The evaluation of training programs especially of active labour market programs has attracted a lot of attention the last coupl e of years. In particular, analyses of on-the-job training have come up with encouraging results regarding job stability and reemployment chances after a preceding job loss. The study at hand is the first one to scrutinize the influence of on-the-job-training on the labour market outcome of elder employees by estimating its effect on the age
structure of the establishment. The empirical analyses rely on the IAB firm panel which has been augmented by information on the age struct ure from individual data of the Linked Employer Employee Data Set LIAB. To estimate the causal effect, matching estimators have been applied to control for endogeneity of firm s' decision to provide further on-the-job training. The expected positive effects of on-the-j ob training on the share of elder employees could not be confirmed in this study. Neither a variation of the timing of effects nor of the type of training yielded any significant positive estimation of the effect. The results are robust against different specifications of the matching estimator