December 12, 2017
A set of seven recommendations for "pay-for-success" interventions identified by researchers at the University of Michigan can help governments and private investors work together to effectively use this emerging financing model that has bipartisan support.
Pay-for-success, also known as social impact bonds, is designed to address social problems, especially health and social welfare issues in vulnerable populations. The initiatives tackle issues such as homelessness, early childhood education, job training after prison, and housing improvements for kids with asthma.
Under the model, private investors cover the startup costs of new or innovative programs, and the investors get their money back from the public sector — potentially with interest — if the program is successful.
"Often, cash-strapped governments are challenged to maintain existing programs, let alone launch new ones, however promising they might be," said Paula Lantz, associate dean for academic affairs at the Gerald R. Ford School of Public Policy and director of the U-M Policies for Action research hub that conducted the analysis.
The recommendations are published in the current issue of Stanford Social Innovation Review.
In the past five years, excitement around pay-for-success has grown across the globe. Governments have used it to finance more than 87 projects in some 20 countries. The U.S. has seen 18 begin, with dozens more at some stage of development.
"We believe this model can provide funding to help vulnerable populations, and there's a lot of interest in the approach," said Samantha Iovan, co-author of the analysis and project manager at Policies for Action. "But what we're seeing is that these projects can be hard to get off the ground."
One of the reasons they're difficult to put in place is a lack of agreement on what makes an intervention or program a good candidate to finance through pay-for-success. Ideally, the program works — and investors get their money back. But when a program doesn't meet its goals and investors lose out, stakeholders can misplace their criticism on pay-for-success, rather than on the program that failed.
"In those cases, we should resist the urge to say pay-for-success doesn't work," Iovan said. "Pay-for-success is a financing model, not an intervention. The model itself works well, but the interventions need to be designed and implemented in a way that ensures their success."
The U-M researchers set out to identify common elements of programs that paired well with pay-for-success. To accomplish that, they developed a "landscape surveillance" system that identifies new projects and tracks their progress over time. The tool allowed them to examine every domestic and international pay-for-success initiative that has secured funding, signed contracts and launched services.
They found that interventions with the greatest chance for a successful payout:
1) Address an important public problem. They take aim at an issue of significant concern to local, state, or federal government partners and their constituents.
2) Use a tested approach. They've been shown to be effective through strong research evidence. Ideally that's through randomized controlled trials, but quasi-experimental studies can also work.
3) Save money, or are worth the money. They hold promise of cost-savings or cost-effectiveness for the public sector. Cost-effectiveness refers to generating benefits that are seen to be worth the costs.
4) Can be measured. It's possible to evaluate them through well-defined, quantitative metrics.
5) Succeed quickly. Outcomes should materialize within years, and not decades. For this reason, programs to reduce smoking in adolescents might not be good pay-for-success candidates if the desired outcome is a savings in health care costs later in life.
6) Are feasible to implement. They don't come with big administrative challenges. These programs shouldn't be put in place with shortcuts and they shouldn't employ people who don't have solid training.
7) Are not politically divisive. They face no significant political challenges. A program might meet all other criteria, but not appeal to stakeholders for political reasons. For example, an evidence-based teen-pregnancy-prevention program that would increase women's access to birth control might be met with religious objections from elected officials or community-based groups.
"Intuition and wishful thinking are not substitutes for high-quality intervention research," Lantz said. "If we're going to invest government time and investor resources to bring services to people in need, we should be willing to go the extra mile to ensure that pay-for-success interventions will actually bring positive results to program participants, taxpayers, government agencies, and private investors."
Funding was provided by the Robert Wood Johnson Foundation. Policies for Action is a Robert Wood Johnson program with hubs at five U.S. universities, including U-M.