How do employment effects of job creation schemes differ with respect to the foregoing unemployment duration?

Based on new administrative data for Germany covering entrances into job creation schemes between July 2000 and May 2001, we evaluate the effects of this active labour market policy programme on the employability of the participating individuals. The programme effects are estimated considering the timing of treatment in the individual unemployment spell. Applying propensity score matching in a dynamic setting where the time until treatment in the unemployment spell is stratified into quarters, regional (East and West Germany) as well as gender differences are considered in the estimation. As matching is concerned with selection on observables only, we test the robustness of the estimates against possible unobserved influences. The results in terms of employment present a mixed picture. For West Germany, most of the estimates are insignificant at the end of the observation period, but positive exceptions are found for persons starting in the fifth or ninth quarter of the unemployment spell. For East Germany, none of the groups experiences an improvement of the labour market situation. Instead, the majority of the estimates establish negative employment effects until the end of the observation period (30 months after start of programmes). Hence, job creation schemes decrease the employment chances of the participating individuals.