Nearly half of 16-to-18-year-olds released from New York City’s Rikers Island prison are back within a year—a trend the government doesn’t have the cash to reverse. So last fall, Mayor Michael Bloomberg inked an unusual contract with Goldman Sachs. The bank would put up a $9.6 million “investment” to teach 10,000 young offenders moral reasoning before August 2015. If these teens stay out of jail, the city saves money and Goldman Sachs will make a 22 percent profit. If not, the government pays Goldman nothing.
The new social-impact bonds—also known as pay-for-success contracts—use private-sector money to run prevention programs that would save the government money down the road. Taxpayers reimburse investors only if a program saves money compared with government services, bringing market discipline to the social sector and, advocates hope, attracting long-term funding to proven interventions. After Rikers, the first SIB contract in the United States, others followed: President Obama requested $485 million for SIBs in his last budget, and several federal agencies have already made money available, including $20 million from the Labor Department. Massachusetts has authorized $50 million to finance two social-impact bonds, New York state has requested suggestions, and Illinois plans to do the same.
At this early stage, there’s no standard SIB formula. But the basic structure may mirror the Rikers Island contract. In that case, a research organization called MDRC identified cognitive-behavioral therapy as a proven approach to reducing recidivism, citing studies such as a 2007 meta-analysis from the Vanderbilt Institute for Public Policy Studies in which teaching detainees how to better think through their behavior reduced recidivism by 25 percent. MDRC acted as an intermediary, coordinating the deal between the city and Goldman. It will manage the loan and oversee the Osborne Association, a nonprofit prison-reform organization, as it brings cognitive-behavioral therapy to Rikers.
Meanwhile, an outside group, the nonprofit Vera Institute of Justice, will evaluate how much time therapy recipients spend back at Rikers within two years of their release. A similar population first detained at the jail five years ago will be the comparison group, and savings will be calculated using the cost of a day’s incarceration. If recidivism drops by 8.5 percent, the city will save less than $1 million and pay Goldman only $4.8 million. If recidivism falls by 10 percent, the city will see similar savings but will return the bank’s $9.6 million investment. If recidivism drops by 20 percent, the city will save $20.5 million and Goldman could make a $2.1 million profit.
It’s an ingenious bit of innovation with little cost to government—but it’s unclear how many private investors might buy in. Goldman itself was leery of a high-risk, low-return loan, so Bloomberg Philanthropies offered to guarantee a $7.2 million repayment even if the intervention fails. The foundation is also covering MDRC’s costs. Without a deep-pocketed philanthropy to back the investment, the risk could make financiers skittish. “These first deals could easily end up being $500 million,” says Harvard’s Jeffrey Liebman, director of the Social Impact Bond Technical Assistance Lab at Harvard’s John F. Kennedy School of Government. “The question is, is there $5 [billion to] $10 billion ready to do this?” For that kind of money, SIBs may need to rely on wealthy individuals and corporate social-responsibility programs. The Rockefeller Foundation and J.P. Morgan’s investment bank estimate that market size for impact investing will reach between $400 billion and $1 trillion by 2020.
It’s also unclear how many programs could use this kind of financing. Interventions must be backed by substantial evidence—ideally, multiple randomized or quasi-experimental trials. They must be scalable, able to deliver a quantifiable impact on government spending by a given date, and inexpensive to evaluate. It’s much easier to calculate successful payments if savings are captured by a single government agency—a rarity, because recipients typically draw on social services from all levels of government.
Programs that reduce time in jail, emergency-room visits, and homelessness are three possibilities for an initial wave of SIB financing. For other programs, “the time frame is too long, and the equation is too complex,” says Mark Rosenman, director of Caring to Change, which promotes more-effective grant-making. Advocates have floated broader access to high-quality prekindergarten as a candidate for SIB financing, but it may not be a good fit: Pre-K education potentially saves money the government would have spent over a recipient’s lifetime, not over a few childhood years.
Once the first social-impact bonds provide proof of concept, governments and investors may be more willing to take this route, and the time frame of contracts may grow, says Bill Pinakiewicz, vice president of the Nonprofit Finance Fund’s eastern region. Social services are under increasing pressure to quantify outcomes. “The ability to ascribe economic value and to actually monetize social impact” is a conceptual breakthrough, he says. “The risk in this thing is that all the economic value comes from savings, avoided costs.” There is not yet a holistic way to measure the value of changing a Rikers prisoner’s life. Tying payments to a specific metric could also backfire, perhaps by leading providers to cherry-pick participants most likely to succeed.
Social-impact bonds have promise, but they probably can’t meet the entire demand for social services. And their reliance on philanthropic guarantees could cannibalize money that would otherwise be spent on nongovernmental programs. So, for now, SIBs are the solution to only a narrow set of problems.