David Ainsworth: Six issues with social impact bonds

27 May 2014 Voices

Peterborough Prison, site of the first social impact bond

In the space of seven days the government pulled the plug on its flagship Peterborough social impact bond (SIB), then ploughed £30m into a SIB fund for Neets. David Ainsworth explores what's going on.

A few years ago, the government launched a pilot social impact bond. It wanted to test whether SIBs were an effective mechanism to tackle intractable problems.

A couple of weeks ago, just over halfway through this pioneering proof-of-concept trial, the Ministry of Justice scrapped it, and replaced the SIB with an “alternative funding arrangement”, as yet undefined.

It’s not being canned because there’s a problem, however. Rather the reverse. No further testing is needed, it appears. It works.

Just to prove it, the government almost simultaneously unveiled another £30m of funding for SIBs.

The reason for scrapping the pilot SIB was that it was not compatible with the funding mechanisms of the Transforming Rehabilitation programme, a nationwide drive to replace the probation service with a group of 21 community rehabilitation companies. The funding model for these companies will include a payment-by-results programme to reduce reoffending. But as it uses different metrics to the SIB, the two can’t run simultaneously.

All of this sends very mixed messages.

What I want to try to do now is to break down half a dozen of the key issues around SIBs, and look at whether they’re something the sector should welcome.

Are they something we should be doing a lot of? Are they a useful product for niche situations? Or are they a potentially useful example of a funding tool which in the hands of government could go badly wrong?

But first, a brief explanation. What is a SIB anyway?

The basic model is that the charity agrees with a public funder that it will try to solve a problem. Obesity in Rochdale, problem children in Leeds, repeat offenders in Peterborough. If the charity succeeds in reducing the problem, the public body will make a payment.

So far, this is a regular payment by results contract. And PBR has problems – one, the fact that you risk not getting paid if you fail, and two, the fact that even if you succeed, you won’t get paid for a long time.

But a SIB differs from other PBR in that social investors get involved. They front the money to deliver the contract, and they get the profit if the intervention works. If the intervention fails, they lose out. Either way the charity is financially secure.

So a SIB offers long-term, low-risk funding. But it also has one other big advantage for charities: it uses a “black box” approach. This means, basically, that the charities involved can use whatever interventions they like, so long as they work.

It sounds good. But the big question about SIBs remained unanswered – do they succeed in producing improved outcomes, compared to other types of funding?

So far, there’s some anecdotal evidence to say they do, but the first formal evaluation of the effectiveness of the first SIB isn’t likely to appear until July.

Despite this, they have already caught on. The government has already launched a score of social impact bonds, plus a number of funds to promote their use. Worldwide, more than a hundred are in the offing.

Given that it’s a format designed to promote evidence-based contracting, it seems curious the government has launched so many without waiting for any evidence about their effectiveness. But actually, there are strong reasons to believe that SIBs are really powerful tools, if they are constructed effectively.

The “black box” model

SIBs "black box" approach says the people who know most about a problem should be allowed to do what they need to solve the problem. Everyone else gives them money and gets out of the way. They are measured not on processes, but on results.

Letting the experts do what they want to solve the problem seems like a good idea, but it’s surprisingly rare.

For the government, this is down to trust and accountability. You can't just hand someone the cash and say “Best of luck” unless you’re really sure they know how to solve the problem. And you’ve got to be able to show the public, and the EU, that you haven’t just given the money to your mate.

The service provider, on the other hand, wants to know he has as long as needed to solve the problem, and won’t have to reset his priorities halfway through because a new chief executive takes over at his funder.

One positive thing to emerge from the SIB model is that it has helped popularise outcomes-focused commissioning.

You could strongly argue that if a charity’s intervention is already proved to work, though, it makes a lot more sense to just give that charity a long-term grant.


While departments want to solve problems before they come up, they only get one year’s money at a time. Most parts of government can’t borrow. This year’s money has to be spent on this year’s problems, not next year’s.

In addition, it’s easy for a government to justify spending money on solving a problem, but hard to justify preventing it existing in the first place.

All of this leaves government willing to offer a very high premium for the right to pay later, out of savings already made. This, plus the cash crisis in government at the moment, suggests that PBR is here to stay. The sector’s focus has to be on developing PBR which works for charities and beneficiaries.

Social investment

One of the best things about SIBs is that it transferred risk to the right place – investors, who in theory are professional risk takers.

The term “risk capital” is beginning to be used a lot in the sector – it just means investors who take a risk on the success of your operation.

In a SIB, the investor takes most of the financial risk, in exchange for a big chunk of the profit. This is pretty similar to what shareholders do in conventional businesses.

The charity sector has historically been a cash business. You identify a project. You raise the cash to do the project. You spend the cash. Lather. Rinse. Repeat.

But service delivery isn’t like that. When delivering government contracts you often need balance sheets and scale. And to do that you need to attract finance. Lots of it.

The SIB pioneered this idea of transferring risk. But there ought to be cheaper, simpler, better ways of doing so, because a SIB is a complex, expensive operation, with a lot of moving parts.

There aren’t better ways, though, at the moment.

One problem with SIBs is that they all require social investors. But so far, the pool of investors is limited. Social investment is a bit like going for an early morning run – many people like the idea, but few people do it.


One incredibly large problem for the SIB, and for all PBR, is that you need to measure the right thing. Often it’s quite difficult to find exactly the right metric.

A long time ago, I saw some graffiti on the side of the road, which read “Keep Britain tidy – picnic in France”. This is known as displacement – the cheapest way to hit your target is often not to solve the problem, but to move it down the road.

Another problem is dead weight loss. How do you measure what would have happened if the charity hadn’t intervened? It’s like claiming you have a lucky hat which keeps the elephants away. How many elephants would have come by if you’d done nothing?

With a PBR contract, it’s crucial not to P for the wrong R. The Work Programme is the prime example. It paid for getting people into jobs, and this led to creaming and parking – claiming credit for the people who would have got jobs anyway, and ignoring the hard-to-help.

Statistical noise

Another problem is that you need to separate out the difference you’ve made from what's called "statistical noise". There's a certain amount of random variation between sets of clients, just due to the fact that people and circumstances are different.

It's like flipping a coin. You know if you flip it 100 times it'll be roughly 50-50, but you wouldn't be surprised if it was 53-47 either way. So if you come up with a magic method for improving your coin flip chances by 2% (there is one, actually - always bet on the face-down side) it could get swamped by natural variation.

The trouble is, setting the threshold of success above anything which could be down to random variation, sets an unrealistically high bar.

Noise can generate wilder swings than intervention, and the government will often only start paying after you’ve got over the noise threshold – ie when you’re generating some pretty significant improvements. If the public sector gets too hung up on this, they can put the kibosh on the whole thing.
SIBs are also expensive to set up. They take a lot of paperwork, and a lot of people have to agree to something. By the time you’ve got everything agreed, you need to have a lot of cost-saving to make it worthwhile.


For SIBs to get backing, savings also have to accrue in just the right way. They have to be cashable, for a start. Or to put it another way, it’s not good enough to reduce obesity; you have to show that as a result you can close a hospital, sack a few doctors, or in some other way make real, cash savings. It’s not enough to make people’s lives better, you have to help government spend less on sorting them out.

And they have to accrue to the right people. Your intervention may save cash for many agencies – the police, the courts, the NHS, the local council, and Uncle Tob Cobbleigh and all. But none of them will pay a brass farthing to prevent cost to another agency. One agency needs to see enough savings to pay for it alone.


Okay, so where does that leave us?

With a potentially really good idea, but with an awful lot that can go wrong.

Basically, given the complexity and cost of a SIB, in a sensible world we would only use them, or any other PBR, to fund things when we didn’t know if they would work. We would leave the design to independent experts, and accept the cost because they were experimental.

Ideally, SIBs and PBR shouldn’t be used as a permanent solution, at scale, because of the extra cost involved.

Once you have a proven set of effective interventions, the sensible thing to do is to fund them through a long-term contract, or better still, a grant.

But there are seemingly innocuous but actually devilish problems of government accounting and accountability which make it easier to pay for an ambulance at the bottom of the cliff than a fence at the top. So long as these are with us, the SIB, or something like it, may be the best alternative.

More on