Not-for-profit provision of job training and mediation services: An empirical analysis using contract data of job training service providers

This paper analyses the relative performance and selection behaviour of not-for-profit (NFP) and for-profit (FP) job training service providers, using contract data of welfare-to-work trajectories of the Dutch social benefit administration. Our analysis takes full account of selection effects, both ex ante (i.e. during the contracting process) as well as ex post (i.e. at the start of the program). First, for each cohort type of unemployed clients, service providers are only informed on the broad characteristics of cohorts that are contracted, thus limiting the room for adverse selection here. At the start of trajectories, however, selection may occur (directly) by service providers sending back clients or (indirectly) by encouraging clients to start a program, so as to receive additional (fixed) payments at the start of the program. We find strong evidence for FP service providers to be more selective than NFPs in both ways. Regarding the estimation results for the job placement rates, however, we only find weak evidence for NFP job training service providers to outperform FPs in the durability of job contracts.