The high-profile failure of a Social Impact Bond (SIB) experiment at Riker's Island begs the question: how can we de-risk investment in SIBs, and social programming generally? SIBs are largely predicated on private investors assuming the risk that programs will succeed at producing desired social benefits. In the Riker's Island example, Goldman Sachs lost $1.2M and Bloomberg Philanthropies lost $6M because a juvenile recidivism intervention failed to work. Some backers hailed it a success nonetheless, insofar as the government didn't have to pay for the program. But was it really a success? Will investors be more attracted to SIBs in the future?
Investors can only assume risks that they understand, and quantify. In the case of SIBs, it appears unlikely that Goldman, or other SIB investors were adequately able to evaluate the risk. And if they did try to assess the risk, they certainly didn't guess right. While the program underlying the Riker's SIB, Adolescent Behavioral Learning Experience (ABLE) intervention, was purported to be "evidence-based," the designation remains fairly arbitrary. What does it take to deem a program "evidence-based"? Does the program need to be 20% effective? 50% effective? 80% effective? Proven effective once? And how reliable does the evidence have to be? Indeed, a comprehensive academic meta-analysis of the program's core model, Moral Reconation Theory (MRT), suggests that the intervention was not very effective. In fact, the meta-analysis, based on 33 underlying studies, found an overall r-squared of .16 (a statistical measure of efficacy), indicating that the MRT intervention was barely effective in the past. The researchers even suggest that the MRT was "more successful with adult than juvenile defenders" in institutional settings, such as Rikers. Finally, the vast majority of the evidence on MRT was produced by Correctional Consulting, Inc. the for-profit company that invented the program.
So how can an investor, with no formal training in social science, be in a position to interpret fairly dense academic literature and make a judgment about the likelihood of success for a particular intervention? They can't. No more than an investor can personally assess the underlying creditworthiness of an individual loan or mortgage. These financial investors typically rely on a qualified third party like a rating agency or credit bureau to appraise that credit risk. This is absent in the structure of SIBs. The focus of measurement is ex-post, not ex-ante. And therein lies a major challenge of SIBs preventing mass adoption. The rigor, the measurement, happens on the back-end, not the front-end.
"This can unlock new pieces of funding, private capital especially," said Jim Anderson, who leads the government innovation program at Bloomberg Philanthropies. "It also brings a laser-like focus to measurable data. Everyone has an incentive. We're not doing this to feel good. We want positive results." The fact is, failures like Rikers won't unlock new pieces of funding like private capital; it'll scare them away. And that is the entire proposition of SIBs.
Here's what we can do about it:
- Bonds need ratings - standardized, independent, predictive assessments of risk and return. There is no other reliable way for investors to assume capital markets risk. To date, SIB investors like Goldman Sachs have assumed philanthropic risk, and maybe some reputational risk. But the only way that SIBs or any form of social finance can attract true risk-capital is if investors are able to pre-determine the likelihood of success for an investment. With advances in meta-analysis and evaluation modeling, like the Impact Genome Project® investors can derive directionally-right estimations of effect sizes of a social program, before it runs its course.
We believe that the future of SIBs and social finance is brighter than ever. But if we are to create a true social capital market, we must be more rigorous, intentional and outcomes-driven than the current approaches.