Association for Public Policy Analysis & Management

Transitional jobs programs have been demonstrated to improve employment and earnings, but only during the period of transitional employment. Longer term, those impacts do not hold up on average. This paper examines the extent to which those who get a larger, potentially better “dose” of a transitional employment intervention experience relatively greater impacts. Using data from the seven-site Enhanced Transitional Jobs Demonstration (ETJD), the paper uses variability in individuals’ experience with the intervention to assess associated variation in long-run labor market outcomes and impacts. The four patterns of subsidized job participation examined are: rapid entry with long retention, rapid entry with short retention, slow entry with long retention, and slow entry with short retention. After defining and mapping out descriptively the employment and earnings trajectories of individuals in those four groups, the paper will analyze impacts on these groups using an analysis of symmetrically-predicted endogenous subgroups (ASPES). The result is to report the impact of the ETJD program on each of the four sets of workers: workers who begin their transitional jobs immediately and remain in them a comparatively long time, workers who start early but stay in transitional jobs a short time, workers whose entry into transitional employment is delayed but then lasts a comparatively long time, and workers whose entry is delayed and whose tenure is short. The distinctive element of the ASPES approach is the symmetry with which subsets of the experimental sample are selected for impact comparisons. As a result, the approach leverages the experimental evaluation design and establishes a strong case for causal inference subject to assumptions about the homogeneity of impacts for workers with similar transitional employment experiences but different background characteristics. This paper makes an important contribution to the field’s knowledge of the effects of transitional jobs programs.