Job mobility and employment durations can be explained by different theoretical approaches, such as job matching or human capital theory or dual labor market approaches. These models may, however, apply to different degrees at different durations in the employment spell. Standard empirical techniques, such as hazard rate analysis, cannot deal with this problem. In this paper, we apply censored quantile regression techniques to estimate employment durations of male workers in Germany. Our results give some support to the job matching model: individuals with a high risk of being bad matches exhibit higher exit rates initially, but the effect fades out over time. By contrast, the influence of human capital variables such as education and further training decreases with employment duration, which is inconsistent with the notion of increasing match-specific rents due to human capital accumulation. The results also suggest that the effects of certain labor market institutions, such as works councils, differ markedly between short-term and long-term employment, supporting the view that institutions give rise to dual labor markets.


Job Durations, Mobility, Matching, Human Capital, Quantile Regression