Seminar: Mass layoffs and early retirement: Employment losses by skill profiles

A large and growing literature documents the biasedness of shocks from ongoing technological change in the labor market towards particular groups of workers (Acemoglu and Autor, 2011). Yet, little is known about the adjustment process of workers. Meanwhile, the size and timing of the adjustment costs play an important role for policy that aims to account for the distributional impact on individuals. Many general labor market policies are already in place that aim to mitigate the effects of negative shocks, e.g. general unemployment insurance. In the occurrence of mass layoffs, multiple European countries include voluntary early retirement in the policy options for displaced workers. However, policies like early retirement often involve crude measures to determine eligibility, for example the age of the worker. We investigate a case study of a car manufacturer closure that allows us to estimate how costly age-based early retirement is in terms of re-employment losses. The sharp agebased criterion allows us to exploit locally randomized variation in the eligibility for early retirement by means of a regression discontinuity design. Our analysis shows that take-up of this scheme was very large. Moreover, the loss in re-employment rates measured around the age cut-off is estimated to be 42.9%-points. Importantly, we document an interaction with ongoing technological change by differentiation between worker skill types. Workers of all skill types equally opt into the early retirement program. However, routine and low-educated workers have lower re-employment rates also when falling below the eligibility threshold, in line with the hypothesis of routine-biased technical change (Goos et al., 2014). This leads to a lower estimated loss of employment for these workers. Lastly, we provide suggestive evidence on the role for search motivation in our results. This points towards large discouragement among workers early on in the process, which propagates over the period that we observe.