This paper studies the effect of employer-provided training on the probability of subsequent job exit. Empirical evidence usually shows that the probability of receiving training by the employer is higher among those employees with the lowest expected rates of turnover. Therefore, it seems that firms provide training selectively. In this paper, we address the empirical question of to what extent this endogeneity problem leads to a spurious correlation between training receipt and job mobility. Using Spanish Data from the European Community Household Panel, we provide estimates that ignore the selection bias and compare the results with the ones obtained when correcting for the possible nonrandom selection between trainees and non-trainees. Overall, our results show that there is a negative correlation between on the job training and job mobility, but only for fired workers, and not for voluntary movers. Nonetheless, once the endogeneity problem is accounted, the negative effect becomes statistically nonsignificant for all types of movers.

Key words: on the job training, turnover, job mobility.
JEL Classification: M53.