Beschreibung

"This paper uses a dynamic model for employment and training transitions in discrete time to estimate average employment effects of training incidence and planned training duration for the treated. We account for the dynamics of program start and continuation while treatment and outcome transitions are linked through correlated individual- specific effects. Identification relies on a consistency condition, conditional sequential randomization, time-varying covariates, and exclusion restrictions. We implement our framework applying MCMC methods to rich administrative data for a large training program in Germany. In addition to finding a positive training effect on employment between 6 and 12 percentage points 2.5 years after program start, we demonstrate that a longer planned training duration typically results in higher medium- to long-term employment gains" (Author's abstract, IAB-Doku, © Elsevier) ((en))