Revisiting Returns to Education

Abstract

This paper takes advantage of the college enrollment expansion law of 1980 that essentially raised educational attainment for a particular set of sample cohorts. This paper estimates the impact of the quasi-experimental increase in college education on earnings by using the college enrollment expansion law as an instrument for education. This paper's approach shares a similar spirit with Angrist and Krueger (1991) but identification uses a much large-scale supply-side policy change in the face of chronic excess demand for education that substantially raised educational attainment, i.e., a strong instrument. Estimates combined with non-parametric Kernel estimation results represent a local average treatment effect of education around the upper 17% of the applicants. This paper's instrumental variables estimate of the return to education is a bit larger than ordinary least squares estimates and also larger than usual IV estimates reported in the literature, suggesting heterogeneous returns to education.