Beats | July 6, 2015

The Prison Reform #Fail That is Shaking the Social-Impact Bond Market

The team at


Green bonds have grown into a more than $40 billion annual market, while social-impact bonds have had trouble getting out of the starting gate.

It doesn’t help when early and high-profile offerings not only fail to deliver returns to investors, but fail to deliver the social impact that the bonds were designed to finance in the first place.

Last week’s news that the plug had been pulled on the first U.S. social impact bond, or SIB, shook up the nascent marketplace. The failure of the New York City project, launched in 2012 to reduce the rate at which juvenile offenders go back to jail after release from Riker’s Island, followed last year’s report that the Peterborough, England, granddaddy of all social impact bond-financed projects, failed to meet the 10 percent recidivism-reduction targets needed to repay investors, though it did show an 8.3 percent drop.

I haven’t had the chance to dig in as deeply as is necessary to piece the full story together. Some of the details will have to come from New York City reporters and bloggers who know their way around City Hall and the ongoing scandals at Rikers Island and can reconstruct how the project came unglued. For impact investors, the important question is what can be learned for future SIBs, which account for less than 1 percent of impact assets under management, according to a recent report.

The New York City project was always a chronicle of a death foretold. It was intended by then-Mayor Michael Bloomberg as a conversation-starter and demo of a concept that could be picked up elsewhere. From its initiation by the mayor’s offices, through its underwriting at Goldman Sachs, to its loss-guarantee of more than 75 percent by Bloomberg Philanthropies, is was one-of-a-kind.

One of the outcomes of the deal is that the billionaire’s Bloomberg Philanthropies will now cut a $6 million check to Goldman Sachs to repay three-quarters of the bank’s investment in the program. Goldman took a $1.2 million (or 17 percent) loss on a $7.2 million investment. If the program had continued for a fourth year, Goldman’s exposure would have increased to $2.4 million, or 25 precent, as the loss-guarantee from Bloomberg Philanthropies dropped to 75 percent (or $7.2 million).

(Note: The New York City project shouldn’t be confused with a separate social-impact bond sponsored by New York State that also aims to reduce recidivism. The state project, financed with $13.5 million raised from investor by Bank of America Merrill Lynch and launched last year, was the subject of ImpactAlpha’s piece last December, “Rebuilding Lives, Reducing Costs: Social Impact Bond Finances Jobs Not Jail.” More on the differences between the projects below.)

Change of Plans

The program’s fate may have been sealed even before it began when the operator originally sought to run the program was replaced on the eve of the financing. One person in a position to know told me back in 2012 that Linda Gibbs, then deputy mayor of health and human services (and now a principal at Bloomberg Associates) had reached out to Barbara Gunn, president and CEO of Structured Employment Economic Development Corp., or Seedco, to run the program.

Seedco was widely reported as a Bloomberg administration favorite. And Andrea Phillips, the Goldman vice president who led the social-impact bond financing, had previously been president of Seedco Financial, a community development financial institution (CDFI) that was spun off from Seedco to provide capital to small businesses and nonprofits in disadvantaged communities, and before that, executive vice president for programs at Seedco.

Then Manhattan U.S. Attorney Preet Bhahara filed civil fraud charges against Seedco for filing false reports of job placements to meet goals of a separate federal contract contract goals.

“How could you trust Seedco to run a program about measurement and results, when they’re in paper for cheating on the test?” this person said at the time. “So Seedco was quietly let go.”  (Seedco paid a $1.7 million fine and six agency managers settled with the government. The trial of a seventh ended in April with the jury finding one manager accountable for a small number of false placement reports.)

When the deal was announced, MDRC, more of an advisory and intermediary than an operator, was named to run the program. The Vera Institute of Justice, a research and technical assistance organization, was selected as the evaluator. “Vera and MDRC are in non-typical roles,” said my source, whom I’m leaving unnamed since the conversation was so long ago. “They’re not in the seats you’d expect to see them.”

Scaling Challenges

The principals are not saying much beyond their official statements. MDRC said it contracted with The Osborne Association, which ran the program in collaboration with Friends of Island Academy. Instead of a jobs program, the groups offered Cognitive Behavioral Therapy (CBT) and in particular, Moral Reconation Therapy (MRT). MDRC says those programs had proven effective elsewhere, but “had never been implemented at scale in as challenging an environment as Rikers,” and detailed some of the obstacles.

On its website, MDRC describes itself as an “intermediary” that designs, evaluates and offers technical assistance “to build better programs and deliver effective interventions at scale.”

Vera compared “recidivism bed days” for 16- to 18-year-olds eligible to participate in the program in 2013 with a similar group of Rikers’ prisoners from 2006 to 2010, before the program was established. The program reached most of its target population, and 44 percent of participants hit at least one milestone but concluded bluntly, “The program did not lead to reductions in recidivism for participants.” That is, unlike the Peterborough pilot, the New York City program did not show any statistically significant reduction in recidivism.

In the Huffington Post, Phillips, who leads Goldman’s Urban Investment Group, and James Anderson, who leads Bloomberg Philanthropies’ government innovation portfolio, put on a brave face to say, in effect, the patient died and the operation was a success. They acknowledged the “evidence-based cognitive behavioral therapy that’s been effective in reducing recidivism in many other correctional settings, did not work at Rikers Island.”

But Anderson and Phillips stress that as a public-private partnership to test an innovative model, the program actually succeeded. “And we can now say that this unique financing model made it all happen — and stands ready to help public sector leaders finance more efforts to better serve people who need support and better results.”

Increased Scrutiny

Talking points from other proponents of social-impact bonds, or pay-for-success contracts as they’re more accurately called, also said the pilot demonstrated the essential aspect of social-impact bonds: that risks are borne by private investors, not taxpayers. On its site, the Nonprofit Finance Fund stated, “Taxpayers did not pay for a program that appears not to have generated the desired outcomes set out at the pilot’s launch. That risk was borne by investors willing to put their capital at risk for the sake of demonstrating whether a social program achieved a societally beneficial result.”

“Great to see stakeholders happy but have we learned WHY this SIB failed?” tweeted Cathy Clark, director of the CASE Initiative on Impact Investing at Duke University and author of The Impact Investor. “Where is story on intervention learning?”

The end of the Rikers Island program puts other social-impact bond-financed projects under increased pressure to deliver. Investors in the New York State anti-recidivism SIB, including former Treasury Secretary Larry Summers, will be closely watching for the results of that program, which target older ex-offenders returning from state prison and offered job training and job placement.

Goldman Sachs itself has invested in three other social impact bonds, including to tackle early childhood education in Utah and Chicago and homelessness in Massachusetts, that will report some results later this summer. Other financial institutions are also underwriting deals. Bills have been introduced in the Senate and House to target federal dollars to social impact bonds.

It’s too early to say the bloom is off the rose for social-impact bonds. The New York City experience may say more about the challenges in helping ex-offenders avoid the revolving door back to prison than it does about the efficacy of social-impact bond financing. Most early SIB investors are interested less in the financial returns than in the promise of early intervention to address chronic social challenges. If those promises fall flat, so too may the innovative financing model that has so far produced more fanfare than results.