Back in 1988, New Zealand economist Ronnie Horesh was thinking hard about how to fix social policy. Ronnie believed that the most effective catalyst for the generation of wealth was self-interest, and that this could be harnessed so that “the achievement – but not the setting – of social objectives [would be] contracted out to the private sector via a free market in a new sort of financial instrument: government-issued Social Policy Bonds.”
More than 20 years later, a less ambitious variant of Ronnie’s idea – called Social Impact Bonds (and Pay for Success Bonds in the US) – was first tried in the UK as part of the Cameron government’s Big Society agenda. A £7 million privately financed bond was contracted with a network of service providers to reduce the recidivism rate among 3000 short-stay Peterborough prisoners by 7.5 percent. If this happens, the first of the government-funded bond payouts will trigger mid 2014 and the investors will be in the money.
But it would also be a win for everyone else, because taxpayers pay only if the outcome is successful (if it fails, it’s the private investors who lose): achieving the recidivism target would save future taxpayers over £10 million in reduced costs of crime. The service providers get funded either way, but they too will prefer private finance because they are free to pursue the agreed goal the best way they know. They can be more experimental, seek early intervention, use their own knowledge of the situation and not be forced down bureaucratic channels.
The basic idea of a Social Impact Bond it is to take a costly social problem (recidivism, homelessness, and disadvantaged youth are the main applications thus far) that is currently addressed via publicly funded social services, and to replace this mechanism with private finance. The reason we do this is to get the incentives right – which was Ronnie’s key insight.
It is important to appreciate the substantial incentive problems with the standard public financed model of social service delivery. First, it is based on funding the inputs or costs of the service provider. So the government pays regardless of the outcome, and failed outcomes usually attract more funding. Process becomes the focus, not outcomes, with little incentive to ever solve the problem.
This encourages a bureaucrat-centred social-service ‘industry’ that grows ever stronger through alarmism and increased dependency, with substantial benefits to government connections. Next year’s funding (often a multiple of last year’s funding) is usually a higher priority than this year’s deliverables.
Good people find it hard to implement effective, innovative solutions because of all manner of expert-drafted guidelines and protocols and other tools of the bureaucratic service that are supposed to promote consistency, transparency and accountability. This ‘best-practice’ high-transparency model can work well when there is a single, clear solution to a specific social problem. But it’s a poor model for figuring out how to solve more complex social problems. The ‘government-pays-and-delivers’ model is extremely risk averse. No politician wants to be the sponsor of a failed social experiment – and the subject of journalists’ gleeful reporting. Such perverse incentives make it risky to innovate or achieve productivity improvements in social service delivery.
With a Social Impact Bond, potential solutions to complex social problems come bottom-up from a local community: people who, in other words, are familiar with the problem, and who work closely with service providers.
That’s precisely how you want it to be, with decisions made closest to the people with local knowledge and a stake in the outcomes.
Further, because private investors (including charities) carry the risk, they’re also going to be well motivated to do due diligence on the significance of the social problem, and on the management and delivery team tasked with solving it. They will be acting like ‘social venture capitalists’, meaning a better selection of problems and teams, and a better matching of problem and solutions. In principle, this constrains government spending because the decision to green-light a project is made by private investors, and because it unglues some links that hold together the various taxpayer funded social service industries.
An ideal Social Impact Bond will have two properties: (1) a well-understood and estimated potential taxpayer cost saving (from needing to build fewer prisons or hospitals in the future, perhaps); and (2) an easily verified target outcome. The bonds will also work best in situations with impacted local information and contingencies, so that solutions can be judged readily and adapted or terminated as necessary.
Under a bond regime, the private sector and private financial capital do what they are good at: namely the discovery of effective solutions and coordination of activities focused on a target outcome. The government sector then enters only to scale a successful solution up, to the wider social benefit.
To start this off, existing service providers might ask themselves what they would do if free to focus only on the problem, to be judged entirely by their solution to the problem, and to be privately financed to deliver it. Then they would seek to convince government that this solution saves them money, will cost them nothing unless it succeeds, and will lead to improvement in public sector service delivery.
Now although there are about 40 or so Social Impact Bonds underway or in development worldwide at the moment, including a few in Australia, they are still a largely untested idea. There will be costs involved in going first, and we can expect some steep learning curves along the way. But Ronnie’s idea is a good one: it provides incentives to focus on actually solving problems and doing so efficiently, not just on growing budgets.