Abstract
Using data from the 2010 Graduates Occupational Mobility Survey, this paper examines the wage effects of over-education (OE) among graduates in science, technology, engineering, and mathematics (STEM) disciplines at the early stage of their careers. In the pooled OLS analysis, the negative correlation between OE and wages remains even if we were to estimate an augmented specification where OE is disaggregated according to perceived skill mismatch. However, the pooled OLS estimates are changed dramatically when we control for unobserved individual heterogeneity (i.e., ability) using the panel FE estimation procedure; namely, reduced coefficients in the wage equation compared to pooled OLS estimation.