Andrew Keatts | @Andy_Keatts | November 23, 2015
Like the rest of the country, Austin is trying to figure out if an new financing model can crack some of the biggest social problems it faces.
Earlier this year, a group of local nonprofit service providers received $126,000 in grants from Third Sector Capital Partners, Inc. to study whether social impact bonds, also known as “pay for success,” could help solve three persistent local problems.
Through social impact bonds, private investors pay for government-run social programs, with the potential to get a return on their investment from the savings they those programs help create. The government repays the upfront investment only if the program meets predetermined benchmarks. Investors collect a return on their investment if the program is even more successful.
For example, an investor may fund a program designed to reduce prison recidivism. Ostensibly, a government would save more money than the program costs if it can reduce its prison population. The investor would receive a portion of those savings if the project succeeds.
The idea is for the government to test innovative approaches that might not otherwise happen – and at no cost to taxpayers if they fail. If the programs don’t work, governments don’t pay.
“For us, it’s a tool to implement evidence-based policy,” said Josh McGee, vice president of public accountability at the Houston-based Laura and John Arnold Foundation*, a foundation that has contributed funding to the Harvard Kennedy School’s Social Impact Bond Lab, which assisted with the country’s first two major social impact bond projects, in New York and Massachusetts.
“We want government to try new things and study what works at a greater frequency so we can learn what works, for whom, faster than we do today,” McGee said.
Still a new tool
It’s a hot financial topic, touted as a means of mobilizing private funds to solve intractable problems. But with only a few domestic examples, right now, they still rely more on hope – or hype – than known results.
“There’s more talk of them than there actually are examples of them,” said Donald Cohen, executive director of In the Public Interest, a research center on privatization and contracting. Cohen has written skeptically of the promise of social impact bonds. “They have an appeal because it sounds like a solution to the problem of underfunded social services. People desperate for funding will look to anything for help.”
Travis County, home to the Texas capital, is studying whether it could use the model to reduce teen pregnancies among Hispanic youth, improve the health of newborn African-American babies, and increase permanent supportive housing, a form of government housing for chronically homeless people that also provides ongoing services like health care and employment assistance.
The Austin-area’s exploration comes as the national conversation about the financing tool – once greeted with incredible promise – has been complicated by conflicting reports from a Utah program.
In a social impact bond program there, Goldman Sachs and a family foundation paid $7 million to send 600 kids to preschool in 2013. Of those, 110 were expected to need special education in kindergarten.
When only one of those students needed special education by first grade, the investors were paid back 95 percent of the $285,000 the public school system saved from the kids avoiding special education. They’re scheduled to keep receiving 95 percent of the district’s savings until their investment is paid back with interest, the Salt Lake Tribune reported.
But a New York Times investigation published earlier this month alleged the program’s success was trumped up, renewing debate over the model’s promise.
Another high-profile social impact bond has received renewed attention lately as well. Prior to the Utah dust-up, the country’s highest profile social impact bond was an attempt to reduce recidivism among juvenile offenders at New York’s Rikers Island prison. Officials this summer announced the program didn’t reduce recidivism of its participants.
Still, despite those apparent setbacks for social impact bonds, Travis County officials say they’re continuing to take a rigorous look at whether they hold promise. Officials acknowledge the fact that despite high hopes, the tool may not be a good fit.
“The answer could be ‘no, it doesn’t work,’” said Lawrence Lyman, director of Travis County’s health and human services division. “We either couldn’t find an economic model that works for an upfront investor, or that we can’t make it feasible for the end payer.”
Lyman says one of the biggest questions to sort out is a fundamental issue with the typical structure of social impact bonds.
Once you do the math, he said, it turns out that a program needs to produce “relatively huge” savings to be able to repay the upfront investment and pay for ongoing savings.
But there’s an even more serious issue to address, Lyman said. The hardest social problems to solve tend to involve multiple public agencies. Take, for example, a program aimed at reducing recidivism that would be implemented by a social services department. If successful, it would save the Sheriff’s Department money by reducing incarceration costs, but paying for the success might fall on the social services department that implemented it.
“I’m still just trying to figure out in my own head how the math can even work,” he said.
Another challenge, he said, is finding an intervention – a new approach to solving a social problem – that has a reasonable expectation of success. He said attempts at reducing teen pregnancy are the most speculative, while solutions for improving birth outcomes and supportive housing are better established.
“The more of an evidence base you have to start with, the more likely it is to work,” Lyman said.
But that introduces another hurdle: when there’s especially strong evidence that an intervention will work, it might make more sense for the public sector to just fund it outright.
So social impact bonds are looking for the narrow realm in which there’s enough evidence that a program can work, with enough ambiguity that a government is wary of self-funding it.
Still, none of the uncertainties scare off McGee, of the Arnold Foundation. It’s a tool for governments to combat social ills, he said, but it’s still a tool in its nascent stages.
One early lesson, he says, is the importance of beginning with a pilot stage that works out the kinks of a program before the countdown clock towards a program’s success or failure begins ticking. He also said rigorous evaluation metrics are integral.
Significantly, McGee said, the programs offers another clear benefit that no existing financing tool can offer local governments: it’s the only way a locality can avoid paying for a program that doesn’t work.
“It lowers the economic and political barriers to try new things,” he said. “We can say, ‘try this thing,’ that we might not otherwise.”
And New York’s willingness to declare the Rikers Island project a failure is a proof of concept in another way, too, he said. “It wasn’t delivering outcomes, and it was shut down,” McGee said. “That means government wasn’t overly-invested in a program that wasn’t working.”
Critics remain skeptical
Still, it’s exactly that sort of thinking – that social impact bonds as a concept are successful even when their specific program fails – that drives Cohen nuts.
“That’s true market thinking,” he said. “The market will determine success or failure, therefore anything is a success? It’s really tortured logic.”
By their nature, social impact bonds search for narrow, easily-defined interventions to address social problems. But those social problems are by their nature complex and multi-dimensional, he explains. An approach to recidivism focused on behavioral therapy ignores the many reasons a person does or doesn’t go back to jail, and they aren’t easy to isolate.
Likewise, he warns that they take an inarguable concept – metrics are good – and inflates what governments can learn from those measurements.
The dueling philosophies illustrate why, exactly, the finance, public policy, and philanthropy worlds – as well as Travis County –are all trying to figure out whether social impact bonds have potential.
“I don’t want to overstate things,” Cohen said. “I don’t know if they can turn out to be great. The world is a complicated place. If you need cash for something and this helps you get it, so be it. But ours is a deep skepticism because, in the end, nothing is free.”
Meanwhile, Lyman says he still has more questions than answers. “But another thought is, if it’s a catalyst to get new players together, then it’s a big win even if we never even sign off on a deal.”
*Note: The Arnold Foundation is a contributor to the Kinder Institute for Urban Research.