Identification of the private-public wage gap

We suggest an identification strategy for the private-public sector wage gap to correct for the bias resulting from the heterogeneity of unobservable characteristics between shifters and stayers. The analysis applies a fixed effect difference-in-difference model with event study design to estimate the wage gap. As the parallel wage trend assumption between shifters from the public to the private sector and public sector stayers is rejected, late shifters still in the public sector are used as counterfactual group for early shifters. The estimates are based on rich register data for high-educated workers in Norway 1993-2010. Using this novel identification method, we show that due to positive selection, the private-public wage gap is overestimated by about 20% in the standard model comparing shifters with stayers. In an extension of the analysis, we show that the overestimation is the same for male and female workers and is robust across business cycles, although the size of the wage gap is pro-cyclical.