"It is generally found that overeducated (undereducated) workers earn less (more) than adequately educated workers with similar years of education. We investigate the importance of ability and measurement error bias for these outcomes by applying a fixed-effects instrumental variable approach on data for Flemish young workers. This approach results in substantially higher overeducation penalties and undereducation bonusses than a standard random effects approach. This suggests that the upward bias resulting from unobserved worker heterogeneity is more than compensated by the negative bias resulting from measurement error. Further, we also find some evidence on heterogeneous effects of mismatches accross job levels and years of experience."

Support: Paper