Decomposing the Gender Wage Differences using Quantile Regressions

Using linked employee-employer data, this paper measures and decomposes the differences in earnings distribution between male and female employees in Germany. I propose to extend the traditional decomposition to disentangle the effect of human capital characteristics and the effect of firm characteristics in explaining the gender wage gap. Three decomposition methods are used: Oaxaca-Blinder decomposition using OLS and quantile regressions as well as the decomposition proposed by Machado and Mata (2002). I find that the difference in the returns to the firm characteristics forms the largest part of the gender wage gap while the difference in returns to the human capital characteristics constitutes the smallest part. Furthermore, the observed gender wage gap and the decomposed parts of them vary across the wage distribution