Data Insight: From custody to community: Does post-release supervision reduce re-offending?

This Data Insight by ADR UK fellow Markus Gehrsitz draws on Ministry of Justice Data First linked administrative datasets on courts, prisons and probation services to examine whether post-release supervision for offenders released from short prison sentences reduces re-offending in England and Wales.

Read publication

This study examined the impact of post-release supervision and license conditions on re-offending. Specifically, it analysed the introduction of supervision for offenders released from short prison sentences, by way of the 2014 Offender Rehabilitation Act (ORA).

Prior to ORA, offenders who served sentences of 12 months or less had been released unconditionally. By comparing these offenders to offenders who were affected by ORA, we were able to gauge the importance of supervision.

We found that supervision reduced re-offending by around 15% in the first month after release, and by around 5.5% in the long run, three years after release. While the short-run effectiveness was partly due to prison recalls for violations of license terms, the long-run reduction suggested that post-release supervision genuinely changed offenders’ behaviour.

The crime-reducing effects of supervision were strongest among first-time offenders and those subject to more intensive supervision.

A cost-benefit analysis suggested that the economic value of crime-reductions clearly outstrips the costs of supervision.

What we found

Post-release supervision and license conditions reduces re-offending

People released under supervision were significantly less likely to re-offend in the weeks immediately following release: they committed 15% fewer offenses than similar, unsupervised offenders and were also 10% less likely to commit any (re-)offense at all. The size of the effect shrank as we expanded our follow-up window. One year after release, our supervised treatment group committed 7.6% fewer offenses than our unsupervised control group. This effect remained persistent even after supervision ceases.

Crime-reducing effect of supervision persists in the long-run

Even three years after their release offenders who were supervised and subject to license conditions for only the first year had re-offending rates that were substantially lower than those of initially unsupervised offenders. They committed on average 7% fewer offenses years after their supervision had ended and were also around 2.5% less likely than their unsupervised peers to have committed any offense.

Prison recalls help reduce re-offending in the short run

During the early part of supervision, breaches of licence conditions can result in short prison recalls, typically lasting up to 14 days. Around one in seven supervised individuals experienced a recall. Re-offending fell most sharply during periods when recalls were most common. This suggests that temporary incapacitation play an important role in the short-term crime-reducing effects of supervision.

Supervision has lasting effects on re-offending

Importantly, re-offending remained lower among supervised individuals well beyond the period when recalls could occur.  Even years after release, those who had received supervision committed fewer offenses than similar individuals released without supervision. These persistent reductions indicate that supervision does more than temporarily delay offending — it appears to contribute to longer-term behavioural change.

Supervision works best for first-time and lower-frequency offenders

The largest long-term reductions in re-offending were observed among first-time offenders, and individuals with no previous prison spell. For people with extensive criminal histories, on the other hand, supervision mainly reduced re-offending in the short run through prison recalls, with much weaker long-term effects.

Effects differ by offence type

Reductions in re-offending were strongest for less serious offences, such as theft, and offences that do not result in re-incarceration. There is little evidence that supervision reduced serious offences that lead to further prison sentences. This suggests supervision is most effective at preventing frequent, lower-level offending rather than the most serious crimes.

More intensive supervision is more effective

People who served slightly longer prison sentences experienced longer periods of close supervision immediately after release. These individuals showed larger and more persistent reductions in re-offending than those whose supervision was less intensive. This suggests that the intensity and structure of supervision matter, not just whether supervision exists.

Why it matters

Evaluating criminal justice policies is typically very difficult. Often it is offenders who are most likely to relapse that receive the most attention. This tends to induce a positive but spurious correlation between interventions and re-offending. The Offender Rehabilitation Act (ORA) created a rare natural experiment that broke this spurious correlation between supervision and reoffending. The ORA cut-off date quasi-randomly split our population of interest into two comparable groups, one experiencing supervision and an all but identical one that was released unconditionally. This allowed us to reliably determine whether offender supervision is in fact effective.

This is substantial because until now, policymakers had little guidance on the issue. In fact, high re-offending rates among prisoners released from custodial sentences may well have been interpreted as evidence that their supervision was ineffective. Our study shows that this is not the case.

What is more, our study illustrates the value of leveraging large administrative datasets for policy-focused research. Only the detail and longitudinal dimension of the Ministry of Justice’s Data First Cross-Justice datasets allowed us to construct offender journeys and to efficiently use the natural experiment created by ORA. The detail of the data also helped provide guidance on where to shift probation officers’ attention and resources. For example, only through the large sample size of our data were we able to uncover that first-time prisoners are more receptive and responsive to supervision than prolific offenders.

Our study also shows that supervision passes a cost-benefit test. Our calculations suggest that the value of crimes avoided due to supervision exceeds the cost of the policy. However, one caveat here is that these calculations refer to prisoners released between 2015 and 2016. One aspect of supervision that has changed since then is that probation officers make use of prison recalls for license violations more frequently. Because prison space is scarce and expensive, this can shift the cost-benefit calculus.

In sum, our study was the first of its kind to robustly assess the effectiveness of offender supervision and license conditions. Our findings are useful in informing policy and practice, and they have the potential to help lower high re-offending rates and to alleviate Britain’s prison overcrowding crisis.