We show how significant may be the difference in the estimated returns to education in Poland conditional on the measure of wages used and the estimation approach applied. Combining information from two different Polish surveys from 2005 and taking advantage of the Polish microsimulation model (SIMPL) we demonstrate how different the results can be depending on whether we use net or gross, and monthly or hourly wages, and show how important selection correction is for the conclusion. While there are several papers examining the wage equation in Poland, so far none of them has provided a comprehensive analysis of the effects of using different methods and the issue of selection-correction in the estimation of the wage equation in Poland has not been examined in detail. Annual rates of return to university education for men vary from 6.7% to 9.7% and for women from 8.0% to 13.4% when we compare results using net monthly wages without correcting for labor market selection to those from a selection corrected specification using gross hourly wages. We also demonstrate that simple linear estimation performs relatively well for men in comparison to our preferred selection corrected estimation, while using family demographics as exclusion restrictions seems to be the second best in the case of the wage equation estimation for women.
Returns to education wage equation selection models instrumental variables