The Heterogeneous Effects of Training Incidence and Duration on Labor Market Transitions

Training programs are an important tool of active labor market policy in Germany as in many other countries. There exist different types of training programs, but also important heterogeneities within a given program type, in particular with regard to the duration of program participation. The question how the impact of a program of a certain type varies with the length of the scheme is an important dimension of program heterogeneity that has received little attention in the literature so far. One reason for this may be that actual program lengths are endogenously determined. Participants frequently drop out before reaching the planned end of the program or they stay longer on the program than originally scheduled. The decision to shorten or prolong program participation depends on the success in job search, an issue that must not be neglected when studying the effect of different program lengths. In addition, the assignment to training depends on the success of job search, because training programs in Germany may be assigned to the job-seeker at any point in time but only as long as he or she has not yet reentered employment. Furthermore, selection into program participation and employment may depend on characteristics observable to the researcher (like age and education, for example) and characteristics unobservable to the researcher (like personal traits, for example). In this paper, we propose and estimate an econometric model that takes all these issues into account. Furthermore, we develop a strategy to simulate the treatment effects of interest using our estimation results. We study a large-scale German program involving training in professional skills (Förderung der beruflichen Weiterbildung). Participation in the program lasts eight months on average, but scheduled program lengths vary between a couple of weeks and more than one year. We use rich administrative data and perform separate estimations by gender and region (East and West Germany). Our findings imply positive effects of training on the employment probability emerging three to four quarters after program start in all four subsamples considered. First, our estimates suggest that ten quarters after program start employment rates of the participants are on average 12 to 21 percentage points higher, than they had been in the counterfactual scenario in which participants had not participated in the program. The effects are higher for women than for men and higher in West Germany than in East Germany. Second, we use our model estimates to simulate the effect of treatment start at a given date versus not starting a program at that point in time but maybe later. We find that this effect – the effect of participating versus waiting – is about one third lower than the effect of participating versus not participating. Third, we use our estimates to analyze how training effects vary with the planned program duration. During training, participants generally search less intensively for a new job than comparable nonparticipants. Therefore, employment effects of training are typically negative in the short run, and positive effects may unfold only some time after the completion of the program. Our analysis provides evidence to address explicitly the question whether negative short–run effects are necessary to achieve economically important positive employment effects in the long run, or whether comparable long-run effects can be obtained with shorter programs at lower costs. Our findings suggest that longer planned enrollment lengths of three and four quarters, respectively, as opposed to just two quarters lead to an increase in employment rates by four to six percentage points and six to eleven percentage points, respectively, in the medium and long run. This suggests that, on average, longer training programs translate into higher long-run employment gains.

Fitzenberger, Bernd, Aderonke Osikominu and Marie Paul (2010), The Heterogeneous Effects of Training Incidence and Duration on Labor Market Transitions, ZEW Discussion Paper No. 10-077, Mannheim.