Reducing Youth Reoffending: Quantitative Impact Evaluation

Social Bonds are a type of pay-for-success contracting approach that have been used internationally to fund a variety of social interventions, including youth recidivism prevention. In 2017, Oranga Tamariki entered into a six-year Social Bond agreement with Genesis Youth Trust to deliver a programme which aimed to reduce the frequency and severity of youth offending in South Auckland.

This evaluation utilises data up to June 2023 from Police and the Youth Justice National Minimum Dataset to quantify the impacts of the pilot on participants’ reoffending outcomes. The findings complement those in two previous evaluations Oranga Tamariki commissioned on the Social Bonds pilot.

In the 12 months following enrolment:

• over half (54%) of the participants did not reoffend,
• about 70% reduced the frequency and/or seriousness of their offending (compared to the 12 months before).

In the 24 months following enrolment:

• 46% of participants did not reoffend,
• about two-thirds reduced the frequency and/or seriousness of their offending (compared to the 24 months before).

Relative to non-participants from a matched control group of rangatahi who received ‘business-as-usual’ type interventions, Social Bond Pilot participants:

• Were 28% more likely to not reoffend, and 24% more likely to reduce the frequency of their offending in in the 24 months following enrolment.
• Were 14% more likely to reduce the frequency of their offending in the 12-months following enrolment.
• Recorded on average nearly two fewer offences in the 24 months following enrolment. Additional exploration of the data indicates that fewer offences were driven both by a greater share of participants who did not reoffend, as well as by fewer offences committed by those who did reoffend.
• While indicative of positive change, the impacts of the pilot on reoffending severity were not statistically significant.

Given that at the time of the evaluation, not all participants had completed the two-year intervention (or had two years of offending data available post-enrolment), a reasonable next step would be to re-examine the pilot impacts in the future. A supplementary benefit of this would be the ability to explore longer-term impacts (i.e., beyond two years).

In addition, given that the pilot seems to be reducing offending frequency to a greater extent than business as usual interventions, we recommend that the pilot’s return on investment should be formally assessed via a Cost Benefit Analysis.