Estimation of educational returns using university and labour market administrative archives

This study estimates the rate of return to education for University of Milan (Italy) graduates who were active in the labour market during the period 2003-05, using official administrative data (University archives, Regional Labour market archive and Italian National Internal Revenue Service archive). The rate of return is measured in terms of differences in wage rates associated with differences in education. Both the ‘years of schooling completed’ and the ‘highest qualification obtained’ dimensions of education are considered. Methodologically, we propose a longitudinal extension of the Correlated Random Coefficient Model that, in contrast to alternative approaches for estimating educational returns, permits simultaneous evaluation of education effects on the earnings of graduates in both a cross-sectional (income differences between years or levels of education in 2005), and in a longitudinal framework (differences in income growth rates during the period 2003-05, between years or levels of education). Furthermore, the problem of self-selection (non-randomness of income earners), as well as education endogeneity bias, is taken into account. Empirical results, based on three institutional administrative archives, demonstrate that workers sort themselves into higher paying work experiences and income growth trajectories, while also providing strong evidence for a positive ability bias. Secondly, cross-section education returns confirm that graduates receive an income advantage in proportion to the educational level achieved, whereas longitudinal returns do not confirm this finding.