Transforming Rehabilitation is being dressed in sheep’s clothing
Initial results from the world’s first social impact bond (SIB) were announced last week. The Peterborough SIB’s One Service achieved an 8.4% reduction in the reoffending of prisoners sentenced to less than 12 months. This result exceeded the 7.5% minimum required over the life of the project for investors (of which LankellyChase is one) to get their investment back, but fell short of the 10% stretch target needed to trigger an outcome payment this time around.
Inevitably a great deal of the commentary that followed the results – including whether 8.4% was a success or failure – ranged along the battle lines drawn by the wider payment by results (PbR) debate. The landscape has changed radically since the SIB was first dreamt up, so that it gets cited as evidence for or against reforms that its creators could not have anticipated. The uses and abuses of its name have been legion and nearly all opportunistic. All of which obscures any attempt to judge the merits of the SIB on its own terms.
Background to the SIB
Some of the commentators who have been quick to dismiss the SIB would do well to return to the original problem it was trying to solve – and perhaps they should answer for how they would have solved it. The absence of support for prisoners sentenced to less than 12 months has been a nonsense and a disgrace for decades. Reviews have come and gone recommending change, such as the Halliday review Making Punishments Work (2001) and the Social Exclusion Unit report Reducing Reoffending by Ex-prisoners (2002). It has been abundantly clear that the social and financial costs of releasing prisoners into destitution are incredibly high, and yet no one has been able to work out where the additional resource should come from to supervise so many additional people.
It is worth recalling that, until recently, every additional responsibility on public services has been issued with a substantial price tag. The idea that the probation service might reconfigure existing resources was complete anathema and fiercely resisted. As such, the government’s main change lever was to pour in extra cash. This is why we became obsessed with piloting ‘invest to save’ approaches. The hope was that if we could prove that supporting the resettlement of short-term prisoners reduced their reoffending, and if we could also prove that this saved the Exchequer money, perhaps the Treasury would find a way of expanding the public money available to resettlement services.
Many pilots have come, gone and been forgotten. Quite ambitious regional schemes funded by Europe were set up only to be dismantled. None of these (in marked contrast to the SIB) caused the slightest stir among policy makers or other opinion formers. Such pilots have become almost piecemeal stopgaps rather than serious attempts to model change. A great deal of money has already been spent in this vein to little or no effect.
What did all these failed initiatives have in common? All had at least one of the following limitations: short timescales, poor resourcing, tiny and/or partial cohorts, poor or no through-the-gate support. Even when they got these elements right, they didn’t have anything like the rigour around data to nail the cash saving argument sufficiently. And all were characterised by predetermined inputs and outputs: numbers found a hostel on release; numbers attending a job centre interview, numbers referred to drug treatment, numbers starting or completing a cognitive behavioural programme.
The input/output model has acted as a particularly dead hand. As so often with social policy, the outcome (reduced reoffending) could not be contractually controlled and so it was reduced to proxy measures which appeared to correlate closely to it and could be counted. What this inevitably drove was (a) a whole lot of box ticking whereby providers were incentivised to chase numbers of inputs and outputs but not people and (b) a whole lot creaming and cropping, with the easiest to help being helped first. Classic stuff, and especially problematic for pilots trying to test the effectiveness of services for people with multiple needs.
You might wonder why the proven capabilities of the probation service, who after all were supervising all other offenders, weren’t trusted with the task. The expense of extending a public service aside, the levels of reoffending of released prisoners under probation supervision have never been its strongest suit. The review of probation’s resettlement practice and prioritisation in HM Inspectorates of Prisons and Probation’s Through the Gate (2001) wasn’t exactly a ringing endorsement. And those voluntary sector organisations who had been attempting to fill the gap for years were chomping at the bit to take the challenge on themselves and, candidly, were clear they could do it better.
What the SIB offered
Tens of thousands of victims later, with all the misery and cost that this implies, Social Finance entered the fray proposing a social impact bond mechanism. What immediately recommended the SIB was that it offered a method of unsticking the seemingly intractable resourcing problem. The tax payer would only foot the bill if the cost was offset by savings generated through improved outcomes. Whether or not you believe that a SIB mechanism could underpin an entire resettlement system, at the very least the financial circle was squared so that an ambitious pilot could be afforded. And critically, the government was tied into validating and responding to its results.
Several other things recommended the SIB: (i) the timescale (six years) and scale (3,000 prisoners) allowed the service to be tested thoroughly; (ii) the cohort was the entire short-term prisoner population of HMP Peterborough matched to a control, so cherry picking was avoided; (iii) the scheme was conceived as a fully resourced prison to community service; and (iv) the outcome payment mechanism meant that the initiative didn’t have to play the input/output game, and was free to focus on and follow the person, moving resources between services depending on emerging evidence of need. Significantly, the financial mechanism drove an entirely new level of rigour around data which would allow a clear assessment this time of whether the service was working and whether it was generating cashable savings.
In other words, its design overcame all of the limitations that had dogged so many previous pilots. As a result, the workers have been freed up to contend with and learn from the real challenges of the lives of short-term prisoners. They have been able to build relationships of trust and have moved well beyond previous models by engaging people with lived experience to support the service and offer mentoring. They have been able to make mistakes, learn and change the service, and have had a clear incentive to do so. All the things that you hope for from a public service but which we frequently find so hard to get right, especially when the service landscape is highly complex. I have visited the service and thought it was the best realisation of a short-term prisoner resettlement service that I have seen, with strong values and hard won expertise running through it.
All of which speaks to the fact that how you fund a service fundamentally determines the quality and effectiveness of the results that it can achieve. I write this in the full knowledge that many PbR mechanisms lead to all manner of perverse consequences and gaming behaviours. In the SIB, however, Social Finance seems to have found a way of shaping and protecting a space in which front line workers can focus on doing the right thing, not just through gut instinct or individual judgement, but through ongoing iterative data gathering and learning.
Conclusion
What conclusions can we draw? Social Finance, to its credit, has been very cautious. We only have results from the first cohort, hence it is still far too early to pronounce on its cost benefit and replicability. So 8.4% is certainly an encouraging reduction, especially as engagement levels are continuing to grow since the first cohort; but even when later results come through there will still be a lot of cost data that will need to be crunched. There is also a lot more analysis needed on the relationship between the financial model of the SIB and the practice model of the One Service. Is the SIB mechanism the only model that could incentivise the same behaviours and performance on the ground? We just don’t know.
No such caution has been evident in the government’s response. The SIB was being repeatedly cited by justice ministers as evidence for their Transforming Rehabilitation (TR) reforms long before its initial results were available. In which case, they were never actually interested in what the SIB might teach them. And this is seen in the way that TR appears to be rewarding results, with a rigidity that will make it very hard for the community rehabilitation companies (CRCs) to emulate the crucial flexibility of the SIB. What they are rolling out is not what is being piloted.
The full resourcing of resettlement which the SIB made possible through the outcome mechanism is also going to be exceedingly tough to replicate under TR. Short-term prisoners have simply been added to the responsibilities of the CRCs without any new resourcing. Following further public spending cuts, the CRCs’ budgets will be significantly lower than the former probation trusts and yet they will have far higher caseloads. Once short-term prisoners are subject to statutory license on release, we face the prospect of overstretched staff reduced once again to the bureaucratic enforcement of sentences. In other words, reconvictions triggered by breached licenses are set to sky rocket.
Within this much tighter financial context it was vital that incentive structures forced CRCs to make short-term prisoners a priority. News emanating from contractual negotiations, however, suggests that penalties and rewards are stacked towards managing the public safety risks posed by more serious offenders. This is understandable, but TR was justified on the basis of reducing high reoffending rates. Given that the SIB was dreamt up to address the existing system bias towards public safety, then this is simply a reversion to type.
TR is being dressed in the sheep’s clothing of the SIB, but risks systematically disabling most of what the SIB has been trying to achieve. For those of us who have been trying for years to find a workable way of resettling short-term prisoners, this is quite a blow. Whatever your views on the PbR mechanism that underpinned the SIB, here was a scheme that was finally grappling with the core questions of what it takes to do the job properly. It is deeply frustrating, to put it mildly, to see the Ministry of Justice welcoming the Peterborough SIB results when in fact it should be explaining why it has chosen to ignore the opportunity that the SIB represented.
It is not too late in the TR process to get some of this right. Peterborough type schemes could be stimulated elsewhere in the system through enhanced incentive structures. If not, and if MoJ drops this ball completely, it is going to be very interesting to see how it welcomes the next set of results in 2016 when TR is well under way.
