Does work-related training reduce the discrepancy between function requirements and competencies?

The issue of lifelong learning is high on the political agenda. However, despite this political interest and the large economic literature on human capital, the impact of work-related training on the discrepancy between function requirements and the skills of the employee has been ignored. In this paper we use an ordered probit model to analyze the perceived change in discrepancy. Based on the bi-annual OSA panel from 1998 till 2002 for The Netherlands, we show that taking a work-related course decreases the discrepancy significantly. We correct for the endogeneity between the decision to take a course and the change in discrepancy and we argue that ignoring the selective decision to take a course leads to misleading conclusions about the effect of these courses on the change in discrepancy. Some respondents of the OSA-panel drop out between two waves. To correct for the possibility of selective attrition we develop an Inverse Probability Weight (IPW) estimation method for the ordered probit with an endogenous binary regressor. From the implied marginal effects of the IPW estimation we conclude that taking a course increases the probability to change the fit between skills and function requirements from Bad to Good with 16~percent-point.