Annual Report 2018

34 Economics FAMILY ORGANIZATION AND INCOME INEQUALITY IN BRAZIL The analysis of demographic census data from 1970 to 2010 reveals a pattern of increasing marriage between couples with similar characteristics during this period. The increase of selective marriages, in principle, should not diminish income inequality. Nevertheless, counterfactual exercises reveal that the improved income distribution in this period could have been even greater had this marriage trend not occurred. OBJECTIVE • To investigate the evolution of the marriage market and its impact on income distribution in Brazil. RESEARCH METHOD • • • To verify the number of selective marriages, Brazil's socioeconomic classes were divided into four educational levels - incomplete primary school, complete primary school, com- plete secondary school and university degree - and the Kendall correlation coefficient was estimated between the educational levels of the husband and wife for each census sample studied (1970, 1980, 1991, 2000 and 2010). To verify how the marriage market trend affects income inequality, statistics were ana- lyzed on the income of married people at the different schooling levels. To check whether Brazil's income distribution improved during the period, the GINI index was computed and Lorenz curves were constructed. To examine what would have happened regarding income inequality in Brazil without this marriage market phenomenon, some counterfactual experiments were performed. First, the effect was measured on the GINI index assuming that all marriages had occurred at random in 1970 and 2010. To distinguish the effects on income distribution caused by changes in the educational levels of men and women and by the greater tendency for selective marriages, an ex- ercise was performed in which the distributions of marriages between 1970 and 2010 were switched. Another experiment analyzed what would have occurred with the income distributions in 1970 and 2010 when exchanging the percentage of unmarried people between these two years, i.e., assuming the size of the unmarried cohort of 1970 was equal to that of 2010 and vice versa. These counterfactual exercises only make sense and produce an effect on the income distribution if the participation of women in the labor market is growing. Based on this, an exercise was conducted assuming that all marriages occurred randomly in 1970 and 2010, but exchanging the participation of married women in the labor force between 1970 and 2010. RESULTS The analysis of the percentage of marriages between people of the same educational level, taking into account that the marginal distributions of men and women among the different educational levels changed over the years reveals an increase in the proportion of selective marriages in Brazil. Despite the increase in the number of selective marriages in Brazil, at first there was im- provement in income inequality, a finding that runs counter to observations in developed countries. However, Brazilian income inequality would have declined even more if this conjugal diversity had not decreased during the period studied. There are strong indications that the presence of married women in the labor force im- pacts income distribution. Besides, in the counterfactual exercise involving exchange of the number of single people between 1970 and 2010, it was possible to note that a larger number of marriages helped to improve Brazil's income distribution, although this effect I was not very substantial. CONTRIBUTIONS OF THE STUDY According to the World Bank, Brazil is one of the ten countries with the greatest income inequality in the world. Understanding the determinants of this high inequality is essen- tial. Although many studies have been conducted in this direction, this project is the first to examine the impact of family organization on income inequality in the country. The methodology employed, recognized internationally, had never been applied to the Brazilian case. APPLICATIONS OF THE RESULTS AND POSSIBLE EXTENSIONS OF THE STUDY • The literature on questions of selective marriage and income distribution has only an- alyzed developed countries. It would be interesting to study these questions in other emerging countries and compare them with Brazil. 1 The Lorenz curve is a graph used to represent the relative distribution of a variable in a determined domain. The domain can be the set of people of a region or country, for example, while the variable of interest can be per capita income, among many others. AUTHOR: Cézar Augusto Ramos Santos. RESEARCHER: Luciene Torres de Mello Pereira. ORGANIZATION: EPGE Brazilian School of Economics and Finance (EPGE). SUPPORT: Applied Research Fund (FPA FGV). Annual Report 2018 35 RESEARCH

View entire presentation