Duncan Green recently took on the subject of Development Impact Bonds and impact investing in this blog post, and raised a few reasons for skepticism. As he anticipated, we didn’t really agree with the concerns that came up and wanted to explain why.
DIBs are based on Social Impact Bonds first launched in the criminal justice sector in the UK, and now being implemented in other sectors in the UK and in the US, Australia, Ireland, and elsewhere. There are high hopes in the impact investing world that SIBs will bring new money to social sectors, and generate cash savings (used to pay back investors) for government by funding preventative measures – two very promising features of SIBs and reasons why they might appeal to governments in particular. But these aren’t where the only or the biggest potential benefits of SIBs lie, or why CGD got interested in looking at how the SIB model could apply in development.
In his blog post, Duncan quotes a colleague who says that the Peterborough Social Impact Bond hasn’t really attracted new money. May or may not be true but this wasn’t the main point. The interesting thing about DIBs is their potential to do what the Peterborough Social Impact Bond is doing: bringing together the public, private, and non-profit sectors and aligning incentives towards achieving a social outcome.
It is too early to offer an assessment of how well the Peterborough SIB is working but the early evidence is that services are being managed well under the SIB – recidivism has gone down among Peterborough prisoners (6%) while it’s gone up nationally (16% - the numbers are here), and the program seems to be better than traditional public sector programs at identifying and responding to individual ex-offenders’ needs and the things that might lead to them to reoffend.
There’s a lot more on this at Social Finance´s website and Owen Barder and Toby Eccles have written blog posts, here and here, about how DIBs are not just a financing tool but a new business model for development.
On the question of whether the Peterborough SIB is attracting new money, this evaluation reports that a lot of the Peterborough investors are foundations using their endowments to invest (so not grants), or are foundations or individuals investing in criminal justice for the first time. And the SIB is attracting new money in the sense that preventative programs are getting funding under the SIB that they weren’t getting before. So SIBs/DIBs can change what gets funded in the first place (government pays only for proven results so doesn’t have to take the risk of paying for programs that don’t work) as well as how services get delivered (with more flexibility than traditional government contracts).
The second argument by Duncan’s colleague is about the challenge to identify sectors where reduction in future costs can be clearly demonstrated. It’s an advantage of SIBs/DIBs that they can help to shift more resources into prevention, reducing the costs of treatment later, but “future cost savings” is not actually the defining criterion for either SIBs or DIBs.
In the Peterborough SIB, the idea of saving money over time (preventing reconvictions rather than paying for prisons) makes it a good “value for money” case for the government, but the real focus is on whether the overall business model leads to a better social outcome (reduced recidivism). Success in the SIB, and repayment to investors, is based on the rigorous measure of that outcome, and the government is committed to paying for success regardless of what future costs savings turn out to be.
It would be even harder in developing countries than in developed countries to identify sectors where future cost savings could be calculated. In some sectors, for example education (see the case studies in the DIB Working Group consultation draft report) savings would be harder than for others to quantify. What’s more, in developing countries donors are providing external financing. It’s not a question of how much future spending they will save in a country but a question of whether current spending is being used as efficiently as possible.
And one final note: we’re not arguing that DIBs will make sense for every development problem –but they could help to solve a lot of current problems in development and are an approach worth testing.