Research

While social skills have become increasingly important in the labour market, other skills may have lost relevance. Estimating these changes is challenging because skills measured before tertiary education affect wages both directly and indirectly through educational sorting. This paper develops a sequential model with cognitive, social, and diligence skills measured at age 17 to estimate direct and total early-career returns across recent German cohorts, while accounting for unobserved ability. Direct returns to social skills increased by 6 percentage points, whereas total returns to diligence declined. Among individuals with low cognitive skills, returns to diligence fell by 10 percentage points, consistent with high-diligence workers sorting into routine-intensive occupations whose value declined under deroutinisation.

When individuals risk being overeducated for their jobs, returns to education might be lower and heterogeneous. To investigate this, we develop a novel framework that decomposes returns using an expected value conditional on overeducation risks and penalties. We estimate these components using Belgian data and a dynamic model of endogenous educational choices, overeducation, and wages. Overeducated individuals experience a persistent wage penalty. However, as also medium levels of education are associated with an overeducation risk, this risk usually plays a limited and sometimes even positive role in explaining average returns. Additionally, overeducation generates heterogeneous realized returns among Master's graduates.

Returns to higher education vary substantially between and within fields of study. This paper studies whether occupational sorting at labor market entry contributes to this heterogeneity in early-career realized returns. We estimate a generalized sequential Roy model of yearly higher education enrollment choices, skill mismatch at entry, and subsequent labor market outcomes. Identification combines persistent latent heterogeneity with exclusion restrictions based on distance to programs and residual variation in graduation timing. Individuals with a higher probability of enrolling in higher education obtain higher returns, but also experience the largest losses in realized returns when they start in mismatched occupations. In counterfactual decompositions, entry mismatch accounts for a substantial share of low, and in some cases negative, early-career realized wage returns.