Anything that evaluates criminal reformation programs rigorously, as this article from Atlantic Magazine suggests, is good.

An excerpt.

Every year, the government spends billions of dollars on programs designed to help America’s neediest citizens. In many cases, whether these programs work is anyone’s guess.

Less than $1 out of every $100 of federal government spending is “backed by even the most basic evidence that the money is being spent wisely,” wrote Peter Orszag, the former head of the Office of Management and Budget, and John Bridgeland, the former director of the White House Domestic Policy Council, in a 2013 piece in The Atlantic.

In their article, Orszag and Bridgeland advocate for a “moneyball for government,” arguing that an era of fiscal scarcity should force Washington to become more results-oriented.

A new partnership among New York State, 40 private investors, and a nonprofit called the Center for Employment Opportunities seeks to apply this sort of thinking to an area of policy that has been particularly resistant to interventions: lowering the recidivism rate in an era of growing prison populations.

The investors, including private philanthropists and former Treasury Secretary Larry Summers, have put up a total of $13.5 million to fund an expansion of the work that an organization called the Center for Employment Opportunities (CEO) already does with people coming out of prison. CEO’s model is simple: It prepares people who have criminal records for the workplace, gives them up to 75 days of temporary employment, and then helps them find jobs of their own. With the $13.5 million, CEO will work with an additional 2,000 clients, targeting the highest-risk people.

But the expansion of the program isn’t charity: The project is a so-called “Pay for Success” initiative, modeled after social-impact bonds, which were first used in the United Kingdom five years ago. The basic idea is that investors fund a program that has a promising approach, putting in place extensive data-collection points so that they can track the program’s results. The investors are betting on the idea that the program can do a better—and less expensive—job of providing a given service than the status quo. If they’re right, and the program meets certain expectations—in this case the benchmarks for success are to reduce recidivism by eight percent and increase employment by five percent—the government will have saved money in less prison spending. The government then pays back the investors with its savings. If the program succeeds, investors can earn a return. If it exceeds those goals substantially, investors can get a bigger return, which in this case is capped at 10 percent. The state at no point spends more money than it would have spent incarcerating the 2,000 individuals anyway.

The Pay for Success strategy isn’t just a way to test CEO’s model. It’s a way to bring careful, data-based monitoring of a program’s effect into government spending.

There are dozens of programs that seek to help people re-enter the community once they’re released from prison. They provide job training and housing assistance and college-preparation classes and counseling. But a lot of people still end up back in prison. About 700,000 individuals are released from prison nationally each year; the national recidivism rate is about 40 percent.