Search

The Supplemental Nutrition Assistance Program (SNAP) is the largest anti-hunger program in the United States. SNAP benefits are rapidly depleted immediately after disbursement. This phenomenon, known as the “benefit cycle,” is associated with adverse dietary, nutrition, and health outcomes. Despite significant interest in policy interventions to address the benefit cycle, longitudinal research evaluating cyclical food purchasing and the underlying mechanisms driving such behavior is limited. The primary aim of this dissertation is to evaluate trends in food expenditures after households receive nutrition assistance, and examine the benefit cycle using behavioral and traditional economic frameworks. This dissertation uses data from a prospective trial of low-income households receiving nutrition assistance. The first paper describes patterns in food purchasing before and after households receive benefits. After benefit distribution, cyclical expenditures were observed for all food categories. Although cyclical spending behavior was observed broadly across different households, lower food security and income was associated with more severe expenditure cycles. The second paper evaluates the association between food expenditures and shopper impulsivity, which has been cited as a potential driver of the benefit cycle and is a target of SNAP educational interventions. Total food expenditures were cyclical for all households, but patterned by impulsivity. Shopper impulsivity was associated with expenditures on foods high in added sugar, but not fruits and vegetables. Findings suggest that shopper impulsivity likely exacerbates behavioral mechanisms underlying the benefit cycle, but is not the primary driver. The benefit cycle has also been attributed to time-inconsistent preferences, which suggests that households demonstrate a preference for immediate benefit consumption. The third paper evaluates how households discount utility across time, and whether observed time preferences explain cyclical food expenditures. Contrary to theoretical predictions, time-inconsistent preferences did not fully account for the benefit cycle. However, households with greater resources mitigated the benefit cycle and prioritized healthy foods when doing so. The final manuscript evaluates the impact of benefits over the course of the benefit cycle. Standard economic theory states that extramarginal households who receive more benefits than their typical food budgets respond sharply to increases in benefits. Given the relatively small portion of SNAP households who are extramarginal, the high marginal propensity to spend (MPS) out of benefits observed among SNAP households is incongruous with predictions. The fourth manuscript estimates the MPS out of benefits and cash across the benefit month by household extramarginality. Compared to other households, extramarginal households had a higher MPS out of benefits throughout the benefit month, but a lower MPS out of cash in the final two weeks. Extramarginal households also demonstrated a cyclical pattern in the MPS out of benefits. Results suggest that the benefit cycle can help explain the outsized MPS out of benefits reported in the literature, and a subset of extramarginal households may be driving more severe fluctuations in spending patterns. Understanding patterns in cyclical food purchasing and the mechanisms driving such behavior is critical for evaluating interventions to address the benefit cycle. Results from this dissertation suggest that interventions targeting households that are resource-constrained or extramarginal may help mitigate severe fluctuations in spending. Findings also suggest that policy proposals to help attenuate the benefit cycle—such as educational interventions for impulsive shoppers or more frequent benefit distribution to help households smooth consumption—should be directly evaluated to ensure their effectiveness.

This dissertation consists of three chapters, each of which provides causal evidence on the impacts of the Supplemental Nutrition Assistance Program (SNAP) in a distinct, policy-relevant area. The first chapter provides evidence of SNAP’s effects on the food retail industry. I combine data on SNAP participation, industry-specific retailer outcomes, and state SNAP expansions from 1998 to 2016. To address the endogeneity of SNAP participation, I employ a novel simulated eligibility instrumental variables framework exploiting variation in state SNAP eligibility expansions. I find that higher SNAP participation leads firms to operate more stores in industries where benefits are typically accepted, especially smaller general stores – a category dominated by dollar and discount stores. The second chapter provides evidence of SNAP’s effects on Medicaid enrollment and spending. I combine state-level information on SNAP eligibility expansions, SNAP participation, and Medicaid enrollment and spending from 1999 to 2012. I summarize diverse SNAP eligibility expansions through a novel simulated eligibility measure. I find that SNAP expansions increase Medicaid enrollment and decrease Medicaid spending per enrollee, suggesting that SNAP participation lowers barriers to enrollment in Medicaid for groups who cost lower on average to cover. The third chapter provides evidence of SNAP’s effects on mortality during its introduction in the “War on Poverty” era. SNAP was introduced as the Food Stamp Program (FSP) on a county-by-county basis from 1961 to 1975. I combine county-level information on the timing of FSP introduction with death counts from 1969 to 1978. I estimate the impacts of the FSP’s introduction on various mortality rates over time, including the overall mortality rate, population subgroup-specific rates, and cause-specific rates. I find that, among a subsample of high-poverty counties where the program’s introduction is likely to have a larger impact, the FSP reduced mortality over time. This effect was largely driven by reductions in the black, male, and age 0-19 mortality rates.

In the United States, grocery food tax policy varies at both state and county levels with 17 states having grocery taxes in the year 2020 and several states are engaged in active debates about whether to remove or impose such taxes. Although there is extensive economic literature evaluating multiple factors that may contribute to food insecurity, an examination of the linkage between grocery food sales taxes and food insecurity is lacking. In this thesis, county-level panel data on grocery taxes from 2006 through 2017 are developed and are merged with household-level data on food insecurity to examine whether grocery taxes are associated with food insecurity. The regressive nature of grocery taxes makes it plausible that taxing groceries could theoretically exacerbate food insecurity. Combining our county-level tax data with food insecurity measures and other data from the Current Population Survey, we estimate that a one-percentage-point increase in grocery tax rates is associated with a 0.84% increase in the probability of a household being food insecure. Using these estimates, we conduct policy simulations of grocery taxes that have been recently considered in six states and assess the potential impacts on food insecurity. Results show that these proposed grocery taxes may exacerbate food insecurity by 1 to 5 percentage points.

This dissertation explores two applications of behavioral economics. The first chapter investigates whether the composition of a choice set alters the relative importance of different attributes. I derive the distinguishing predictions of several prominent theories of choice-set-dependent preferences, including the focusing model of Koszegi and Szeidl (2013), the salience theory of Bordalo, Gennaioli, and Shleifer (2013), and the relative-thinking model of Bushong, Rabin, and Schwartzstein (2019). I test these predictions in a laboratory experiment in which I vary the prices of high- and low-quality variants of multiple products. The data provide clear evidence of choice-set dependence: price increases that expand the range of prices in the choice set lead to more purchases, and increasing prices while holding price differences fixed induces substitution to higher-quality options. Structural estimates imply that most participants exhibit choice-set dependence consistent with relative thinking, albeit with substantial heterogeneity in the intensity of the effect. The consequences of this choice-set dependence are economically meaningful: my estimates imply that the average participant was willing to pay around a 20% premium relative to their intrinsic value due to the presence of seemingly irrelevant options in the choice set. The second chapter investigates how seemingly trivial components of the recertification process of the Supplemental Nutrition Assistance Program can impact program participation. We document low recertification rates, which we attribute partly to procedural issues associated with the recertification process. Current recipients, who must complete a caseworker interview by recertification month end, are 20 percent less likely to recertify when assigned an interview at the end rather than the beginning of the month. We provide evidence that a substantial fraction of recertification failures appear due to mistakes: the majority of these marginal cases successfully reapply to the program, with larger effects for higher need cases. We find that the data are most consistent with a model of inattention in which individuals are inattentive to, or unaware of, critical components of the rectification process. We estimate costs to participants and administrators created by rigid renewal deadlines.

The participation rate of non-citizen households eligible for the Supplemental Nutrition Assistance Program (SNAP) is much lower than the average participation rate of all eligible households. Using the Survey of Income and Program Participation (SIPP), this thesis explores the effect of household members' language use and English proficiency on their SNAP participation decision by estimating panel data econometric models. The main finding is that households whose members speak English at home are 5.1% more likely to participate in SNAP. The result implies that non-native households may have difficulties in applying for SNAP because of higher transaction costs of application. Therefore, policies targeted to reduce transaction costs facing non-native households would likely help increase their SNAP participation, which may lead to less food insecurity. However, the results also show that the effect of household members' English proficiency on their SNAP participation is not statistically significant. Since this result could be due to the limitations of the data, it would be worthwhile to conduct future research using more reliable measurement of English proficiency such as a well-designed test score.

The purpose of this research was to analyze the monthly dietary and food expenditure patterns of women participating in the Supplemental Nutrition Assistance Program (SNAP). The secondary goal was to identify factors that influence weight status of women in SNAP. In Study 1, a 95-item, semi-quantitative food frequency questionnaire based on a 1-week reference period was developed. It was validated against 3-day diet records in a sample of 70 female SNAP participants. The food frequency questionnaire was administered again after one month in a subgroup of 40 women to evaluate its test-retest reliability. Mean daily nutrient intakes obtained from the two tests of the 95-item food frequency questionnaire, and 3-day diet records were calculated. The average correlation and weighted Cohen’s kappa for the validity study were 0.60 and 0.39, respectively. The mean correlation and kappa statistics for the reliability study were 0.66 and 0.50, respectively. Thus, the 95-item, 1-week food frequency questionnaire effectively measured diets of women participating in SNAP. In study 2, a sample of 151 SNAP women were administered the 1-week food frequency questionnaire four times, each with an interval of one week, so that it reflected weeks 1, 2, 3 and 4 of the diets of participants. A significant decrease in intakes of fruits, vegetables, whole grains, dairy and total protein foods were observed with an increase in time since receipt of benefits. Moreover, a decline in diet quality towards the end of the SNAP month was determined. In Study 3, 160 women participating in SNAP saved their grocery receipts for one month, and were measured for height and weight. Spending patterns on 29 food categories were determined from the receipts, and compared to the Thrifty Food Plan (TFP) recommendations. Some of the food categories for which the expenditure was significantly greater than the recommendations included: refined grains, red meat, frozen entrees, soft drinks, and sodas, fruit drinks and ades. A greater amount spent on monthly groceries relative to the TFP recommended total cost was significantly associated with higher compliance of spending on low fat dairy, vegetables, whole grains and fruits to the recommendations. In Study 4, 152 women in SNAP completed the multi-dimensional home environmental scale and adult food security module and were measured for height and weight. Multi-dimensional home environmental subscales that were significantly associated with body mass index, and also differed according to food security status were: availability of unhealthy foods at home and neighborhood safety. Factors measured by this scale significantly mediated the relationship between food insecurity and body mass index

The purpose of this research was to analyze the monthly dietary and food expenditure patterns of women participating in the Supplemental Nutrition Assistance Program (SNAP). The secondary goal was to identify factors that influence weight status of women in SNAP. In Study 1, a 95-item, semi-quantitative food frequency questionnaire based on a 1-week reference period was developed. It was validated against 3-day diet records in a sample of 70 female SNAP participants. The food frequency questionnaire was administered again after one month in a subgroup of 40 women to evaluate its test-retest reliability. Mean daily nutrient intakes obtained from the two tests of the 95-item food frequency questionnaire, and 3-day diet records were calculated. The average correlation and weighted Cohen’s kappa for the validity study were 0.60 and 0.39, respectively. The mean correlation and kappa statistics for the reliability study were 0.66 and 0.50, respectively. Thus, the 95-item, 1-week food frequency questionnaire effectively measured diets of women participating in SNAP. In study 2, a sample of 151 SNAP women were administered the 1-week food frequency questionnaire four times, each with an interval of one week, so that it reflected weeks 1, 2, 3 and 4 of the diets of participants. A significant decrease in intakes of fruits, vegetables, whole grains, dairy and total protein foods were observed with an increase in time since receipt of benefits. Moreover, a decline in diet quality towards the end of the SNAP month was determined. In Study 3, 160 women participating in SNAP saved their grocery receipts for one month, and were measured for height and weight. Spending patterns on 29 food categories were determined from the receipts, and compared to the Thrifty Food Plan (TFP) recommendations. Some of the food categories for which the expenditure was significantly greater than the recommendations included: refined grains, red meat, frozen entrees, soft drinks, and sodas, fruit drinks and ades. A greater amount spent on monthly groceries relative to the TFP recommended total cost was significantly associated with higher compliance of spending on low fat dairy, vegetables, whole grains and fruits to the recommendations. In Study 4, 152 women in SNAP completed the multi-dimensional home environmental scale and adult food security module and were measured for height and weight. Multi-dimensional home environmental subscales that were significantly associated with body mass index, and also differed according to food security status were: availability of unhealthy foods at home and neighborhood safety. Factors measured by this scale significantly mediated the relationship between food insecurity and body mass index

It is unknown if participation in the Supplemental Nutrition Assistance Program (SNAP) influences the magnitude of improvement in dietary intake in response to dietary interventions. Adults with low socioeconomic status (SES) tend to have lower overall dietary quality as compared to those with higher SES. However, low SES adults are more likely to receive benefits from SNAP, which gives nutrition assistance to millions of eligible Americans. The objective of this investigation is to examine differences in dietary intake between 1) SNAP participants, 2) those eligible for SNAP but not receiving (nonparticipants), and 3) those ineligible for SNAP, in response to an intervention targeting a reduction in sugar-sweetened beverage (SSB) consumption. Adult participants (n=146) from Southwest Virginia were enrolled in a 6-month, community-based trial, SIPsmartER. Participants provided SNAP enrollment status and 3 24-hour dietary recalls at baseline and 6-months. Dietary variables (SSB, macronutrients, etc.) and dietary quality data (Healthy Eating Index [HEI-2010]) were derived from nutritional analysis software (NDS-R 2011). Statistical analyses included descriptives and repeated-measures ANOVA. Although SNAP participation and eligibility status did not impact the overall effectiveness of this dietary intervention, the within group data suggests that those eligible for SNAP but not participating (n=30) may be at a disadvantage to improving their dietary intake as compared to those at a similar household income who receive SNAP benefits (n=56) or ineligible individuals at a higher income level (n=60). Future research is needed to explore if participant's ability to maintain long-term adherence to the dietary changes differs between groups.

Food Security is a situation that exists when all people, at all times, have physical, social, and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life (FAO, 1996). 17% of Kentuckians are food insecure (Kentucky Department of Agriculture, 2016). This study explored the quality of life (QoL) impact of the Bluegrass Double Dollars (BGDD) program on Supplemental Nutrition Assistance Program (SNAP) participants through secondary data analysis. Utilizing the categories of quality of life indicators established by The Commission on the Measurement of Economic Performance and Social Progress (CMEPSP) the results from this study concluded that participating in the BGDD program provides some level of quality of life benefits.

This research examines how aspects of a weakened social structure potentially impact SNAP participation over the course of the Great Recession among elderly low-income heads of households. The objective is to understand how individual characteristics, as well as structural factors of a weakened economy, a growing low-wage labor market, and a shrinking safety net, impact participation in SNAP.This research will specifically ask the following questions:How does participation in SNAP change throughout the course of the Great Recession when the structural vulnerability of poverty framework is examined?Do differences in SNAP participation change by structural factors associated with poverty across states?This research brings a new approach to investigating the relationship between determinants of SNAP participation as a function of poverty in relation to structural factors throughout the Great Recession. Instead of asking how SNAP participation affects labor force participation, this research adds to the literature by exploring what factors act as a determinant to SNAP participation in times of economic uncertainty.A binary logistic multilevel regression was conducted using secondary data from the Survey of Income and Program Participation 2008 Panel. The sample included low-income (150% of the Federal Poverty Line) heads of households, aged 18-64.

This dissertation takes a distributional approach to examining dietary quality in the United States. Diet quality is a direct input to health, is often used as a proxy for well-being, and is an outcome variable for a wide variety of economic interventions. This makes diet quality a particularly important, yet understudied, outcome for program evaluation and describing food bundles that individuals choose. The first chapter describes the evolution of adult dietary quality in the U.S. over the last two decades. Contrary to popular wisdom, there have been statistically significant improvements at all levels of diet quality. For the population as a whole, we find significant improvements across all levels of diet quality. Further, we find improvements for both low-income and higher-income individuals alike. Counterfactual distributions of dietary quality are constructed to investigate the extent to which observed improvements can be attributed to changes in the nutritional content of foods and to changes in population characteristics. We find that 63% of the improvement for all adults can be attributed to changes in food formulation and demographics. Changes in food formulation account for a substantially larger percentage of the dietary improvement within the lower-income population (19.6%) as compared to their higher-income counterpart (6.4%). The sheer myriad of overlapping policies and public awareness initiatives during this time period make it difficult to pin down the exact causes behind such improvements. This chapter motivates two program evaluation studies in the two chapters that follow.The second chapter estimates distributional effects of food consumed at school and away from home on child dietary quality. Using a fixed-effects quantile estimator, two non-consecutive days of food intake are used to identify the effect of eating away from home and at school. I find considerable heterogeneity in the estimated impacts. The study finds that food away from home, as compared to home-prepared food, has a negative impact on the distribution of dietary quality except at low quantiles. Main results suggest that school food has both positive and negative impacts across the distribution of dietary quality. I find positive impacts on dietary quality at low quantiles of the outcome distribution, whereas food from school has a negative impact at the upper end of the distribution of diet quality. While food consumed under the National School Lunch and Breakfast Programs may not benefit every child, especially the average child, it does improve the diets of many children who otherwise would have poorer dietary quality. The implication is that U.S. schools are fertile grounds to improve nutrition skill formation, especially for the most nutritionally disadvantaged. This final chapter estimates the effect of replacing food assistance benefits, which typically come in the form of a food voucher, for an equal value of cash on the quantity and quality of food consumed in a household. We utilize an experiment in which a portion of beneficiaries were chosen at random to receive their benefits in the form of cash. We take a distributional approach because we believe it is important to analyze low-consuming households separate from high-consuming households. We find some evidence that a cash system would increase kilocalorie consumption in the portion of the distribution below recommended levels of consumption and decrease consumption in the portion of the distribution well above any reasonable threshold. This finding implies that a cash transfer system may both alleviate food insecurity and decrease overconsumption. The cash system appears to have a positive impact on the distribution of dietary quality in quantiles above 40. Virtually all of the improvement in quality comes from a decrease in consumption of less-healthy foods by the cash receiving group. Overall, these findings imply that beneficiaries are no worse off under a cash transfer system and in fact, may be better off.

The first study investigates the factors that impact the duration of U.S. fresh fruit and vegetable imports. We employ both survival analysis (Kaplan Meier estimates and Cox proportional hazards model) as well as count data models. Our results indicate that SPS treatment requirements positively impact the duration of trade while new market access has the opposite effect. Other factors typically included in trade duration models (such as: GDP, transportation costs, tariff rates, etc.) were also investigated. We also employ a probit model to understand the factors impacting the probability that a country selects into exporting fresh fruits and vegetables to the United States. The second study estimates the goods-time elasticity of substitution for Food Stamp/SNAP participants versus non participants. We find that the elasticity of substitution for SNAP participants is not statistically different from zero. This indicates that SNAP participants have Leontief production function in household food production, implying that increasing the amount of SNAP benefits paid to participants will not lead to more food production if the time households dedicate to food preparation remains unchanged. This finding extends the analysis done by Baral, Davis and You (2011) and offers insights for policies related to the SNAP program.