UC Davis

The social insurance system in the United States is a broad and varied set of programs, transfers, and tax credits that serve as a social safety net, often through the US tax system. The US spent around 13% of its GDP, or around $2.7 trillion, on social insurance programs in 2019. This dissertation explores how the social insurance system in the US functions as anti-poverty policy. The first two chapters focus on the Earned Income Tax Credit (EITC) and examine the mechanisms through which the program works. These two chapters use different approaches and data to understand household expenditure decisions because of EITC receipt. I focus on food expenditures, as food insecurity and lack of nutrition remain challenges for low-income households in the US. The third chapter explores the role of food and nutrition programs in the US social insurance landscape, pointing out common threads in the literature, established patterns in economic research, as well as the remaining gaps for further analysis.

Chapters 1 and 2 explore the impact of the Earned Income Tax Credit on different measures of household expenditures. Prior literature has established clear short- and long-run benefits of the Earned Income Tax Credit (EITC), but the mechanism behind these effects is unclear. These papers provide evidence that these benefits occur by increasing nondurable expenditures for recipients. Both papers consider the same demographic group---low-income, single women---which has technical and empirical advantages for the context of my analysis. Focusing on single women is a critical contribution of these papers, as they tend to be economically vulnerable and have different spending patterns than married families or single males.

In Chapter 1, I examine the impact of the EITC on the food expenditures of single female headed households. Using longitudinal data with a dynamic difference-in-differences design and a pooled estimation model, I comprehensively study all EITC policy changes over time. This includes the 1975 introduction, a currently understudied aspect of the program. This paper additionally estimates the marginal propensity to consume out of transfer income, providingnew evidence on the elasticity of household expenditures to a large lump-sum transfer. I find that an increase of $1 of EITC benefits leads to $0.39 more of food expenditures, significantly higher than the proportion of disposable income typically spent on food. This research strengthens the literature on how public assistance changes the spending decisions of low-income households and furthers the research on a population not typically the focus of the household finance literature.

In Chapter 2, I examine how the Earned Income Tax Credit affects all nondurable expenditures, moving beyond only food expenditures. I use data with a monthly frequency, to better see expenditure patterns in the year following EITC receipt. Using the major expansion of the EITC in 2009, which increased benefits by $1025 (worth about $1250 in 2020) for one particular demographic group. This provides an ideal quasi-experimental setting to explore the effect of increasing the benefit amount for one group, with another unchanged, and very similar, comparison group. Using both a static and dynamic difference-in-differences design, I find an $86 increase in monthly nondurable expenditures, and then further decompose that category into key areas of spending. This includes grocery spending, where there is an increase of about $36 as well as total food spending, with an increase of about $65. These numbers are in line with the findings in Chapter 1. Studying how EITC receipt influences household consumption decisions can strengthen our understanding of the impacts of means tested government transfer programs and provide new evidence onconsumption elasticities for low-income populations as an outcome of social insurance policies.

Finally, Chapter 3 summarizes the literature on US public food and nutrition assistance programs. This review focuses on the health and nutrition impacts of food assistance programs. We focus particular attention on the United States, both because of the plethora of types of programs and associated variation and because spending on these programs is a large share of the nonmedical safety net there. We begin by reviewing the theoretical predictions concerning health and nutrition effects of these programs, also paying attention to potential mediators such as education and income. We then discuss program eligibility and size, both as caseload and in terms of spending. We next touch on identifying causal variation and opportunities for further research. The review concludes by discussing the existing literature in five broad areas: take-upand use of the programs; effects on nutrition and food consumption; other immediate effects on short-run health; impacts on other contemporaneous outcomes such as income and labor supply; and longer-run health and nutrition effects.