Rikers Island's social impact bond paid for prison education from savings.

Young people leaving Rikers Island reoffended at high rates, and the city paid the costs of that recidivism, yet prevention programs for them were chronically underfunded.

In August 2012 Goldman Sachs lent $9.6 million through a social impact bond: MDRC and the Osborne Association delivered education and therapy to 16-18 year olds on Rikers, and New York City repaid the loan only from the cost savings created by a real drop in recidivism.

Because repayment depended on a measured outcome, the risk shifted from taxpayers to private investors and philanthropy, financing a program that government might never have bought on inputs alone.

Pay only for measured outcomes, not promised programs

Government paid only for outcomes, not for inputs

Private capital absorbed the risk of a preventive social program

Measured recidivism savings made the result verifiable

A bank and philanthropy funded what tax dollars would not

Make the payer the beneficiary of the outcome: tie repayment to measured savings and private capital will fund prevention it would otherwise never buy.

The Rikers bond was the first social impact bond in the United States and helped seed the pay-for-success field, showing that private capital and philanthropy could fund prevention when repayment was tied to a measured outcome; the model later spread to dozens of projects in health, housing and criminal justice.