This paper studies the role of training vouchers and caseworkers in public training programs. Using a rich administrative data set, we apply matching and regression methods to measure the effect of the Hartz reform in Germany, which introduced training vouchers and imposed more selective criteria on participants. Besides estimating the overall reform effect, we isolate the effect induced by changes in the composition of program participants due to stricter selection by the caseworkers (selection effect) from the effect based on the introduction of vouchers (voucher effect). Analyzing the most important type of training in Germany, we find a slightly positive impact of the reform. Our decomposition results suggest that the selection effect is ? if at all ? slightly negative, and that the voucher effect increased both, the employment probability and earnings of the participants.