The distribution of the gender wage gap in Italy: does education matter?

In this paper we evaluate the wage gap due to differences in rewards to characteristics by studying the entire distribution of the individual unexplained wage gap. We use quantile regressions and an adaptation of the procedure suggested by Machado and Mata (2005) to derive the marginal distributions of predicted and counterfactual female wages. Then, we estimate probability distributions of unexplained wage gap conditional to observed characteristics. The methodologyallows to evaluate the probability of women with different characteristics to experience any level ofdiscrimination. The main focus of the paper is on the relationship between human capital characteristics and outcomes in differences in pay. In particular, we focus our attention on differenteducational levels. Under the hypothesis that women invest in education to signal their productivity, we should detect a lower wage gap −due to differences in rewards to characteristics− among higheducated females. Our analysis suggests that education can be a good signal but not for all females.We also show that education interacts differently with other human capital characteristics, such as general experience acquired in the labour market. The analysis is carried out on Italian data drawnfrom the last available cross-section of the European Community Household Panel (2001