Studies of on-the-job training have found that low skilled workers participate less in further training. In this paper, we develop a signalling model of training where training can increase productivity when workers’ prior ability matches the level of the training course. Consequently, employers can use observed training participation as information about the quality of their employees when job performance is unobservable. As a result, programs aimed at low ability workers will be underutilized because of the negative signal participation conveys about their initial ability. We show that offering training has spillover effects: introducing high-level courses to a choice set of only low-level courses increases participation in low-level courses, and improves sorting efficiency. We discuss the implications of these results for optimal training provision, and make several testable predictions for assessing the validity of our model.
|GSBE Research Memoranda
- j24 - "Human Capital; Skills; Occupational Choice; Labor Productivity"
- j45 - Public Sector Labor Markets
- m53 - Personnel Economics: Training
- human capital