In most countries the number of places at the universities is restricted. This paper estimates the effect of university enrollment constraints in 1982 on years of education and earnings in Sweden 1981-96. The effect on educational attainment is related to labor market performance, to estimate the effect of education on earnings. The variation used is driven by discrete jumps in the admission selection to university. The results show that university enrollment constraints affect educational attainment over the entire period studied. In 1996, admitted applicants in 1982 have about one quarter of a year longer education than screened out applicants. The effect of enrollment constraints changes with time. In the end of the panel, admitted applicants in 1982 are no better off than screened out applicants. The estimated return to education in Sweden is very low, both with least square and instrumental variable techniques.
Education and earnings admission selection regression-discontinuity design long panel