Which Gender Wage Gap Estimates to Trust? A Comparative Analysis

The aim of this paper is to compare estimates of the adjusted wage gap from different methods and sets of conditioning variables. We apply available parametric and non-parametric methods to LFS data from Poland for 2012. While the raw gap amounts to nearly 10 percent of the female wage; the adjusted wage gap estimates range between 15 percent and as much as 23 percent depending on the method and the choice of conditional variables. The differences across conditioning variables within the same method do not exceed 3pp, but including more variables almost universally results in larger estimates of the adjusted wage gaps. Methods that account for common support and selection into employment yielded higher estimates of the adjusted wage gap. While the actual point estimates of adjusted wage gap are slightly different, all of them are roughly twice as high as the raw gap, which corroborates the policy relevance of this methodological study.