University drop-out. The case of Italy

We combine individual and aggregate-level level data on educational attainment to study the determinants of university drop-out in Italy, one of the worst performers among developed countries. Based on detailed information on a representative sample of secondary school graduates and on local university supply we first show that family and educational background are relevant determinants of continuation probability. In particular, our results show that accounting for enrollment-induced sample selection significantly enhances the estimated coefficients with respect to standard probit analysis. We then combine our estimates with data on family and educational backgrounds of secondary school graduates in comparable European countries and find that differences in endowments only explain a minor fraction of the observed cross-country gap in students� attainments.