Dinardo et al applied nonparametric methods to measure

Dinardo et al. (1996) applied nonparametric methods to measure the effects of institutions and labor market factors on the changes of income distribution in the U.S. in the period 1979–1988. They concluded that labor unions and supply/demand shocks are vital to explain the income inequality rise in that period. Besides, there was also evidence that the decline of the minimum wage contributed mostly to such inequality increase, especially due to the gender factor. Based upon the Blinder–Oaxaca decomposition with the quantile regressions procedure, Albrecht et al. (2003) found in the Swedish workers’ income distribution what is conventionally named in literature as the glass ceiling effect. This means that the difference between men\'s and women\'s income is positive, it grows along the distribution and intensifies at the top. The technique proposed by Machado and Mata (2005) has been applied in some recent articles. It utilizes quantile regressions in order to identify and decompose income gaps among distinct categories of works at different points in time. Their aim was to verify the changes in income distribution in Portugal regarding the period 1986–1995, and concluded that the changes of the individuals’ attributes as well as the returns to such attributes do contribute to the increase of income inequality, besides emphasizing that education plays a decisive role to this increase. This technique has been applied worldwide with distinguishing approaches. Rica et al. (2005) found a steep pattern of gender wage gap in Denmark, Sweden, United Kingdom, Italy and Greece in somehow with the flatter evolution of the gap in Spain, where for higher education range there is evidence for glass ceiling effect, while for lower education group the gap is much higher at the bottom than at the top of the distribution. Albrecht et al. (2009) provided greater theoretical support to the methodology of Machado and Mata (2005) by evidencing the asymptotic consistency and normality regarding such technique, besides taking into consideration sample selection adjustments. The empirical application of such technique aimed to test the hiatus between genders among Dutch workers’ income, and concluded for the presence of a glass ceiling effect on that distribution. Using these authors’ methodology, Paredes (2012) also analyzes the gender discrimination in the Chilean economy by grouping up occupations in sets of nine, and identifies that the income hiatus between men and women has been narrowed in categories such as managers, upscale self-employed professionals, and especially in occupations dealing with finances and insurance. Similarly for Brazil and using data from PNAD/1999, Santos and Ribeiro (2006) found in the aggregated data of PNAD/1999 a glass ceiling effect in the female earnings (top of the distribution), but was uncertain about a glass floor effect (bottom of the distribution). Coelho et al. (2010) have also identified such an effect through the earning distribution gap of black and white Brazilian women by performing quantile regression with PNAD/2007 data. Their main finding is that the returns to education are higher at the top of the distribution. This work also follows the Methodology of Machado and Mata (2005) to Brazilian workers’ income distributions, although placental mammals distinguishes from others by considering occupational groups which shall provide valuable information has more accurate results in the labor market. So it will benefit policies that aim to reduce salary inequalities as they might be driven to specific working categories. The first versions of occupational distribution were proposed by Erickson et al. (1979) and improved by Goldthorpe et al. (1987), a scheme internationally known as EGP (first authors’ initials), which is comprised of seven classes. Based on this aggregation, occupations were then grouped in several countries studied by the CASMIN Project (Comparative Analysis of Social Mobility in Industrial Nations), which was coordinated by Goldthorpe in the early 1980s.