Education expansion, college choice and labour market success

We apply the unordered monotonicity setting of Heckman and Pinto (2018) to estimate the distribution of response types and the counterfactual outcomes associated with the choice of a STEM or non-STEM college. Instrumental variation is induced by the proximity to universities offering STEM and/or non-STEM degrees. The empirical analysis uses confidential survey data for Italy, combined with administrative information about the founding dates of all Italian universities and faculties. We find that about 2.4% of individuals in our sample would choose a STEM college if there was one in their province of residence. The corresponding share of non-STEM switchers is 1.1%. We simulate the effects on a number of labour market outcomes of various reforms improving the geographical distribution of colleges and we find sizeable impacts, particularly for women.