Avaliações de abordagens metodológicas da equação de rendimentos aplicadas em cenários selecionados no mercado de trabalho

• Ano de defesa: 2013
• Autor(a) principal: Guimarães, Daniel Barboza
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Renda - Distribuição; Transmissão de renda; Mercado de trabalho
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/9488

This study comprises two analyses to investigate some specific scenarios related to Brazilian workers’ earnings. The first analysis focuses on the main determinants of the sons’ and daughter’s earnings, among which the incidence of transmission of parental income. Estimations of earning equations with different methodological procedures allow to capture the transmission effect and the dynamic effects (analysis of overlapping generations), whose empirical findings are based on data from the National Household Sample Survey (PNAD) for the years 1989 and 2009. Methods of Heckman's two-stage and instrumental variables models are used, in order to correct for sample selection bias and endogeneity in the model. In addition, it is applied Klein-Vella method as a corrective procedure to obtain the true average effect, not only an effect for individuals who are truly affected by changes in the instruments. Then, due to the impossibility of obtaining perfect counterfactual from PNAD data, quantile regressions are estimated to increase the reliability of estimates through homogenization of workers in the sample by income ranges. It was found that instrumental variables method presented the worst predictive fitting amongst the methodological procedures. Regarding the other approaches, the estimates did not show any statistical superiority of one over the others for inference purposes. Therefore, based upon anyone of these procedures it could be concluded that the parents’ incomes, especially the mother's ones, play a positive and significant transmission impact on their sons’ and daughter’s earnings, although it has presented a declining tendency over time. The second analysis aims to test earning differences by gender at three regional levels (Brazil, Northeast and South-Southeast) and four occupational categories (all occupations, managers, technicians, general services). Three methodological approaches were evaluated in the estimation of Mincerian earnings equation. Firstly, the OLS and two-stage Heckman methods were performed to correct for selection bias. In order to test the existence of gender discrimination in the labor market it was initially employed the propensity score matching method, and then applied the Oaxaca–Blinder method to measure the magnitude of such discrimination, since it allows decomposing the earnings differences due to individual characteristics (explained) and discrimination (unexplained). Regardless of the method used in the estimations carried out, there is strong evidence toward an earning gap in favor of men, whose superiority ranges from 36.32% (managers in Brazil) to 118.72% (technicians in the Northeast). These differences are due to the presence of discrimination by gender since the effect of discrimination, which varies from 100.5% (services in South-Southeast to 153.8% (technicians in Brazil), is greater than the difference in earnings because women have, on average, acquired attributes (education, for example) higher than those of men. Therefore, if there were no discrimination in the labor market, there is evidence in the sample that men's earnings tend to be lower, on average, than those of women.