Wage gender discrimination and segmentation in the Brazilian labor market

Abstract

This study calculates the returns to education and experience, wage gender discrimination and market segmentation, based on the earnings equations coefficients, obtained by using a sample selectivity bias correction model. To this end, a detailed analysis using a multinomial logit model is developed, where the dependent variable takes value 0 if a person is not working, 1 if he or she is in the formal market place, or 2 if the individual is employed in the informal sector. Based on the coefficients estimated in the polychotomus choice model, a lambda variable (inverse of Mill’s ratio) is calculated and used in the wage equation to obtain consistent estimates without sample selectivity bias. Large differences are observed when comparing the returns to education and experience obtained from the coefficients of the earnings equations when both correcting and not correcting for sample selectivity bias. The men's return to education in the informal sector, for example, more than double when excluding lambda. Large gender discrimination was found when the women's average estimated wages is obtained by substituting their characteristics in the men's structure. While the actual average women’s earnings is approximately 25% below the men’s wage, the estimated average women’s earnings surpassed the average men's earnings. Finally, it was observed that 20% of the earnings differentials between formal and informal sectors is attributed to market segmentation.