are by definition progressive. Furthermore, such contributions should include those made by the state as
an employer. If one were to measure the net payout
by the state, also netting out the contributions made
by the state as an employer, the distributive impact
of pension payments would look much better than
it does in figure 1.
Even here, though, there is a strong case to be
made for expanding the coverage of pension systems.
Retirement benefits in many Latin American countries are related to formal employment, thus severely limiting their progressiveness. A truly progressive
pension system lies in the design of a program
financed by the government that provides pensions
for informal sector workers. 11 Such a pillar would be
highly progressive, as reflected already in those countries that have some basic pensions that are universal
in character or some form of non-contributory pension system (Argentina, Bolivia and Brazil).
Compare the overall contribution of social spending to improving income distribution with the human
development index of the United Nations Development Program and you see that the most effective
form of targeting is, in fact—and, for some, perhaps
ironically—a universal social policy. The two are
Figure 2 Links Between Human Development
and the Distributive Effect of Social Policy
0.0 net Effect of social spending on income Distribution
(points of the gini coefficient)
-0.02
Guatemala
-0.04 Honduras
El Salvador
Peru
Bolivia
Ecuador
Mexico
-0.06
Colombia
Brazil
Panama
-0.08
Costa Rica
Uraguay Chile
Argentina
0.90
-0.10
0.70 0.75 0.80 0.85
Human Development index (education and health)
highly correlated as demonstrated in figure 2 below.
The largest redistributive effect of social spending is
achieved in those countries that had an early development of more universal systems of social policy:
Argentina, Chile, Costa Rica, and Uruguay (Cuba
should be added to this list but is generally excluded
from this type of estimates.) Countries with an intermediate level of development—Brazil, Colombia and
Panama—have intermediate levels of redistribution
associated with social spending, and the lowest level
occurs in countries that have a lower level of development of their social policy instruments: Bolivia, El
Salvador, Honduras, and Guatemala. In three countries—Ecuador, Mexico and Peru—the redistributive
effects of social spending should be higher given their
level of human development.
Targeting alone has a relatively limited impact.
But it can play a subsidiary role in three specific areas.
First, social assistance (conditional subsidies, nutrition programs, pension transfers for poor old people)
can serve as a pillar of a broader universal system.
But even in these cases these programs must aim
for the eventual universal coverage of the targeted
population. Second, targeting can also enhance the
access of the poor to universal social programs. This
is indeed a particular advantage of the recent conditional transfers; they tie the assistance to access of
the population to universal programs of education
and health. And third, targeting can also be used to
differentiate the programs for specific groups of population, particularly indigenous peoples. In the last
two cases, targeting must serve as an instrument of
universalism and not as its substitute.
The Paradox of redisTribu Tion
Universal systems are associated with a better primary distribution of income across
the population. In the industrial countries of
continental Europe more universal welfare systems
have gone hand-in-hand with a better distribution of
income compared to countries that use more means
testing (targeting) in their social policy, such as the
Anglo-Saxon countries. 12 Causality goes both ways in
this case: more equal societies demand more universal systems of social policy, but the latter contribute
in turn to equality. In contrast, the extensive use of
