Research

We investigate the intersection of family size, food security, and the efficacy of public benefits, especially with respect to the Supplemental Nutrition Assistance Program (SNAP). Food security literature pays scant attention to the role of number of children in a household – an important dimension for understanding family resource and food assistance adequacy in the context of child well-being. We exploit longitudinal food security data within the Panel Study of Income Dynamics to explore how food security status and family resources change when families change in size, particularly with the addition of children. Gundersen et al. (2018) flagged large families as a group for consideration in any future SNAP reform, which motivates the need for evidence on the dynamics of family size, program benefits, and child food security. We focus on the subsample of SNAP recipients to address the question of how well program benefits meet the needs of families of varying sizes, as defined by a geographically price-adjusted Thrifty Food Plan, as well as how an additional family member, or child, affects the probability of being food insecure and how family size intersects with the likelihood of being extramarginal (no food spending beyond SNAP assistance), the size of the average food resource gap between spending and needs, and the adequacy of SNAP benefits in meeting food needs. Our findings provide key insight on the responsiveness of food assistance programs to changes in family composition and needs. Importantly, this study supports future research and policy design with respect to child well-being in larger families.

This paper introduces a new measure, the probability of food security (PFS), to study food security dynamics in the United States. PFS represents the estimated probability that a household's food expenditures equal or exceed the minimum cost of a healthful diet, as reflected in the United States Department of Agriculture's Thrifty Food Plan monthly cost estimates. PFS matches the official food security prevalence measure in a given period but enables richer study of the dynamics and severity of food insecurity. Applied to 2001-17 data from the Panel Study of Income Dynamics, we find that roughly half of households that become newly food insecure resume food security within two years. But the positive association of persistence with prior food insecurity means that half to two-thirds of food insecure households at any given time remain food insecure at least two years later. PFS varies dramatically with income and demographic characteristics, such that inter-group prevalence and severity measures differ by one or two orders of magnitude. Households headed by non-White women with low educational attainment disproportionately suffer persistent, chronic food insecurity, while White-headed households without a college degree account for most of the business cycle-associated variation in national food insecurity.

In the United States, almost one in seven households with children have limited access to food. The problem of food insecurity is closely tied to a household’s financial circumstances. Yet, prior research has paid insufficient attention to the financial risk factors beyond poverty that impact food insecurity. Lack of liquid financial assets may compromise a household’s ability to smooth consumption during income shortfalls, while debt obligations, such as debt from credit cards or medical bills, may deplete financial resources and constrain food budgets. Using longitudinal data from the Panel Study of Income Dynamics (PSID) and mixed effects growth curve models, I estimate associations of household debt and liquid assets with food insecurity among households with children. Additionally, I disaggregate household debt by amount and type of debt incurred, including debt from housing, student loans, credit cards, medical bills, and other sources of unsecured debt. Results indicate significant heterogeneity in wealth and debt profiles based on food security status. Further, debt from unpaid medical bills, other sources of unsecured debt, and student loans increase the odds of household food insecurity net of income and other household characteristics. I also find that lack of liquid assets is an important determinant of food insecurity, independent of household income and debt. Considering the full spectrum of household finances, including lack of financial assets and specific debt obligations, is essential for identifying at-risk households and alleviating the problem of food insecurity.

Using data from the Consumer Expenditures Survey, we document the level and volatility of quarterly consumption across the socio-economic distribution. While the measurement of economic well-being in the United States is focused on income, the secular and policy discourse prioritizes income-adequacy to meet family needs. This concern over income adequacy centers on the capacity of individuals and families to predictably consume minimally acceptable levels of basic needs, and the social and economic mobility consequences of low levels of consumption. Our results show a clear socioeconomic and demographic gradient of lower consumption levels amid higher consumption volatility for economically disadvantaged groups. Of particular concern is the stylized fact that, among the categories we track, food and clothing exhibit relatively high levels of consumption volatility among low-income households.

I examine trends in the material well-being of working-class households using data from the Current Population Survey in the two decades surrounding the Great Recession. Average earnings, homeownership, and insurance coverage all fell, while absolute poverty and food insecurity accelerated leading up to the Great Recession. After-tax incomes were stagnant for much of the distribution across and within skill groups. On the contrary, the economic hemorrhaging either abated or reversed in the decade after the Great Recession, especially for the least skilled and for households headed by a Hispanic person. This includes robust earnings growth resulting in falling lower-tail earnings inequality, absolute poverty, and food insecurity, coupled with increased insurance coverage and a modest rebound in after-tax incomes. As many of these recent advances likely stalled with the onset of the Covid-19 Pandemic, I discuss various policy options.

I compare the extent of food hardships in the United States among all adults, and separately for seniors, in the two decades before and during the onset of the Covid-19 Pandemic. The data come from the 2001-2019 December Supplements of the Current Population Survey, as well as the newly released Census Household Pulse Survey. The results indicate that food insufficiency among all adults increased three-fold during the Covid period compared to 2019, and more than double that observed during the Great Recession. Over 1 in 5 Black adults were food insufficient in mid 2020, a rate double that of white adults. Food insufficiency among seniors increased 75 percent during the Covid period, but when broadening the definition to also include reduced variety of foods, the share of seniors food insufficient also more than doubled compared to 2019 and the Great Recession. Receipt of charitable foods among disadvantaged adults spiked over 50 percent in the Covid period, but the initial response among seniors was a sharp reduction, before rising. These patterns, which hold in richly specified regression models, are consistent with strong shelter-in-place and other social distancing measures enacted at the state and local levels in response to the Pandemic that were gradually relaxed over time.

There is a large literature on earnings and income volatility in labor economics, household finance, and macroeconomics. One strand of that literature has studied whether individual earnings volatility has risen or fallen in the U.S. over the last several decades. There are strong disagreements in the empirical literature on this important question, with some studies showing upward trends, some downward trends, and some flat trends. Some studies have suggested that the differences are the result of using flawed survey data instead of more accurate administrative data. This paper provides an overview of a project attempting to reconcile these findings with four different data sets and six different data series--three survey and three administrative data series, including two which match survey respondent data to their administrative data. Using common specifications, measures of volatility, and other treatments of the data, the papers show almost uniformly a lack of any significant long-term trend in male earnings volatility over the last 30 years. Moreover, the survey and the administrative data almost entirely agree on that long-term stability when the comparison is done properly. Several possible explanations for the differing finds in past work are suggested by the papers. The stability of earnings volatility raises many questions for future research on trends in the U.S. labor market.

Previous research has shown that investments during the early childhood period are likely to have the highest social return. We use administrative data from Virginia to document participation in SNAP and TANF among children born between 2007- 2010 during their early childhood period, which we define here as birth to age six. We find that participation in SNAP is about four times greater than participation in TANF and that most children begin their connection with the social welfare system in their birth year. Children who participate earlier in life tend to stay connected over a longer portion of the early childhood period, although SNAP participation peaks around ages 3-4 while TANF peaks earlier, around ages 2-3. In terms of joint participation, most households on SNAP do not receive TANF and about 1 in 12 children on TANF do not receive SNAP. Finally, over the early childhood period, on average, just under 1 in 2 children in Virginia participated in SNAP or TANF but demography plays an important role in this process: The level of cumulative receipt is 1 in 4 among White children, 1 in 2 among Hispanic children but rises to 3 in 4 for Black children; cumulative receipt is also higher in nonmetropolitan counties than metropolitan counties. This study documents the reach of the existing social welfare system during the early childhood period, underlining the importance of race and space in 21st century America.

Health insurance improves health and reduces mortality. Expanding insurance is a central feature of the Affordable Care Act (ACA). Persons who use drugs (PWUDs) have historically been at high risk of being uninsured. It is unknown if Appalachian PWUDs, who live in an extremely economically distressed region, are more likely to be insured since implementation of the ACA. Data from a cohort of 503 PWUDs from eastern Appalachian Kentucky, who were interviewed at seven time-points between 2008 and 2017, were analysed using mixed effects regression models. At baseline, only 33.8% of participants were insured, which increased to 87.3% of the cohort at the last follow-up interview. The final multivariate model, which included baseline characteristics and interactions by time, indicated there were significant baseline differences in insurance status by gender, age, education, income, and history of injection. Differences in the predictive margin probabilities of being insured across these groups had dissipated by the final follow-up interview. After Kentucky’s implementation of the ACA, this cohort of Appalachian PWUDs made substantial gains in obtaining insurance that far exceeded the increases reported in national studies.

As of June 2020, the coronavirus pandemic has led to more than 2.3 million confirmed infections and 121 thousand fatalities in the United States, with starkly different incidence by race and ethnicity. Our study examines racial and ethnic disparities in confirmed COVID-19 cases across six diverse cities – Atlanta, Baltimore, Chicago, New York City, San Diego, and St. Louis – at the ZIP code level (covering 436 “neighborhoods” with a population of 17.7 million). Our analysis links these outcomes to six separate data sources to control for demographics; housing; socioeconomic status; occupation; transportation modes; health care access; long-run opportunity, as measured by income mobility and incarceration rates; human mobility; and underlying population health. We find that the proportions of black and Hispanic residents in a ZIP code are both positively and statistically significantly associated with COVID-19 cases per capita. The magnitudes are sizeable for both black and Hispanic, but even larger for Hispanic. Although some of these disparities can be explained by differences in long-run opportunity, human mobility, and demographics, most of the disparities remain unexplained even after including an extensive list of covariates related to possible mechanisms. For two cities – Chicago and New York – we also examine COVID-19 fatalities, finding that differences in confirmed COVID-19 cases explain the majority of the observed disparities in fatalities. In other words, the higher death toll of COVID-19 in predominantly black and Hispanic communities mostly reflects higher case rates, rather than higher fatality rates for confirmed cases.