Returns to education and gender wage gap across quantiles in Italy

The quantile regression approaches are implemented to analyze the characteristics of Italian data on earnings at various quantiles. A changing coefficients pattern across quantiles due to heteroskedasticity is uncovered, which exhibits varying returns to education along the wage distribution. Next, by accounting for endogeneity in education, wider returns at all quantiles are disclosed. Finally, a quantile regression decomposition approach shows the presence of non-linearity. The latter displays that higher education grants higher return at all quantiles, thus implying non-linear returns throughout the entire pattern of the earning distribution. All the above technical features bear consequences on the results and uncover different sources of inequality in returns to education.