding poverty and inequality in
Latin America. This includes macroeconomic fluctuations such as high inflation periods, as well as
structural and demographical trends like migration. Although these trends have certainly influenced
the evolution of poverty in the region, the main objective of the current study is to first observe the
relationship between economic growth and poverty alleviation and, second, to study the development
of physical and human assets and its incidence in the well-being of Latin America.
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
Poverty and Inequality 603
primarily interested in the growth–inequality relationship (with one
inequality measure being the income share of the poor group), but they also
asked whether the poorest 20 percent of the population shared the benefits
of growth proportionally.
16
These authors conclude that the income share
of the poor tends to decline in the early stages of development but increases
in the long run.
This approach has received renewed attention recently. Romer and
Gugerty, Gallup, and Dollar and Kraay state that the growth elasticity
of the incomes of individuals in the bottom quintile is essentially equal
to one.
17
Timmer obtains a more modest elasticity of around 0.9.
18
Even
though these studies use the same data and similar econometric techniques,
they disagree on the issue of whether growth leads to a proportional increase
in the income of the poor – in other words, if the gains for the poor are
smaller than those of other groups.
The second approach to poverty and inequality measurement tracks
income poverty levels using an absolute poverty line and a standard poverty
measure. Ravallion, Ravallion and Chen, and Bruno employ absolute
poverty lines of one and two dollars a day to identify the poor and then
aggregate, using the most common measures of poverty, the head count
ratio and the per capita poverty gap.
19
These studies find that the growth
elasticity of the head count ratio is normally below a minus-two level. In
other words, when average income increases by 1 percent, the proportion
of poor declines by more than 2 percent. Hence, the results not only show
that the relation between growth and poverty reduction does not function
on a merely one-to-one basis, but they further imply that growth is the
main vehicle for poverty reduction.
Other authors such as Morley, De Janvry and Sadoulet, and Smolensky
also use poverty lines that combine an absolute and a relative compo-
nent, but their elasticities are highly sensitive to the location of the poverty
16
Irma Adelman and Albert Morris, Economic Growth and Social Equity in Developing Countries (Stan-
ford, CA, 1973); Montek Ahluwalia, “Inequality, Poverty and Development,” Journal of Development
Economics 2 (1976): 307–42;Montek Ahluwalia, Nicholas Carter, and Hollis Chenery, “Growth and
PovertyinDeveloping Countries,” Journal of Development Economics 6 (1979): 299–341.
17
Michael Roemer and Mary Kay Gugerty, “Does Economic Growth Reduce Poverty?” (Discussion
Paper no. 4,Harvard Institute for International Development, Harvard University, 1997); John Luke
Gallup et al., “Economic Growth and the Income of the Poor” (Harvard Institute for International
Development, Discussion Paper no. 36,Harvard University, 1999); David Dollar and Aart Kraay
“Growth Is Good for the Poor” (Mimeograph, Washington, DC).
18
Peter Timmer, HowWell Do the Poor Connect to the Growth Process (Cambridge, MA, 1997).
19
Martin Ravallion, “Growth and Poverty: Making Sense of the Current Debate” (Mimeograph, World
Bank, Washington, DC, 2000).
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
604 Miguel Sz
´
ekely and Andr
´
es Montes
line.
20
The growth elasticity of poverty ranges from −2.59 to −0.69,
depending on whether the threshold is established at 50 percent or 100
percent of the average income at the initial period of observation.
Foster and Sz
´
ekely propose an alternative methodology to track low
incomes, based on Atkinson’s family of “equally distributed equivalent
income” functions denominated “general means.”
21
Based on this method-
ology, they estimate the growth elasticity of the general means and use 144
household surveys from twenty countries that cover the last twenty-five
years of the twentieth century. Among other results, they found that poor
individual incomes do not grow on a one-to-one basis with the increases
in mean income. Therefore, their conclusions differ from others that use
per capita income of individuals in the first quintile as an income standard
of the poor and argue that the growth elasticity of this income standard is
unity. Foster and Sz
´
ekely estimate the relationship for a set of Latin Amer-
ican countries and conclude that the elasticity of the income of the poor to
economic growth is also lower than one.
In sum, even though the relationship between growth and poverty is
evidently close, the available evidence indicates that growth itself cannot
fully explain changes in poverty in the region.
some final considerations
From the previous discussion, it is possible to assert that even in cases where
the inequality–development relationship does seem to follow the inverted
U pattern suggested by the main development theories, extreme poverty has
not declined consistently as a natural outcome of the development process,
contrary to expectations. Moreover, the mechanisms through which poverty
was supposed to change were actually only observed during a relatively short
subperiod (1977–84).
In some cases (e.g., Mexico) it can be found that, contrary to what the
theory predicts, the main cause of the change in the standard of living of
the extremely poor after the turning point in the inequality-development
relationship was that the rural–urban gap continued to expand, generating
20
Alain De Janvry and Elizabeth Sadoulet, “Growth, Poverty and Inequality in Latin America: A Causal
Analysis, 1970–1994”; Eugene Smolensky et al., “Growth Inequality and Poverty: A Cautionary
Note,” Review of Income and Wealth Series 40, 2 (1994): 217–22.
21
James E. Foster and Miguel Sz
´
ekely, “Is Economic Growth Good for the Poor? Tracking Low Incomes
Using General Means” (Working Paper no. 453,Inter-American Development Bank, 2001).
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
Poverty and Inequality 605
significant poverty-increasing effects. The distribution of income within
the subgroups, implicitly assumed constant, also changed considerably in
most of the years.
Overall, this implies that, at least in some cases, the results are not in
line with the Kuznets hypothesis, which would have predicted a consistent
improvement in the standard of living of the poor after the turning point.
Even in the context of positive economic growth and declining inequal-
ities, a deterioration in the standard of living of the poorest is observed.
Uncertainty about the accuracy of the Kuznets hypothesis could be well
extrapolated to an entire region such as Latin America, which has borne the
twin burdens of poverty and inequality for a long period of time, despite
extended periods of economic growth.
INCOME-EARNING ASSETS IN
LATIN AMERICA
assets, prices, and returns: an asset-based approach
Traditional explanations arguing that economic growth can explain the
evolution of poverty and inequality in Latin America and the Caribbean
are clearly insufficient for understanding the persistently high levels in the
region. In the spirit of Engermann and Sokoloff, one could argue that the
main factor driving the persistently disparate income levels is the skewed
distribution of factor endowments, or income-earning assets, among indi-
viduals.
One way to pursue this analysis is to acknowledge that the income of
each individual in a society is a function of the combination of four essential
elements: first, the stock of income-earning assets owned by each individual
(classified as either human or physical capital); second, the extent to which
these assets are employed in producing income; third, the market value
of income-earning assets; and fourth, the income received independently
of income-earning assets, which may include transfers, gifts, and bequests,
among others. Consequently, family per capita income can be expressed in
the following equation:
y
i
=
j
i=
i
a=
A
a.i
R
a.i
P
a
+
k
i=
T
i
n
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
606 Miguel Sz
´
ekely and Andr
´
es Montes
where y represents the household per capita income of the individual i ; A
is a variable representing the stock of asset type a,owned by an individual
i ; R is a variable representing the rate at which asset type a is used by
individual i ; and P is the market value per unit of asset type a. The variable
j represents the number of income-earners in the household to which the
individual i belongs, l is the number of different types of assets, and k is the
number of individuals in the household obtaining income from transfers
and bequests, whereas n is the size of the household to which i belongs.
For simplicity, here we classify income-earning assets into human and
physical capital. Human capital assets are the set of skills endowed by
individuals – such as knowledge, capability, or expertise – and the health
that enables them to produce any good or service. The most widely used
proxy for quantifying knowledge is years of formal education. Other types
of skills acquired through experience or training programs are not always
available and are more difficult to measure. Information on labor market
experience is also seldom available. Therefore, the definition of human
capital is restricted to years of schooling and health.
On the other hand, physical capital refers to the monetary value of any
form of financial asset, be it money holdings, property, rents, capital stock
used for production, or any other form of physical capital used to produce
a good or service. These stocks can be used in several ways; they can either
be invested for production or accumulated to function as savings.
In terms of opportunity to use the assets productively, the two clear areas
are employment and investment opportunities. Employment opportuni-
ties refer to the conditions, costs, and incentives in the labor market that
influence the demand for different kinds of labor and the demand for skills.
Investment opportunity comes from the existence of an efficient financial
market that gives access to credit. Credit can be used to create economic
activity and to take advantage of the economic environment to generate
income.
Ownership, or other avenues of access to any of these income-earning
assets, implies that an individual has the potential capacity to generate
income at some point in time, but the income that is actually generated
depends on the use of the asset. For instance, in the case of human capital,
the years of schooling of an individual will only be translated into income
when that person becomes gainfully employed. Regarding physical capital,
it will only produce income when the dividend or return generated by the
asset is made liquid.
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
Poverty and Inequality 607
Tu rning to the third income component, prices and returns for each
income-earning asset, the behavior of prices is somewhat unique because
it is determined by supply and demand, as well as by institutional fac-
tors. Prices and returns are therefore set by the economic system, and the
individual responds to them in the process of deciding whether to seek
income and accumulate a certain type of capital. Unlike the other two
income components, the individual exercises very little control over prices
and returns.
Finally, the fourth element in the income-generating process described
earlier is transfers, bequests, and gifts that a household receives and that are
not directly related to income, prices, or returns.
There is certainly a link between assets, their use, and poverty. As we
show subsequently – for instance, in the case of human capital – the poor
have smaller stocks. They receive the lowest rewards not only for having a
small stock, but also because the returns are nonlinear and increase with the
size of the stock. Finally, because of the low returns, the poor and, specifi-
cally, women ultimately end up using these assets at a much lower rate.
In an attempt to improve our understanding about why poverty and
inequality in Latin America have historically been so high, we explore long-
term trends in the underlying factors determining incomes: assets, use, and
returns. This section examines the process of human and physical capital
accumulation in Latin America in order to assess how the distribution of
these assets evolved in the region during the past century.
human capital accumulation through
schooling attainment
Because Latin America is characterized by increasingly high wage inequal-
ity, it is essential to know how human capital is distributed as well as how its
distribution changes over time. This section establishes some of the more
general trends followed by Latin America throughout the century, exam-
ining changes in both the flows and stocks of schooling. South Korea and
Taiwan are interesting cases for comparison because they are regarded as
having achieved outstanding schooling progress during the twentieth cen-
tury. Comparisons of Taiwan and the Latin American region are presented
subsequently.
Figure 14.5, taken from Behrman, Duryea, and Sz
´
ekely, plots schooling
attainment for Taiwan and the average Latin American country for all
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
608 Miguel Sz
´
ekely and Andr
´
es Montes
4
5
6
7
8
9
10
11
12
Year Dummy Coefficient, FE regression
1930 1935 1940 1945 1950 1955 1960 1965 1970
Year of Birth
Pattern for
Taiwan
LAC pattern
LAC trend
1940-1960
Figure 14.5.Pattern of schooling progress in Taiwan and Latin America.
Source: Jere Behrman, Suzanne Duryea, and Miguel Sz
´
ekely, “Schooling Investments
and Aggregate Conditions: A Household-Survey Approach for Latin America and the
Caribbean” (Working Paper no. 407,Inter-American Development Bank, Washington,
DC, 1999).
cohorts (or groups of individuals) born between 1930 and 1970.
22
The
figure shows that, on average, Latin America and Taiwan had very similar
levels of schooling among cohorts born before 1940, but from that year on,
progress in Taiwan shot upward. Thirty years later, cohorts in Taiwan were
registering attainment levels almost 50 percent greater than the average
Latin American country. The figure also shows the slowdown in Latin
America for the 1960–70 birth cohorts. Cohorts born in those years were
making schooling decisions that coincided with the early years of the debt
crisis in the region. The figure also plots a trend line in Latin America from
22
Household surveys for years around 1995 are used to produce this figure. Therefore, cohorts born in
1930 were about 65 years of age at the time of the survey, whereas those born in 1970 were around the
age of 25. The figure does not include information on more recent cohorts (born after 1970) because
a considerable share of individuals born after 1970 have not yet completed their schooling; hence,
it is not clear what their attainments will be. The Latin America and the Caribbean (LAC) pattern
was obtained by pooling all the information on the average years of schooling by year of birth, for
18 countries, and estimating a country fixed-effects regression using the average schooling of each
cohort as the dependent variable and dummy variables for each year as the right-side variables. The
figure plots the coefficients for the year dummies.
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
Poverty and Inequality 609
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
13.0
1930 1935 1940 1945 1950 1955 1960 1965 1970
Year of Birth
Taiwan
Mexico
Nicaragua
Brazil
Chile
Korea
Average Year of Completed Schooling
Figure 14.6.Mean years of schooling by cohort.
Source: Jere Behrman, Suzanne Duryea, and Miguel Sz
´
ekely, “Schooling Investments
and Aggregate Conditions: A Household-Survey Approach for Latin America and the
Caribbean.”
1940 to 1960.Had the same trend continued for cohorts born after 1960,
the average years of schooling for the last cohort would have been close to
10 years, rather than around 8.5.
Figure 14.6 plots information on cohorts born between 1930 and 1970 for
a selected group of countries: Chile (one of the countries in Latin America
with the highest current schooling levels and second only to Argentina
for the 1970 birth cohort), Mexico (the country in Latin America with
the greatest growth in mean years of schooling between the 1930 and 1970
birth cohorts), Brazil and Nicaragua (two of the countries in Latin America
with the poorest schooling performances), and Korea and Taiwan. All of
these countries display significant improvements in mean schooling for
individuals born between 1940 and 1960, though more for Taiwan than for
the others. Yet in Taiwan and South Korea, schooling increased at faster
rates for persons born after 1960 than did most of Latin America, with the
exceptions of Mexico and the Dominican Republic. For example, Chile
and Taiwan had similar mean schooling for people born between 1950
and 1955, but those born in 1970 in Taiwan have, on average, one more
grade of schooling than their Chilean counterparts. On the other hand,
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
610 Miguel Sz
´
ekely and Andr
´
es Montes
the large differences between Mexico and the Dominican Republic versus
South Korea and Taiwan that can be observed today are not attributable
to greater progress in these two East Asian countries for the most recent
cohorts but rather to their much higher levels at the start of the period
covered.
Not all countries in the Latin American region registered the same spread
of education progress. For example, Honduras reported no increases in
educational attainment for more than a decade for the cohorts born between
1925 and 1940, and it was not until the early 1950s that significant increases
were reported.
In Table 14.4, information on the average years of schooling is presented
for various cohorts. Seventeen countries in Latin America are listed in
increasing order of mean schooling attainment for those born in 1930.
Similar data for South Korea, Taiwan, and the United States are given at
the bottom of the table. This table clearly portrays the upward trend for new
generations of Latin Americans. Although advancements are notable for all
countries, there is considerable disparity among nations. Brazil entered
the century with one of the lowest levels, much like those of Honduras.
Chile and Panama, on the other hand, had educational attainments almost
three times higher than those of Brazil at the turn of the last century.
Notwithstanding the progress reported, the table shows that many countries
have suffered important setbacks regarding education.
On average, there was an increase of 4.6 years of schooling in the seven-
teen Latin American countries from the cohort born in 1930 to the one born
in 1970. The largest increases were in Mexico, the Dominican Republic,
Chile, Ecuador, Bolivia, and Venezuela, all of which boasted a gain of more
than five years during the period. The smallest changes were in Jamaica,
Paraguay, Brazil, and Nicaragua, all with less than four years. In contrast,
South Korea and Taiwan made impressive strides in schooling attainment,
boosting average years of education by 6.8 and 6.5 years, respectively, dur-
ing the same period. Recent generations in these nations are approaching
the schooling attainment levels of the United States.
Educational attainment levels in Latin America improved by only one
year per decade in the region. Given this sluggish growth, the supply of
the highest skills has also been quite slow to increase and has not been
able to keep pace with the growing demand. This predicament has been
a common one across the region and, as a result, the wages of highly
educated individuals relative to unskilled employees have risen (see the
Cambridge Histories Online © Cambridge University Press, 2008
P1: GDZ
0521812909c14 CB929-Bulmer 052181290 9 October 6, 2005 10:49
Poverty and Inequality 611
Table 14.4. Average years of schooling by birth cohort
Year of birth Change
Country 1930 1940 1950 1960 1970 1930–1950 1950–1970 1930–1970
Honduras 1.43.24.65.66.13.21.44.7
Nicaragua 23.24.35.85.82.21.63.8
El Salvador 2.13.24.15.77 2 2.94.9
Brazil 2.83.65.26.26.72.41.53.9
Mexico 2.94.26.78.29.33.82.66.4
Dominican
Republic
3.24.27 8.69.13.92.15.9
Venezuela 3.25.16.97.98.33.71.45.1
Bolivia 3.34.56.37 8.62.92.35.2
Ecuador 3.94.56.58.59.52.63 5.6
Colombia 3.94.46.27.78.42.32.24.4
Costa Rica 4.35.77.18.88.42.81.34.1
Chile 5.27.18.910.111.13.72.15.8
Panama 5.86.98.810.310.13.11.34.4
Peru 66.37.49.410 1.42.64
Uruguay
∗
6.37.48.810 10.72.51.94.4
Jamaica 6.97.98.39.610.61.42.33.7
Argentina
∗
7.58.310 11 11.32.51.33.8
Average
LAC
4.15.36.98.28.82.71.94.6
Korea 5.37.79.511 12 4.32.56.8
Taiwan 5.85.88.911 12.33.23.36.5
USA 12.312.913.613.313.41.3 −0.21.1
∗
The survey for Argentina includes only Gran Buenos Aires; the survey from Uruguay covers only
urban areas.
Sources: Suzanne Duryea and Miguel Sz
´
ekely, “The Determinants of Schooling in Latin America:
AMicro-Macro Approach” (Working Paper, Inter-American Development Bank, 1999). Data
from Korea were taken from UNESCO Statistical Yearbook,(NewYork,1997).
fourth section), thereby contributing to the ever-widening income gap
between highly and poorly educated individuals.
Schooling progress in Latin America was considerably greater for the
generations born between 1930 and 1950 –again of 2.7 years – than for
those born between 1950 and 1970 –again of 1.9. The deceleration appears
to be steeper in Honduras, Dominican Republic, Venezuela, and Panama,
where progress for cohorts born between 1930 and 1950 was more than 1.5