Are the effects of training programmes in Germany sensitive to the choice and measurement of labour market outcomes?

We reconsider the evidence of Lechner, Miquel and Wunsch (2004, 2005) on the effectiveness of training programmes for the unemployed conducted in East and West Germany in the period 1993-1994 by investigating whether, and if so, how overall policy conclusions depend on the particular choice of the outcome variable. We find that different measures of employment and earnings provide very similar results. In contrast, considering unemployment as an outcome measure shows rather different results as the positive long-run effects on employment are not mirrored by a corresponding decrease in unemployment. We show furthermore that it is important to consider the cumulated (net) effects of the programmes for an assessment of the overall effectiveness of different training programmes because they can yield conclusions which differ from those of the point-in-time estimates.