The distribution of male wages in Brazil

This paper examines the evolution of wage inequality in Brazil in the 1980s and 1990s . It tries to investigate the role played by changing economic returns to education and to experience over this period together with the evolution of within-group inequality. It applies a quantile regression approach on grouped data to the Brazilian case. Results using repeated cross-sections of a Brazilian annual household survey indicate that : i) Male wage dispersion remained basically constant overall in the 1980's and 1990' s but has increased substantially within education and age groups. ii) Returns to experience increased significantly over this period, with the rise concentrated on the iliterate/primary school group iii) Returns to college education have risen over time, whereas return to intermediate and high-school education have fallen iv) The apparent rise in within-group inequality seems to be the result of a fall in real wages, since the difference in wage leveIs has dec1ined substantially over the period, especially within the high-educated sample. v) Returns to experience rise with education. vi) Returns to education rise over the life-cycle. vii) Wage inequality increases over the life-cycle. The next step i~ this research will try to conciliate all these stylised facts.

Is child labor harmful? the impact of working as a child on adult earnings

This paper explores the question: is working as young laborer harmful to an individual in terms of adult outcomes in income? This question is explored through the utilization of a unique set of instruments that control for the decision to work as a child and the decision of how much schooling to acquire. These instruments are combined with two large household survey data sets from Brazil that include retrospective information on the child labor and schooling of working-age adults: the 1988 and 1996 PNAD. Estimations of the reduced form earnings model are performed first by using OLS without controlling for the potential endogeneity of child labor and schooling, and then by using a GMM estimation of instrumental variables models that include the set of instruments for child labor and schooling. The findings of the empirical investigations show that child labor has large negative impact on adult earnings for both male and female children even when controlling for schooling. In addition, the negative impact of starting to work as a child reverses at around age 14. Finally, different child labor activities are examined to determine if some are beneficial while others harmful with the finding that working in agriculture as a child appears to have no negative impact over and above the loss of education.