Douglas Clement offers yet another of his excellent interviews, this one with Hilary Hoynes, in The Region, published by the Federal Reserve Bank of Minneapolis (June 1, 2017).  Here are some tidbits.

Long-Term Returns to Food Stamps

\”Food stamps are a central part of the U.S. social safety net, and it’s a means-tested program—meaning that you have to have low income to participate. And, remarkably, it has remained fairly intact over the past 20, 30, 40 years while other parts of the safety net for low-income families have been restricted and reformed. Also, it’s federal—run out of the USDA—so it doesn’t vary a lot geographically. That’s helpful because it really provides a uniform floor across the United States. In very poor areas, even in states that don’t tend to provide a lot of assistance for the poor, food stamps create a kind of universal minimum across all places. It does, however, create challenges for doing evaluation because it doesn’t vary much across space, and it also hasn’t varied much over time. …

\”Food stamps started under President Kennedy. His first executive action was to start some pilot programs for food stamps. … Those pilot programs eventually led to passage of the Food Stamp Act in 1964. But it wasn’t until 1974, 10 years later, that subsequent legislation compelled all areas to implement food stamps. In that 10-year interim, Congress essentially said to U.S. counties, “We’re going to appropriate these funds for this program. If you’re interested in implementing this program, please apply and we will fund them, subject to our appropriation.” … This resulted in gradual rollout of food stamps across the almost 3,200 U.S. counties. …

\”The “rollout design” is one of the tools in our tool bag for doing evaluation. And, of course, we need to convince ourselves that that rollout was as good as random, that it wasn’t systematic, that certain areas had the rollout earlier than others. In our first paper on this, my co-author Diane Schanzenbach and I really dug into the nature of the rollout and the political economy behind it. At the end of the day, we were convinced that it was as good as random which places got food stamps earlier rather than later. …

\”In the paper in last year’s AER with Douglas Almond and Diane Schanzenbach, we took a long-term evaluation lens to this program. Food stamps rolled out in the ’60s and ’70s, so the cohorts affected, or not, in early or late childhood are in their early 50s today. This presented an opportunity to address a question that no one has ever looked at before in the context of food stamps: What are their long-run benefits? …

\”We couldn’t in our data know precisely which families were on food stamps, so it’s sort of an indirect estimate. But we know whether food stamps were implemented when these individuals were 2 or 4 or 14 or 20 years old. We essentially analyzed the data within that lens: How old were you when food stamps were rolled out in your county?

\”The headline finding was about health. We measured metabolic syndrome, which is essentially a range of conditions including high blood pressure, diabetes, heart disease and obesity. … And we found that the more exposure to food stamps that a person had, the lower their risk of metabolic syndrome in adulthood. In particular, the gains were greatest if the food stamps program was implemented before an individual was 3 or 4 years old. That period between in utero exposure—prebirth—to those first three or four years of life, was the age range where having more exposure to food stamps available led to a more dramatic reduction in the incidence of metabolic syndrome in adulthood. … Then we also looked at effects on human capital outcomes. …  In short, we found that better nutrition in early childhood leads to human capital improvement and better outcomes in adulthood, but that finding was limited to women.\”

On the Earned Income Tax Credit

\”The EITC is the most important anti-poverty program for families with children in America. It removes the most children from poverty, and it’s organized as an in-work benefit rather than an out-of-work benefit. Welfare, for instance, is an out-of-work benefit …  The whole idea of in-work programs like EITC is to respond to that and say, “Well, American voters would rather have a program that redistributes while encouraging work, not discouraging work.” The EITC operates as an earning subsidy on incomes up to about $14,000. For every dollar that you earn, if you’re a single mother with two children or a married couple with two children, for every dollar that you earn up to about $14,000, you get 40 cents added to that dollar through the EITC. It’s a quite powerful increase in your after-tax wage.

\”It still needs to phase out or everybody would get it, so there are some negative work incentives that are faced by higher-income workers at levels where the EITC phases out: between about $15,000 and $40,000—or $18,000 and $45,000, depending on your family size. And it’s phased out at a rate of about 21 cents on the dollar. So, if you earn an additional dollar, your EITC is reduced by 21 cents, a gradual phase-out.

\”Research shows that this program design has a dramatic effect on employment. When the EITC expands, you see more low-skilled workers, particularly single mothers, in the labor market. It has a very powerful effect on transitioning people from out-of-work to in-work. And in so doing, it lowers poverty rates, not just because you’re giving households a tax refund at the end of the year—and of course, if you give someone money, you’re going to reduce poverty—but just as important is the fact that by encouraging work, earnings go up in the household, and that also reduces poverty. It generates a roughly 2-for-1 reduction in poverty for every dollar of federal spending, and that’s every efficient. … It does redistribution within the tax code, rather than a sort of brick-and-mortar social welfare operation that is the model of the state-based social safety net.\”

Head Start Doesn\’t Fade for All Groups 

\”[M]any studies that look at the effects of Head Start (and, to some extent, of programs like Perry Preschool and others) have found that increases in cognitive test scores for children in the years they’re in Head Start seem to fade out once they enter school. … Is it possible that this fadeout is masking the fact that there are gains for some groups that somehow, in the global mean, seem to disappear?

\”It turns out that the story isn’t quite that simple; but we did discover that, yes, Head Start increases cognitive test scores, but those global mean results mask the fact that the gains are very concentrated at the bottom of the skill distribution. The test scores at the bottom of the distribution went up by a lot; whereas, test scores in the middle and the top of the distribution didn’t go up by very much. …

\”Fast forward in this Head Start impact study, and observe kids through grade 1. We found that overall, the fadeout occurs throughout the distribution for the full population. But by looking across groups based on maternal education, race, ethnicity and other characteristics, we uncovered this finding of much larger gains for a specific group: kids who enter Head Start as English language learners—that is, English is not the primarily language at home. And in this experiment, that turned out to mostly be Spanish speakers because of the population in the experiment. These results were contemporaneous when the kids are still in Head Start, but they also persisted through transition to elementary school. Fadeout didn’t occur.\”

Hoynes has written for the Journal of Economic Perspectives, where I work as Managing Editor, a couple of times. For those who would like to sample her work up close, the articles are:

Leave a Reply