The housing voucher program Moving to Opportunity is one of the most famous experiments in the history of the social sciences. The study included 4,600 families from housing projects in Chicago, New York, Los Angeles, Boston, and Baltimore who volunteered to participate. Those families were randomly assigned: One-third got vouchers for housing in low-poverty areas and counseling to find it; one-third got vouchers they could use anywhere; one-third didn't get anything.

From the outlines of the study, you can see the point—does it help to move from high poverty neighborhoods to low poverty neighborhoods? Does it help more than just giving people a free choice? And does it make financial sense for the government to facilitate these moves?

The data it generated has been the foundation of work by some of the most famous researchers in the country. Last year Raj Chetty, a John Bates Clark medal-winning economist from Harvard, led a long-run study that found children whose families got the low-poverty voucher had incomes that were one-third higher in their twenties. Jens Ludwig of the University of Chicago led a study that found reductions in diabetes and extreme obesity among the same group. The data set has been used to investigate academic outcomes and changes to mental health. It factors into policy debates big and small.

But, it's limited. Volunteering to be in the study is a choice. The vouchers might be distributed randomly, but the cohort receiving vouchers is not. And that might matter. Eric Chyn, a Ph.D. candidate at the University of Michigan who will soon be joining the University of Virginia's econ department as an assistant professor, wanted to find out if it did.

"The analog I had in mind, one thing we've seen in other evaluations of social programs—in particular schools—is this pattern of reverse Roy," Chyn says, referring to an economic model of self-selection. "We've seen evidence that suggests that there are certain groups of children that would appear to benefit quite a lot from going to a charter school, but those type of kids are the type that never sign up to go to charter school."

Why? It's hard to say for sure, but even availing one's self of free or subsidized services comes at a cost. Volunteering to leave where you live—even if it's a bad place—requires desire and effort. Moving is logistically and emotionally difficult, and moving from a poor neighborhood to one that's not can involve even more dislocation, especially from support systems that develop in underserved areas. There's often a stigma attached to voucher recipients from the projects. Chyn thought there could be parallels between vouchers and charter schools.

"That would be one way of getting this pattern—you have these people who are incredibly motivated to sign their kids up for a charter school, but these are also the parents who spend time reading to their kids at home, and things like this," Chyn says. "There could be other parents out there in the world that, either they don't have the time and resources to spend time reading to their kids at home, and so, for them, they can't give their kids that type of investment, and those are also the type of parents who are so disadvantaged that they also could never figure out a way to get their kid into a charter school." (Take this piece about how Detroit parents take hours-long journeys to send their kids to good schools, for instance.)

The hardest people to reach are, in theory, the ones who need help the most, and are also the hardest to study. But Chyn found a parallel to MTO—a natural experiment—in which the families moved were not volunteers, because their housing projects were demolished as part of Chicago's Plan for Transformation. Using social assistance records, Chyn was able to identify children and their households that were forced to move—and ended up in lower crime, lower poverty neighborhoods—and a control group.

Chyn studied this unique data set to find out whether children from those displaced households saw benefits from being moved to opportunity (to borrow the title of his study). They did, in interesting ways.

First, the headline numbers: The displaced kids were nine percent more likely to be employed when they grew up, and had 16 percent higher earnings, which translates to an extra $45,000 over the course of a lifetime. This means, Chyn writes, that "the increased tax revenue associated with these earnings exceeds the (average) cost of relocating public housing residents."

The cohort of kids Chyn studied, though, is interesting in and of itself. First, he looked at children who were between the ages of seven to 18 at the time their housing project was demolished. So to begin with, it's not a very young group of kids. And the effects continued even as he sliced up the data, which is not necessarily what you'd guess based on the research on MTO.

"The inclusion of even older kids is important, because it's part of what I'm bringing that's new to the table, which is suggesting that… there's this positive benefit for these older kids," Chyn says. "That's new, and that's different. The latest results for the Moving to Opportunity study show that the kids who were very young when their families moved, but they didn't find any detectable benefit for the kids who were older. I actually do see the benefits for the kids in my sample. Which is interesting, because it implies that even though these kids were older, and they spent most of their time living in a disadvantaged area, changing their neighborhood can have a positive impact on them."

Another interesting thing happened to these kids after they were displaced: Ultimately, the poverty followed them. A year after moving, the displaced kids lived in neighborhoods with poverty rates that were 28 percentage points lower than their old neighborhoods. After three years, that was down to 13 percentage points. Eight years later, there was no detectable difference.

Nonetheless, the labor-market gains persisted.

"Households that were living in these buildings that were not slated for demolition, they move out of public housing over time. So that means that, over time, the difference between the displaced and non-displaced groups actually shrinks. So that's why this attenuating pattern happens." Chyn says. "That's important because it suggests that it's something during the childhood period, that exposure during childhood is important."

But again, Chyn looked at ages seven to 18. So the childhood period is important, but perhaps, for a particular group, that important timeframe runs later than we might assume.

The displacement wasn't a silver bullet. But that's also interesting in its way.

"The other avenue that people talk about is, well, maybe if we change people's neighborhoods, it could have an impact on criminal behavior. So I do talk in my paper about the role of crime potentially being a mediating channel," Chyn says. The study didn't show any differences in criminal behavior between the displaced and non-displaced kids, at least in the 13- to 18-year-old group. But looking beyond those years, there was a detectable difference. "Displaced kids have fewer arrests for violent crime [than non-displaced kids]," he says, which could explain the labor-market effects.

"You change the neighborhoods for these kids and, in the long run, because their neighborhoods change, their peer interactions change, that we're able to have some impact on the propensity to commit crime, and that may mean that you're less likely to be in jail, and that you're more likely to be working because of that," he explains.

Specifically, Chyn found that displaced kids were arrested for property crimes 16 percent more as adolescents, but had 14 percent fewer arrests for violent crime in the long run. The former isn't ideal, but that would be a good tradeoff.

Chyn's opportunity came from a major shift in housing policy. Concentrated, high-rise housing projects are disappearing, and right now vouchers are en vogue. But this makes his research no less relevant.

"When you give people housing vouchers, and about 50 or 60 percent, in general, actually end up using the housing voucher," he says, citing a paper in which people who did not use vouchers said it was too complicated and difficult to use them. For instance, it takes extra effort to find a landlord and communicate with him or her to get the voucher to work.

He adds, "We need to know more [about whether the vouchers are too hard to use] and if they are, is that the reason that people don't want to participate in these programs, despite the fact that they could be very beneficial? If that is the case, we could think about ways to make vouchers easier to use, and things like that."