Sustenance and Stigma: Examining the Determinants of Nutritional Welfare Participation in Minnesota and Tennessee

Abstract

Ever since the first food stamp was purchased in 1939, the United States Department of Agriculture (USDA) has shown a commitment to improving nationwide food security through the expansion of public food assistance. In this thesis, we examine participation in two nutritional welfare programs; the Supplemental Nutrition Assistance Program (SNAP), formerly known as the Food Stamp Program, and the Free and Reduced Price Meals (FARM) program. We investigate participation dynamics surrounding two key uses of these programs; their use as a form of redistributive income assistance, and their use as a safety net in response to economic hardship. We are particularly interested in the presence of welfare stigma: in other words, the embarrassment attached to receiving public assistance. Our work builds on Robert Moffitt’s economic model of welfare stigma. We propose that stigma is not universal and is instead determined by place-based demographic and political factors. To evaluate determinants of welfare stigma, we analyze SNAP and FARM participation differences between multiple population subgroups. While we focus primarily on individuals living in rural versus urban areas, we also include subgroups defined by age, poverty levels, and partisanship leanings. We assume that differing participation rates unrelated to economic factors are due to stigma, allowing us to update dominant theories of welfare stigma. Our main research questions are; holding all else constant, are demographic and political factors correlated with 1) a region’s average nutritional welfare participation and 2) a region’s welfare response to an economic downturn? We hypothesize that after controlling for additional factors: rural regions will have lower participation averages and a larger welfare response to economic hardship; older regions will have lower participation averages and a smaller welfare response to economic hardship; poorer regions will have higher participation averages and a larger welfare response to economic hardship; and politically conservative regions will have lower participation averages and a smaller welfare response to economic hardship. To explore these questions, we use a mixed-methods approach. For our empirical analysis, we compile a longitudinal panel dataset that includes all counties in Tennessee and Minnesota for a period of over ten years. We run four fixed effects regressions – one for each program in each state – to observe subgroup-based participation differences. For the qualitative portion of our analysis, we interview five food-aid experts across three counties included in our study. We find that, contrary to prominent literature, population density does not have a strong relationship with welfare participation. Rather, age appears to be the most significant predictor of participation, with older counties showing lower participation averages across the board. In addition, partisanship has a strong relationship with welfare participation, but only in Tennessee. We also find that there is a significant delay between an economic shock and welfare participation increases. Finally, we observe that the factors related to participation differences are not uniform across states, nor welfare programs, emphasizing the need for context-specific studies. Based on our results, we recommend that welfare offices across the nation prepare for an influx of applications in the years that follow the COVID-19 economic shock; that SNAP asset requirements be made less strict; that the USDA supply annual demographic reports to local welfare offices and/or conduct state-specific research similar to this study; and that efforts be made to reduce welfare stigma in older generations.