Reducing reoffending on release
The payment by results pilots at Doncaster and Peterborough prisons were portrayed by the last Justice Secretary as the rationale for his radical overhaul/privatisation of the probation service via the Transforming Rehabilitation (TR) initiative. The pilots were designed to cut the very high reoffending rate of released prisoners, particularly those who served short sentences (under 12 months) and did not receive the support of the probation service.
The outcomes of the pilots have been variable to date (see below). However, this recent (30 July 2015) report is the final process evaluation of the Doncaster pilot – and we will have to wait for the final reoffending outcomes.
The Doncaster model
HMP Doncaster is a privately operated Category B prison, managed by Serco. In July 2011, Serco’s contract for running the prison was due to expire and the MoJ commenced a procurement process to re-let the contract. Serco’s bid included an “Alliance” with two voluntary sector organisations Catch 22 and Turning Point to deliver a new model of offender management and resettlement intended to reduce re-offending rates on release. They were awarded the contract to continue running HMP Doncaster on a PbR model with ten per cent of the prison’s annual revenue dependent on a 5% reduction in re-offending rates.
The PbR pilot was due to run over four years from October 2011 but was halted one year early because of the implementation of TR when the provision of resettlement services was transferred to the new Community Rehabilitation Companies.
As stated above, this report was not about the reoffending outcomes, but it’s still helpful to be aware of how effective the pilot was to make sense of the findings of this process evaluation.
The PbR target was a five percentage point reduction in offenders reconvicted for an offence or offences within one year of their discharge from custody, compared to the baseline year of 2009 (the latest available at the time).The reconviction rate for the first cohort (October 2011 to September 2012) was 5.7 percentage points lower than the 2009 baseline year so the five percentage point target was met. However, the reconviction rate for the second cohort (October 2012 to September 2013) was only 3.3 percentage points lower than in the baseline year so the five percentage point target was not met.
What did we learn?
The report is based predominantly on a large number of interviews with stakeholders, providers and offenders themselves. There were a total of 81 interviews with offenders, but only 30 of these related to their experience on release. Interviewees identified a number of strengths of the pilot scheme:
- The introduction of a case management approach in custody and community, which was proactive, holistic, flexible and offender led – this meant that all offenders received a service whereas previously only those offenders who sought help were provided with (a much less intensive) service.
- Providing custody based case management for offenders at the outset of their sentence and initiating community support six to eight weeks before their release.
- Having informal relationships with partner agencies ‘helped things to get done’ (housing providers were the most common type of partner agency, which reflected the fact that housing was the most common support need identified by offenders and staff).
- Using volunteer mentors, who met independently with offenders, was regarded as an asset in the delivery of support to offenders.
- The Veterans in Custody (VIC) volunteer scheme, which provided informal mentoring and specific support to offenders who were ex forces, and was delivered by former servicemen.
The same interviewees also identified a number of weaknesses:
- The binary outcome measure did not capture frequency or severity of reoffending. This resulted in the Alliance withdrawing community support from individuals who reoffended.
- There were more custody based staff posts than community based posts, which the staff interviewed felt was an imbalance in the use of resources to reduce reoffending most effectively. Staff found the transition to these new roles challenging at the start of the pilot.
- It was difficult to support offenders released ‘out of area’ effectively because of a lack of face to face contact and limited understanding of the local support landscape.
- The Alliance had limited control over some activity which was likely to influence reoffending. This included out of area cases; those led by probation; and interventions delivered by partner agencies, which varied by area.
- Accessing community interventions was dependent on the availability of existing services, many of which were already ‘stretched’.
- There were problems with the clarity of role and commitment among some volunteers.
- Collecting and managing data was more complex than the Alliance envisaged. The data management system was not fully operational at the beginning of the pilot which meant that staff had to rely on a paper-based system in which they had little faith and which made information sharing difficult.
- The informal partnership arrangements meant there were issues with sharing data and the sustainability of relationships.
It would have been helpful to have more offender views on how helpful they found the support from volunteers.
A lot of things have changed since the Doncaster pilot was launched. The Alliance has been disbanded – Catch 22 and Turning Point walking away from the informal partnership after Serco became perceived as toxic following their over-billing the MoJ for the electronic monitoring (tagging) contract they were delivering.
The main original benefit of the pilot – that those serving short prison sentences would get proper support on release – is now, in theory, available to everyone under the TR reforms.
Sodexo and NACRO are the new partnership running the South Yorkshire Community Rehabilitation Company and it will be interesting to see whether they can have a positive impact on reducing the reoffending of released prisoners – their results will also be subject to a payment by results contracting approach, this time using both a binary and frequency (but not severity of offence) payment model.