A ticking bomb
“We only call him the Exploding Boy now, of course; retrospectively. For most of last year he was known only as Ticking Boy.” That is the first line from a new short story by @nickparker from a collection that got a rave review in Saturday’s Guardian.
It made me think about the bomb ticking away ready to blow up the ever-growing range of payment by results schemes (details here) being developed by different government departments.
That bomb is the method chosen for measuring outcomes which, for Ministry of Justice schemes, are always about reoffending. I wrote recently about my concerns that the MoJ seems to have decided that the outcome measure should be “binary” – a simple yes/no indicator of whether someone commits a further offence. The alternative approach is to measure “frequency” – a reduction in the number of re-offences. I argued that “frequency” is a more realistic (albeit more complex) way of measuring reoffending since we have a very large evidence base that tells us that most criminals who stop offending do so over a period of time.
My concerns have been amplified on hearing that the MoJ is insisting that many of the PbR pilot schemes include Prolific and other Priority Offenders (PPOS) in their target group.
PPOs comprise only 10% of offenders but they are responsible for about half of all crime. They typically have lengthy criminal records and entrenched problems and although many do eventually desist from crime, it normally takes several years before they are totally law-abiding. The national strategy has three strands:
- Prevent and deter (which focuses mainly on young offenders),
- Catch and convict, (led by the police) and
- Rehabilitate and resettle (led by the probation service).
Evaluating the success of PPO schemes is notoriously difficult. Take, for example, a prolific offender who has just been released from prison. As a PPO, he (the vast majority of PPOs are men) will receive increased surveillance from the police and access to more intensive support from the probation service. If he is convicted of a further offence a couple of months post-release, the police will count his case as a success under “catch and convict”, while the probation service will regard it as a failure under “rehabilitate and resettle “.
We often talk about the objectives of criminal justice agencies as “preventing crime” or “reducing re-offending”, using the terms interchangeably as if they were the same thing. Quite often they are not.
The MoJ is keen to ensure that PPOs are included in PbR reducing re-offending schemes because it is, quite reasonably, concerned that organisations delivering PbR schemes will “cherry pick” – target less serious offenders who are more amenable to change to ensure that they receive their premium payments. Indeed, I’ve argued elsewhere that one of the chief benefits of PbR schemes ought to be that it aligns the objectives of both Government Departments and providers of PbR schemes.
Government wants to target the “hard cases” because if they are rehabilitated, they will produce the most savings to the justice, health and benefits systems. And the same is true for providers – the more prolific offenders they help, the more they will reduce re-offending and the more they will get paid.
The attractiveness of the PbR approach is that, for once, we should have a system with a built-in incentive for real effectiveness.
However, that argument no longer holds water under a binary measurement system.
In a recent PbR roundtable discussion convened by @reformthinktank, Crispin Blunt, Parliamentary Under-Secretary of State for Prisons and Probation, justified the government’s rationale for a binary approach:
“because it’s clean and clear if someone has been reconvicted, and of course it’s clear that the savings then accrue to the Ministry of justice because the person definitively is not in prison, has not gone through the justice system…”
Now this is, of course, patently true. However, adopting a binary measure will mean that there is a disincentive for providers to work with PPOs – or indeed any other offenders – the moment that they commit (or, perhaps, are even charged with) another offence – no matter how minor – for the simple reason that will not be paid for working with anyone who re-offends.
Probation trusts and voluntary organisations know that it is vital to develop long-term working relationships with offenders which moves them steadily towards desistance (the current jargon for stopping offending) over a period of months and years. A binary approach will inevitably interrupt this relationship at the first sign of re-offending.
This is getting very close to a situation where reducing re-offending PbR schemes are destined to fail. Insistence on a binary approach might well lead potential providers of PbR schemes, and those who are being encouraged to invest in them via social impact bonds, to decide against getting involved in the first place.
The transcript of that Roundtable discussion makes it clear that many people involved in the delivery of Justice PbR schemes are concerned about the introduction of a binary approach. They include Rob Owen, Chief Executive of @StGilesTrust, the organisation which runs the PbR funded ONE Project which is working to prevent reoffending by short-term prisoners released from HMP Peterborough. He describes himself as getting fixated on the binary versus frequency argument. Clearly, so am I.
However, it’s hard not to get a little pre-occupied when the bomb ticks a little louder every day.
I would broadly agree with the proposition that frequency measures make more sense in a PbR contract than do binary measures. It comes back to the point that Russell makes in the post about whether the focus is on reducing crime or on reducing re-offending. Assuming the former is the focus (which surely it must be – after all the purpose of reducing re-offending is to reduce crime) then frequency measures are more appropriate particularly with high-volume perpetrators such as PPOs
However binary measures do have their place as well: they have the great merit of simplicity as the Minister points out but can run into the cherry-picking problem. One way to resolve this is to use a binary measure as the PbR metric but to correlate the payment to assumed likelihood of individuals re-offending using the OGRS3 scale: thus the contract holder is incentivised to work with more complex cases as they will paid more if these individuals do not re-offend. I am very involved in a contract between a south London Council and Third Sector provider which is testing this approach