This paper analyzes the impact of job creation schemes (JCSs) on job search outcomes in the context of the turbulent East German labor market in the aftermath of the German reunification. High job destruction characterized the economic environment. JCSs were heavily used in order to cushion this development. Using data from 1990-1999 and building upon the timing-of-events approach, we estimate multivariate discrete time duration models taking selection based on both observed and unobserved heterogeneity into account. Our results indicate that participation in JCSs increases the unemployment duration mainly due to locking-in effects. However, twelve months after the program start the significantly negative impact on the job finding probability vanishes. We find evidence for effect heterogeneity. Our results suggest that female and highly skilled participants leave unemployment quicker than other groups, which results in highly skilled women benefiting from participation. However, we find no significant impact on post-unemployment employment stability. Our results are robust to allowing for random treatment effects. Also taking into account endogenous participation in training programs, endogenous censoring, or multiple treatment effects do not change the results.