One of the most important policy goals in industrialized countries is to increase the skill level of the labor force by life-long-learning strategies. In this paper our aim is to explain to what extent the variation in training investments is determined either by (observed and unobserved) heterogeneity of firms or of workers, hence we put a new perspective on the determinants of training. Rather than analyzing single determinants or groups of variables, we decompose the variation into a worker-specific and a firm-specific part and show how much of the unexplained variation is independent of both. Our results show that both firm-, job- and worker-level heterogeneity explains training participation and that firm heterogeneity is far less important compared to the others. Also interesting, is the finding that a large part of the overall variance is not driven by firm- or worker heterogeneity, hence training participation seems to be to some extent an unexplained event which happens by chance.


Steffes, Susanne
Warnke, Arne Jonas


Human Capital, Training, Linked-Employer-Employee Data (LEE), Decomposition, Unobserved Heterogeneity