President Obama and leading Republicans are all looking for ways to foster economic opportunity and tackle pressing social challenges, even as annual non-defense spending is on track to amount to the smallest share of the economy in a half century. While some of us rejoice and others wince at that trend, we can all agree that this means a smart Washington should be shifting funds away from programs that don’t work and into ones that do.

This is easier said than done, but there are promising ideas out there. One is “pay for success,” which seeks to put taxpayer money on the line only after programs have gotten results. That may sound too good to be true--and it could turn out to be. But it’s well worth trying.

The concept behind pay for success is not complicated. Rather than pay for a service up front, the government enters an agreement to pay only after it delivers specified results. Providers raise money from philanthropic or private-sector funders and a third-party evaluation is arranged. If the program achieves its intended outcomes, the funders get back their initial investment and a reasonable return.

If the program fails, the government pays nothing and the funders are out of luck. Contracts can be written so that the government pays only when it will realize corresponding budgetary savings downstream. In such cases, pay-for-success has the potential to finance useful services now without adding to long-term taxpayer costs.

Five years ago, there were no pay-for-success initiatives in the U.S. Today, there are four and counting. Most offer job training and placements for ex-offenders, hoping to cut incarceration rates over time. An effort in Salt Lake City, supported by the J.B. Pritzker Foundation and Goldman Sachs, would expand early childhood programs with the aim of reducing the number of costly special education placements. If the program works, kids will be better off and the state will save funds. Forty percent of any savings will go to investors, but sixty percent will go into expanded early childhood education.

You can see the potential. Supporters of the President’s plan to dramatically increase early childhood investment, and to pay for it with a new tobacco tax, can continue to fight for that approach. Critics can continue to argue that it’s a wasteful white elephant. While the two sides (and the two of us) fight that one out, pay-for-success is something we should all be able agree on. The approach can make it possible to expand early childhood programs while producing new evidence about how to run them effectively.

If the programs don’t work, taxpayers won’t get stuck with the tab.

States and cities are leading the charge on pay-for-success, and that’s as it should be. They get most of the savings from good crime prevention and pre-kindergarten programs.

That said, there’s a modest but potentially important federal role as well. The anti-recidivism pay-for-success efforts got their start with federal innovation funds (at the Department of Labor, in this case), even though this meant Congress allocated fewer resources for the usual block grants.

Absent a dedicated funding source, doing pay-for-success for early childhood education would require trims elsewhere--say, in business development programs that duplicate what the private sector already does. But that’d be worth it: Unlike too much of what government does, pay-for-success has the potential to make a measurable difference, for both beneficiaries and state budgets.

There’s a unique federal role in supporting pay-for-success programs that yield mainly federal savings. The obvious example starts even before pre-kindergarten--in the home visiting programs for at-risk pregnant moms which, evidence shows, can lead to healthier, smarter babies. Right now, Medicaid doesn’t support such success-based payments, but either the Administration or Congress should change that. If Medicaid pays only to the extent that a home visiting program can show savings, there shouldn’t be any federal cost at all.

Congress might even create an all-purpose Pay for Success Fund, along the lines that Conservative prime minister David Cameron created in Britain or that President Obama has proposed. In today’s world, any funds would have to be paid for by offsetting budget cuts. But such a deal would let the President put new funds into early childhood; let Republicans support successful social programs with private support; and make taxpayers pay only after the program delivered results.

There are surely other paths worth pursuing, too. But this much we know: a smart Washington might want to try pay for success.

Note: This article originally appeared in The Atlantic. It was written with Robert Gordon, former deputy director of the Office of Management and Budget and a visiting scholar at the Brookings Institution.