Vertrouwensexperiment Oss: Zelf aan het stuur en op maat. Werkt het?

This third and final report discusses the results of the research on the outcomes or effects of the Trust Experiment implemented in the municipality of Oss in the Netherlands. In this 2-year RCT-experiment (October, 1, 2017 – December, 31, 2019) with three experimental treatments (1. self-reliance and exemption of obligations, 2. intensive mediation or tailored support and 3. a control group) a different way of supporting people in welfare has been applied. The research consists of a product and a process evaluation. The product evaluation examines the results or effects of the experimental approach compared to the regular treatment not only on outflow to work but also on well-being, health, freedom of choice or capabilities, social participation and self-reliance; the process evaluation examines the execution of the experiment by the consultants of the municipality of Apeldoorn/Epe. In other words, how does the process of supervision and support go. The study shows positive results on outflow to full-time paid work, both when compared to the randomised control group and compared to the non-participating social assistance beneficiaries. But with the other outcome measures, a different picture came out of the regression analyses; some of our expectations on the non-work related outcome measures are confirmed but others refuted. However, there is also good news to report. The positive treatment effects on social participation are not insignificant from a policy perspective. Also, the scores that customer managers give to the participants on the participation ladder show a significant increase for the majority of participants. As a result, the distance to the labour market has narrowed and social participation has increased, which was an important objective of the experiment. We also do find positive and significant effects of subjective wellbeing over time but for all groups. Whether this is the result of the experiment, thus a treatment effect, is as yet causally undescided. We find some evidence that selection bias and experiment effects (Hawthorne) might have affected the outcomes because the behavior of the participants as well as the caseworkers appears to have changed to some extent in the standard or controlgroup already from the start of the experiment. Further research is needed with another comparison group of welfare recipients during this period to explore this issue further.