Does Offender Supervision Reduce Re-Offending? Evidence from Novel Administrative Data

This Abstract was Corrected on 02/02/2026 A Correction Notice is available: here ObjectivesWe ask whether supervision and license conditions for offenders reduce re-offending. Our goal is to leverage novel administrative data to estimate the causal effect of the 2014 Offender Rehabilitation Act (ORA), a government flagship policy that introduced supervision and license conditions for offenders released from short prison sentences. MethodsWe deploy a regression discontinuity design exploiting that ORA quasi-randomly assigned offenders to supervision by probation services upon release, depending on whether they committed the offense that led to incarceration before/after a cut-off date. We obtain a causal effect by combining our quasi-experimental method with administrative data from the Ministry of Justice. By linking these data, we construct offender histories and journeys through the criminal justice system, from initial offense, to court appearance, to prison spell, to probationary period (w/wo supervision and license conditions), to – in many cases – re-offending. ResultsWe demonstrate that linkage and analysis for policy evaluation of complex administrative dataset is feasible. We identify the universe of prisoners who were affected by the ORA reform. By linking their prison records with probation records, we verify that they were indeed subject to supervision and license conditions. Via linkage to the court records we identify the type of offense and sentence they received. We are also able to link them to further court appearances. This allows us to not only construct detailed offender histories but also re-offending trajectories at the individual level. ConclusionCombining novel administrative data with a credible quasi-experimental approach, we show whether offender supervision and license conditions are an effective and efficient means of decreasing re-offending. We demonstrate that the use of detailed administrative data can be effectively used to evaluate policy interventions in the criminal justice space.