Over the last 20 years or so, the Supplemental Nutrition Assistance Program (SNAP)—colloquially referred to by its old label “food stamps”—has become one of the most important elements of America’s social safety net. My new research paper (with Dean Jolliffe and Juan Margitic), “Food stamps and America’s poorest” focuses on an aspect of SNAP that has received little or no attention, namely it efficacy in reaching America’s poorest and so lifting the floor to US living standards. While this is a novel feature in the context of the literature on poverty in America, the paper’s focus on the floor has deep roots in theories of distributive justice whereby a society’s progress is judged in part by its ability to enhance the economic welfare of the least advantaged group.
Assessing the impact of any social program on the floor poses methodological challenges, given that conventional surveys are not designed for this purpose. Even in high quality surveys, the tails of the distribution are not likely to be measured accurately. Recorded incomes in surveys can be very low for some household in a survey response period, but this can be deceptive about living standards. Our method recognizes that there are both measurement errors and smoothable transient income effects in the observed survey data. Some averaging is called for. The idea is to measure the floor as a weighted mean of those deemed to be poor, with higher weight on observationally poorer people.
The new paper implements an operational approach to measuring the floor in the US using cross-sectional national surveys spanning 30 years, and uses these data to study the impact of SNAP. We estimate the level of the floor before and after recorded receipts from SNAP at the household level. This requires some potentially strong assumptions, so we provide various robustness tests, including tests of alternative weighting schemes, possible behavioral responses to SNAP among the poorest and the sensitivity of our results to including non-positive incomes. We also provide confidence intervals for the floor, taking account of survey design.
The paper finds that America’s poorest gained from SNAP. Their mean gain exceeded mean spending on SNAP—implying that the program has historically done better than a universal basic income (UBI) with the same budget (and ignoring any differences in administrative cost, which could well be lower for a UBI). Alarmingly, we find that America’s floor has been on a long-term decline, although SNAP helped prevent further decline in some periods, notably during the 2008-11 financial crisis. However, the efficiency of SNAP in raising the floor has also fallen over time, such that the per capita amount received by the poorest in 2016 is about the same as mean spending per capita.
In summary, the paper provides the first assessment of America’s progress in lifting the floor of the distribution of real income, and whether the country’s largest antipoverty program helped do so. Alarmingly, America’s floor has been sinking over the last 30 years. SNAP has lifted the floor, but its efficiency in doing so has declined over time. Full SNAP coverage of the poorest would lift the floor appreciable, from around $5 a day (per person, or about one third of the official poverty line) to $8.