Even with the short-term budget deal reached last month, domestic discretionary spending will be billions of dollars below its level of just a year ago. For some of us, that’s a triumph; for others, it’s a tragedy. But we all can agree on the urgent need to spend as wisely as possible.

No field cries out for reinvention more than federal investment in youths who are out of school and out of work. These efforts on behalf of 7 million “disconnected youth” between the ages of 16 and 24 haven’t yet demonstrated great results. While that doesn’t mean we can give up on these young people, it does mean we need to rededicate ourselves to figuring out how better to serve them.

Here there’s an instructive comparison with early childhood programs. President Barack Obama and the U.S. Chamber of Commerce don’t agree on much, but they do agree on the value of preschool. The reason is high-quality research going back decades. Adults who participated in the Perry Preschool and Abecedarian Project initiatives in the 1960s and 1970s finished more school, worked more, earned more and committed fewer crimes than adults who didn’t. Economists such as Nobel Laureate James Heckman cite these studies to find that effective preschool programs can save more than $7 for every $1 spent. Critics point out that Perry and Abecedarian were small in scale, but bigger programs have also shown striking results.

Programs serving disconnected youth, however, have demonstrated less success. As MDRC’s Dan Bloom notes, many initiatives have increased immediate earnings or GED certificate attainment for young people, but these results fade over time. Heckman argues that the return on investment in teens seems lower because it is lower — teenagers have entrenched habits that are difficult and costly to change. This is a plausible theory, but it’s only one take. After all, some programs serving teens still in school do get results. For example, “career academies” — smaller schools inside schools that teach distinct skills and partner with employers — achieve impacts on earnings that are substantial (more than 10 percent) and sustained (persisting after eight years).

The problem may be that we haven’t yet tested the right approaches to helping disconnected youth.

One of the most promising programs serving out-of-school teens is called National Guard Youth ChalleNGe. It is centered on a five-month, quasi-military boarding school experience. Research shows that three years after finishing the program, graduates have earned more than they would have otherwise. That’s great, but we don’t know if those results persist or fade away. A study that answers the question of persistence wouldn’t cost much and it could tell us a lot about whether ChalleNGe is worth a new investment.

Let’s take another famous program: Job Corps. The most rigorous study of its effects, by Mathematica Policy Research, found that the $1.5 billion program fails a cost-benefit test. Yet Job Corps remains mostly unchanged, operating under a 1,362-page rulebook. Why not try some new approaches? Some Job Corps sites could incorporate elements from ChalleNGe. Others could replace the pursuit of the GED diploma, which has not been linked to higher earnings, with a full-blown residential high school. We could learn what works best and copy it.

In the past two decades, social entrepreneurs have brought tremendous innovation to K-12 education. Standouts such as KIPP schools achieve educational outcomes for poor children that many people thought impossible. Similar breakthroughs may be waiting for disconnected youth, but government programs too often repel, rather than invite, innovation. Instead of funding the same efforts over and over, Congress should shift resources into encouraging new approaches and testing whether they work.

In today’s era of fiscal showdowns and budget cuts, we have to improve our portfolio. We can’t evaluate everything, but we can do so much more to make sure our investments are the right ones. With the right choices today, we could be pointing a decade from now to programs for disconnected youth that are every bit as successful as those now aimed at early childhood education. There’s only one way to get there: Experiment and evaluate.

Robert Gordon served as the acting deputy director and executive associate director of the Office of Management and Budget during the Obama administration. He is currently a guest scholar at the Brookings Institution. Robert Shea is a principal in the Grant Thornton LLP Global Public Sector, where he leads the firm’s performance management community of practice. Previously he served as the associate director for administration and government performance during the second George W. Bush administration.