Research

Using data from the Panel Study on Income Dynamics (PSID), this small grant conducted three studies designed to increase understanding of food-related hardships among older Americans. Study 1 found that persons who had a work limitation or were food insecure in midlife (ages 40-54) had significantly increased odds (OR: 2.20, p<.05 and OR: 4.23, p<.01, respectively) of living in a food insecure household at age 60 to 69, holding all else constant. Those who worked more during midlife had significantly reduced odds (OR: 0.26, p<.01) of living in a food insecure household in their 60s. Study 2 found that older adults who were currently living in a food secure household had higher odds of healthy aging at age 60 to 69. Having higher midlife income or more time employed in midlife was associated with increased odds of healthy aging, while having a work limitation or a limiting health condition in midlife was associated with lower odds of healthy aging.  Study 3 used an instrumental variable approach to examine the probability of living in a food insecure household among households as their family units receive Social Security at retirement age. Study 3 found limited causal evidence that receipt of Social Security at retirement age is associated with an increased probability of living in a food secure household among the full population. These findings were robust to changes of the dependent variable or the endogenous variable but were sensitive to some of the expansions or contractions of the sample.

Senior participation in the Supplemental Nutrition Assistance Program (SNAP) has traditionally been lower than other groups, with historical estimates below 50 percent. We examine the relationship between state SNAP policy changes occurring over the 2001-2014 period and SNAP participation as well the relationship between SNAP participation and a variety of health-related outcomes for senior and non-senior households. Data from three separate sources, the 2001-2014 December Current Population Survey Food Security Supplement (CPS-FSS), the 2002-2014 Health and Retirement Study (HRS), and the 2001-2014 National Health Interview Survey (NHIS), are used to conduct our analyses. We construct simulated eligibility variables (SEVs) and SNAP policy indices to capture differences in state SNAP generosity. In both our CPS-FSS and our HRS samples, we find that eligibility expansions lead to increases in non-senior SNAP participation with some evidence of smaller increases in senior participation. Overall, seniors and non-seniors appear to respond differently to various policies, but this difference is not consequential enough to explain the large participation disparity between the two groups. We also find that SNAP participation is associated with worse health-related outcomes for all samples in all three datasets, which is likely due to negative selection into program participation. The signs of these relationships reverse when we use our SEVs as instruments for SNAP participation. However, these coefficient estimates are not statistically significant and the strength of the SEV as an instrument for senior SNAP participation depends on the dataset and sample used, as well as whether or not we use sampling weights.

The goals of this program of research are to estimate (1) the sociodemographic predictors of food insufficiency among seniors ages 60 and older and (2) the causal impacts of Social Security, Medicare, and Medicaid on food insecurity and/or insufficiency among seniors. I use data from the Health and Retirement Study (HRS), Current Population Survey (CPS), National Health Interview Survey (NHIS) and American Community Survey (ACS). Analyses using HRS data show that, consistent with earlier studies, age, income, work status, disability, education, and race/ethnicity are all significant determinants of food insufficiency; neuroticism is also a significant predictor of food insufficiency. Exploiting the Social Security “notch” in benefits that resulted in lower payments to individuals born just after January 1, 1917 compared with those born just before, I find using HRS data from 1995 that lower income is associated with higher food insufficiency and SNAP use, but these results are imprecisely estimated. Next, using NHIS and HRS data from multiple years, I find no evidence that becoming age-eligible for Medicare at age 65 reduces food insecurity. Finally, a difference-in-difference analysis of the Affordable Care Act’s Medicaid expansion using ACS and CPS data shows that despite significant increases in Medicaid among seniors in states that implemented expansion compared with those that did not, food insecurity among seniors did not decline. These results suggest that public health insurance does not reduce food insecurity among seniors, perhaps because this benefit is not fungible.

This project explores the correlates of geographic and temporal variation in food security using data from the 2008 to 2018 Current Population Survey’s Food Security Supplements.  The focus is on the relationship between State-level availability and accessibility of congregate and home-delivered meal programs, as well as the Supplemental Nutrition Assistance Program (SNAP), the Senior Farmer’s Market Nutrition Program(SFMNP) and the Commodity Supplemental Food Program(CSFP) on food security among lower-income households headed by older adults (ages 60 and up).  Results show some evidence that a State-level food security infrastructure plays a role in the food security outcomes of households headed by older adults.

Housing wealth is the primary source of wealth for many older adults, particularly those with lower incomes, who are more at risk of severe forms of economic hardship such as food insecurity. For housing wealth to directly improve food insecurity, it first must be liquefied. Understanding the role of housing wealth requires careful consideration of home equity and mortgage borrowing. A primary contribution of our report is to model the financial mechanisms through which housing wealth and its components—home value, home equity, and mortgage debt—affect food insecurity in older age. We use panel data on households from the Health and Retirement Study and instrumental variable linear probability models with household fixed-effects to assess the effects of new mortgage borrowing on food insecurity (N=20,421 household-years). Trend analyses reveal that food insecurity increased from the 2008 recession until 2014 and that new mortgage borrowing peaked prior to the recession. The proportion of older homeowners facing credit constraints is highest for those age 65 to 69 in all years. Regression results show that mortgage borrowing has a substantial short-term negative effect on food insecurity. Each additional $10,000 borrowed is associated with reduction of food insecurity of 2.2 percentage points. The effect of new mortgage borrowing on food insecurity is distinct from changes in house prices or changes in home equity, neither of which are statistically significant factors. In a simulation of the effects of relaxing the debt-to-income borrowing constraint, we find that food insecurity is reduced by 2.1 percentage points for previous non-borrowers and by 1.6 percentage points for borrowers. Results support the importance of access to mortgage borrowing to reduce material hardship in older age.

We examine the relationship between the Earned Income Tax Credit (EITC) and Black-White after-tax income inequality from 1980-2020. The EITC lowers overall inequality by 5-10 percent in a typical year, improving the incomes of Black households relative to White households in the bottom half of the distribution. Gains in relative economic status emerged after the 1993 EITC expansion, concentrated among working class Black households, and not extending to those at the very bottom. Estimating the effect of the 1993 expansion on labor supply, we find evidence of a much larger extensive margin employment response for Black households than White households.

This paper examines the effect of the Child Tax Credit (CTC) on the labor supply of single and married mothers using the numerous policy reforms in the credit generosity and eligibility criteria since its inception in 1997. I use variation in the simulated benefits for a nationally representative sample to estimate the labor supply response at the extensive and intensive margins. Using 25 years of data from the Current Population Survey my results suggest that an increase of $1,000 in the average CTC benefit leads to a 1.7 percentage point increase in employment of single mothers and a 35-hour increase in annual work. I find a greater effect when I restrict the sample to single mothers with high school or less education. For married mothers, I find that an additional $1,000 in the average CTC benefit is associated with a 28-hour increase in annual work. The results are robust to an alternative identification strategy using only variation in the maximum credit across time.

This study estimated the effects of welfare reform in the 1990s, which permanently restructured and contracted the cash assistance system in the U.S., on food insecurity—a fundamental form of hardship—of the next generation of young adults. An implicit goal underlying welfare reform was the disruption of an assumed intergenerational transmission of disadvantage; however, little is known about the effects of welfare reform on the well-being of the next generation. Using intergenerational data from the Panel Study of Income Dynamics and a variation on a difference in differences framework, this study exploits 3 key sources of variation in childhood exposure to welfare reform: (1) Risk of exposure across birth cohorts. (2) Variation of exposure within cohorts because different states implemented welfare reform in different years. (3) Variation between individuals with the same exposure who were more likely and less likely to rely on welfare. We found that longer exposure to welfare reform led to decreases in food insecurity of the next generation of households, by about 10% for a 5-year increase in exposure, with stronger effects for women, individuals exposed at least 13 years, individuals exposed at relatively young ages (0-5 years), and individuals whose mothers were not high school dropouts. We found no evidence that Supplemental Nutritional Assistance Program benefits explained any of the observed effects.

We examine intra- and intergenerational food security dynamics in the United States using longitudinal data from the Panel Study of Income Dynamics (PSID) while accounting for measurement error. To proceed, we apply recently developed methods on the partial identification of transition matrices. We show that accounting for measurement error is crucial as even modest errors can dwarf the information contained in the data. Nonetheless, we find that much can be learned under fairly weak assumptions; the strongest and most informative being that measurement errors are serially uncorrelated. While the evidence of both intragenerational and intergenerational is consistent with significant mobility, we also find food security status to be persistent for at least some households in the tails of the distribution. Finally, we document some heterogeneities in the dynamics across households differentiated by race and education.

Food insecurity, a condition of limited access to nutritious food, is a critical issue for college students’ health and well-being. In this report, we present results from our project that examines the long-term effects of college students’ food insecurity on future socioeconomic status, wealth, and food insecurity using nationally representative data from the Panel Study of Income Dynamics. Our study population included 1,574 individuals who were enrolled in college between 1999-2003, aged 16-29 years old during college enrollment, and remained in the sample through 2015-17. We examined the associations of interest using a combination of linear, logistic, multinomial logistic, and fixed-effects regression models, adjusting for sociodemographic characteristics measured concurrent with college enrollment or during childhood. In the sample, the prevalence of food insecurity during college was 14.9%. College food insecurity was associated with lower odds of college completion and lower likelihood of obtaining a Bachelor’s degree or a graduate or professional degree. These associations were more pronounced among first-generation college students. College food insecurity was also associated with lower mean labor wage income and lower wealth accumulation (including home equity) across subsequent survey waves between 2005-2019. Finally, college food insecurity was associated with a higher prevalence of food insecurity in the later adult years (measured in 2015-17). Results suggest that college food insecurity has long-term implications for socioeconomic outcomes in adulthood. Future research should focus on the role of programs and policies that can break the cycle of chronic food insecurity and poverty over the life course.