Understanding the GWG differential between public and private sector in Italy: A quantile approach

The focus of this paper is the analysis of the gender wage gap (GWG) in the Italian public sector from 2005 - 2016 and the decomposition of its determinants. Our source for is the Italian Institute for the Development of Vocational Training for Workers: Participation, Labor, Unemployment Survey. For this task we first consider the public-private pay determination by gender. The analysis of this differential has shown the wage gap tends to be higher among woman and typically narrows as one moves up earning distributions. The difference in pay is usually lower in the public compared to the private sector. Because of the more standardized career path in the public sector and the different hiring selection methods between them, the GWG should be lower in the private sector because the so-called discrimination components should be counteracted.

However, the analysis does not completely support what we were expecting. While we find higher premia for female workers compared to their male counterparts, the public sector wage gap is accounted for completely by differences in returns at the beginning of the income distribution while this component vanishes in the upper part of the distribution. When we analyze the GWG within sectors, the results are in line with those highlighted in the literature. We find a lower level of the GWG in the public sector with respect to the private one. The GWG increases along the wage distribution in both sectors. The Oaxaca-Blinder decomposition shows that the unexplained component mostly exceeds the explained part and the distance grows as the wage increases. This pattern is much more evident in the public sector where we find evidence of a glass ceiling mechanism in action.

The results change dramatically when we take into account the unobserved individual-specic heterogeneity. In order to assess how the GWG varies across the wage distribution we propose a two step procedure for computing the Oaxaca-Blinder decomposition. First, we estimate the public GWG by means of Canay (2011) approach and then we run the Machado and Mata (2005) decomposition for quantile regression. We find a sharp decrease in the public premium which suggests that there is positive selection in the public sector, especially for females. Second, when we look at the GWG within sectors, the evidence of a glass ceiling in the public sector vanishes. However, in both sectors there is a significant unexplained GWG almost stable across the distribution.