The Economic Research Service funded 12 research grants in fiscal 2014 to utilize the new FoodAPS National Household Food Acquisition and Purchase Survey. These research projects will provide rigorous research that expands ERS's understanding of household food behaviors and the Supplemental Nutrition Assistance Program (SNAP), as well as the role of the local food environment and other geographic factors on household food purchase and acquisition decisions. The awards were made by the University of Kentucky Poverty Research Center in cooperation with the University of Illinois.
(Click on the title for project descriptions)
Household Food Behaviors and SNAP
|Dr. Hilary Seligman
|The Relationship of Food Price Variations to Healthy Food Acquisition
Dr. Joshua Berning
University of Georgia
|The Effects of Benefit Timing and Income Fungibility on Food Purchasing Decisions among SNAP Households
Dr. Erin Bronchetti
|Variation in Food Prices and SNAP Adequacy for Purchasing the Thrifty Food Plan
|Dr. Yunhee Chang
University of Mississippi
|The Effect of Food Price on Food Insecurity and Diet Quality: Exploring Potential Moderating Roles of SNAP and Consumer Competency
|Dr. Amy Hillier
University of Pennsylvania
|Influence of SNAP Participation and Food Environment on Nutritional Quality of Food at Home Purchases
|Dr. Conrad Lyford
Texas Tech University
|Do SNAP Recipients Get the Best Prices?
Understanding the Role of Geographic Factors on Household Food Purchases and Acquisition Decisions
|Dr. Scott Allard
University of Chicago
|The Spatial Context of Food Shopping: Understanding How Local Food Retailer Access and Pricing Affect Household Behavior
|Dr. Sarah Bowen
North Carolina State University
|Contextualizing Family Food Decisions: The Role of Household Characteristics, Neighborhood Deprivation, and Local Food Environments
|Dr. Alison Gustafson
University of Kentucky
|Neighborhood Disadvantage, Food Resources in the Environment, and the Influence on Food Store Choice and Purchasing Behavior among Supplemental Nutrition Assistance Program (SNAP) Participants and SNAP Eligible Households
Dr. Barbara Laraia
University of California, Berkeley
|A Spatial Interaction Model of Food Access and Its Relationship to Food Insecurity
|Dr. Michael Kuhn
University of Oregon
|Household Composition and the Calorie Crunch: The Importance of Non-Unitary Models of the Family for Budgeting Over Time
|Dr. Sofia Berto Villas-Boas
University of California, Berkeley
|If You Build It Will They Come? Store Choice Determinants among SNAP and Low-Income Households
The Relationship of Food Price Variations to Healthy Food Acquisition
- Hilary Seligman, University of California, San Francisco (PI)
- Sanjay Basu, Stanford University (Co-Investigator)
- Christopher Wimer, Columbia University (Co-Investigator)
This project will examine how the healthfulness of foods acquired by Supplemental Nutrition Assistance Program (SNAP) households varies between higher- and lower-cost areas of the country. It will also examine how the healthfulness of acquired food varies by: 1) price differences among healthier and less healthy foods within stores frequented by SNAP users, and 2) overall cost of living.
Using the National Household Food Acquisition and Purchase Survey (FoodAPS), the investigators will examine the relationships between food prices and healthy food acquisitions within and among high- and low-food-price areas of the country, defined by Regional Price Parities. Additionally, they will study how healthy food acquisition relates to price variations between healthier and less healthy foods within metropolitan statistical areas, retail market areas, and individual stores frequented by SNAP users, as well as variations in cost-of-living indices and geographic adjustments to poverty thresholds. The healthfulness of foods will be defined by the Healthy Eating Index, and all analyses will compare how food price and food acquisition behaviors differ within SNAP, non-SNAP and low-income, and higher income populations, using a control function approach to account for the endogeneity of SNAP participation.
The analysis will address the question—how much do SNAP adjustments for food price environments encourage healthier food acquisitions? The analysis will also allow for a better understanding of whether interventions should be targeted toward specific parts of the country (high food-cost or overall-cost regions), or toward those stores where SNAP recipients are most likely to shop.
The Effects of Benefit Timing and Income Fungibility on Food Purchasing Decisions among SNAP Households
- Joshua Berning, University of Georgia (PI)
- Gregory Colson, University of Georgia (Co-PI)
- Jeffrey H. Dorfman, University of Georgia (Senior Personnel)
- Travis A. Smith, University of Minnesota (Co-Investigator)
Recent empirical findings have identified an unexpected consequence of the monthly SNAP provision—participants have higher consumption shortly after receiving their benefits, followed by lower consumption toward the end of the month. This "SNAP benefit cycle" has been associated with adverse dietary outcomes among participants. The investigators hypothesize that two behavioral responses by SNAP participants work in tandem to drive much of the cycle: 1) short-run impatience—a higher preference to consume today, and 2) fungibility of income–the degree of substitutability between a SNAP dollar and a cash dollar. The objective of the research is to disentangle these two effects and any compounding effects on the quality and quantity of SNAP households' food purchases.
This study allows for time-inconsistent preferences (such as impatience) by estimating household purchase decisions throughout the benefit month. As benefits become exhausted, it is anticipated that households will substitute higher-quality, higher-cost food for lower-quality, lower-cost food. This study further modifies the model to allow for imperfect substitution between cash and SNAP income. This allows income fungibility to also be a function of the benefit cycle. These modifications allow us to disentangle the behavioral effects on the quantity and quality of food purchases over the benefit month.
This research will deliver several key insights with immediate policy implications. First, the model will estimate the contribution of impatience, income fungibility, and any compounding effects to the benefit cycle. Importantly, the model suggests that the compounding effect can be lessened by a relatively low-cost policy reform—more frequent benefit distribution. Even if benefit frequency remains monthly, the model is able to inform alternative outreach efforts such as SNAP-Ed.
Variation in Food Prices and SNAP Adequacy for Purchasing the Thrifty Food Plan
- Erin Bronchetti, Swarthmore College (PI)
- Garret Christensen, Swarthmore College (Co-PI)
- Benjamin Hansen, University of Oregon (Co-PI)
An important and timely policy question is whether SNAP benefits are adequate to provide food security for eligible households. While the nominal value of SNAP benefits is fixed across States (except for Hawaii and Alaska), variation in food prices across geographic areas is dramatic, and the real value of SNAP benefits varies widely across the U.S. This proposed research provides new evidence on geographic variation in the adequacy of the SNAP benefits, estimating the distribution of the cost of the Thrifty Food Plan (TFP) for households across the Nation and calculating the percentage of SNAP recipients and SNAP-eligible households for whom the SNAP benefit level is adequate to purchase the TFP.
This study will use the cost of the TFP calculated for each store in the FoodAPS-Geographic Component (GC) data and the prices of specific food categories to calculate the cost of the TFP for each respondent in the study using several methods. Stores will include (but are not limited to): 1) the closest store to his/her home, (2) the store where he/she purchased the most food, and (3) the lowest-‐priced store where he/she purchased any food. TFP costs will be computed in multiple ways—one way assumes that all food is purchased at one store, and an alternative way breaks the TFP into food categories and assumes that food is purchased at the lowest- cost store patronized by the customer for each category. For each respondent in the sample, the investigators will calculate the potential SNAP benefit to which the household is entitled, using information on household income, expenses, family size and composition. Finally, the main outcome of the study—estimates of the fraction of SNAP-recipient and SNAP-eligible respondents for whom SNAP benefits, plus 30 percent of income, are adequate to purchase the TFP.
The potential implication is that if there is strong evidence that benefit levels are inadequate in certain areas or overly generous in others, policymakers could tie benefit levels to food prices and provide adequate nutrition assistance to a greater fraction of the population. In addition, by providing a better understanding of variation in the relative generosity of SNAP benefits for potential recipients, these findings may inform future research regarding the effect of SNAP benefits on health outcomes.
The Effect of Food Price on Food Insecurity and Diet Quality: Exploring Potential Moderating Roles of SNAP and Consumer Competency
- Yunhee Chang, University of Mississippi (PI)
- Jinhee Kim, University of Maryland (Co-PI)
- Swarn Chatterjee, University of Georgia (Co-PI)
Higher local food prices not only aggravate household food insecurity but may also hurt diet quality and health outcomes. This study takes advantage of detailed food acquisition and purchase records and geographic indicators in FoodAPS to explore how local food prices affect food insecurity as well as diet quality, and if the effects are adverse, whether consumer competency can mitigate such effects. This study also examines whether lack of consumer competency or effectiveness among some SNAP participants partly explains the program’s inconsistent success in improving nutrition and health, and whether SNAP effects can be improved in high food cost areas by strengthening consumer competency.
Consumer competency scales will be constructed based on the information from food books and interview files to represent a consumer’s effectiveness in food acquisition and purchase, money management, and nutrition information. Food acquisition and purchase records from household food books evaluated against dietary guidelines will be used to construct diet quality measures. Local costs of the Thrifty Food Plan and prices of aggregate food groups as well as other information about local food environments will be obtained from FoodAPS-GC. Regression models will include a series of interaction terms to test moderating effects of consumer competency among program participants in high-price areas. Program participation will be instrumented by State policy variables to avoid endogeneity bias. Fixed effect terms will be included to account for unobserved variations in consumption environment and lifestyle.
This study not only speaks to the adequacy of SNAP benefits and the possibility of indexing benefits based on cost of living, but it also has direct implications for current policies focused on better understanding of program participants’ food-buying behavior and knowledge. Findings will shed light on whether the current program lacks a system that equips or incentivizes participants to become competent consumers who make efficient and healthful use of program benefits when faced with adverse market conditions.
Influence of SNAP Participation and Food Environment on Nutritional Quality of Food at Home Purchases
- Amy Hillier, University of Pennsylvania (PI)
- Benjamin Chrisinger, University of Pennsylvania (Co-Investigator)
- Tony E. Smith, University of Pennsylvania (Co-Investigator)
A growing body of research describes how individuals make food shopping decisions in both time and space, adding needed complexity to our understanding of "food deserts". The FoodAPS Initiative will allow this study to test whether the findings from a previous study conducted in Philadelphia can be generalized and offer insights on the interactions between food environments, food choices, and food assistance.
Three research questions are proposed: 1) where do participants shop for food at home (FAH) and how do individual/household characteristics interact with store characteristics and distance? 2) how does store choice influence the nutritional quality of FAH purchases, controlling for individual and household characteristics? and 3) how does the local food environment influence the nutritional quality of FAH purchases? For all three questions, the study will investigate how SNAP participation interacts with the outcomes of interest. This study will employ spatial and econometric models to estimate the effects of different individual and environmental variables on the nutritional quality of FAH purchases. Using geographic information systems (GIS), it will generate additional spatial variables, including "food shopping sheds," to offer more nuanced food environment indicators.
Findings will offer policymakers additional understanding of how local food environments affect the nutritional quality of food purchases and point to areas for possible health intervention. At present, there is little agreement on which line of policy is most influential for diet (place-based investment in local food environments versus people-based investment in food assistance). This investigation will better describe the protective effects of local food environments and food assistance programs.
Do SNAP Recipients Get the Best Prices?
- Conrad Lyford, Texas Tech University (PI)
- Carlos Carpio, Texas Tech University (Co-PI)
- Tullaya Boonsaeng, Texas Tech University (Co-PI)
- Janani Thapa, Texas Tech University (Co-PI)
One of the key challenges when purchasing food is to consider the relative prices in a particular food environment. The main focus of this research will be on analyzing and quantifying the factors that affect food prices paid by households, in particular SNAP purchases.
The first step of the analysis will involve the calculation of a price index—also called an expensiveness index. This index compares the cost of a household’s food basket at average prices to the cost actually paid by the household. The second step of the analysis will involve regressing the expensiveness index on a set of explanatory variables including income, demographic characteristics (including SNAP participation), and factors characterizing the competitiveness and structure of the retail market. The endogeneity of SNAP participation will be accounted for by using instrumental variables methods with instruments selected from the SNAP policy database also available in the Food APS-Geographic Component (GC).
The policy implications of this study will inform SNAP-Ed efforts by identifying effective approaches by regional differences (for example, supercenter availability) and differences across household types (such as shopping frequency). In addition, the role and value of additional competition at the local food provider level would be examined to determine the relative effect of retail market competitive structure on food purchases and prices. These factors play a role in the effectiveness of the SNAP program and ensuring that households have access to nutritious and balanced meals.
The Spatial Context of Food Shopping: Understanding How Local Food Retailer Access and Pricing Affect Household Behavior
- Scott Allard, University of Chicago (PI)
- Patricia Ruggles, NORC at the University of Chicago (Co-PI)
This project explores the relationship between the food retail environment surrounding poor and near-poor households, household food shopping behavior, and household food security. Drawing on both restricted-use data and prepared measures available through FoodAPS, it will examine the spatial and non-spatial characteristics of the local food retail environment. Particular attention will be paid to how food pricing and product availability vary across types of food retailers within the local food retail environment. Empirical analyses will model how spatial access to food retailers and spatial variation in food pricing is associated with food purchases and household decisions about shopping venue among households within 300 percent of the Federal poverty line.
This study will create retail food access and food pricing measures to reflect the accessibility and affordability of retail food in the surrounding community. Access measures will take into account different commuting times and distance bands. Descriptive analyses will compare food resource access and food pricing across different population subgroups (such as race and poverty status) and geographic locations (for example, urban, suburban, or rural tract, or high or low poverty tract). This study will also estimate a series of multivariate models that explore factors associated with different household food outcomes such as type of retail food store most frequented; total weekly food expenditures; percent of weekly food purchases at supermarkets; percent of weekly food expenditures on fruits and vegetables; and household food security.
Future development of interventions to enhance food security will benefit from clearer evidence on the roles of local retail food availability and pricing in determining food purchases and food security. Improved understanding of how spatial context shapes food insecurity could translate into more efficient and effective allocation of public program dollars, private capital, and philanthropic resources.
Contextualizing Family Food Decisions: The Role of Household Characteristics, Neighborhood Deprivation, and Local Food Environments
- Sarah Bowen, North Carolina State University (PI)
- Richelle Winkler, Michigan Technical University (Co-PI)
- J. Dara Bloom, North Carolina State University (Co-Investigator)
- Lillian O’Connell, North Carolina State University (Co-Investigator)
A growing body of literature focuses on disparities in access to healthy foods and on the relationships between local food environments and outcomes related to diet and health. This work has had direct policy implications, as evidenced by healthier food retail legislation at the State and Federal levels. At the same time, recent research also suggests that the food environment/diet relationship is far from straightforward, and that household finances, not proximity to stores, may be the more important factor. These studies suggest that the local food environment interacts in critical ways with issues related to poverty and household resources.
This analysis employs multilevel models integrated with geographically-weighted regressions to analyze the spatially varied associations between household characteristics, neighborhood characteristics, and dietary quality. Using data from FoodAPS, the key outcome variable is household food purchases in specific categories, used as a proxy for dietary quality. Key explanatory variables at the household level include variables for total household income; housing burden; housing instability; car ownership; and SNAP participation and benefits. The model includes dimensions of food availability and affordability in the local food environment, as well as separate variables to represent neighborhood deprivation.
It is expected that the relationship between household characteristics and dietary quality will be area-specific. If the expected outcomes are verified, this research would challenge public health experts and practitioners to think more comprehensively about how consumers make food decisions. While policies to increase access to retail food stores may be helpful, policies to increase the resources that households have available for food purchases may be more effective.
Neighborhood Disadvantage, Food Resources in the Environment, and the Influence on Food Store Choice and Purchasing Behavior among Supplemental Nutrition Assistance Program (SNAP) Participants and SNAP Eligible Households
- Alison Gustafson, University of Kentucky (PI)
- James Allen, University of Kentucky (Co-Investigator)
- Mark Swanson, University of Kentucky (Co-Investigator)
- Nancy Schoenberg, University of Kentucky (Co-Investigator)
Understanding how low‐income adults decide which food retailers to utilize and what types of food to purchase is fundamental to improving U.S. nutrition programs and advancing economic development in lower‐income neighborhoods. Research to date has reported that neighborhoods with a higher proportion of minority or low‐income residents have fewer healthy food retailers and less healthy foods available within stores. However, even in areas with highly limited access to foods, different households make different choices about what types of stores to utilize. Yet, there is a significant gap in the literature that examines how access to various food venues is associated with utilization of stores and purchasing behavior.
This research seeks to answer the question: What are the interacting effects of environmental-level variables (number and type of food venues within a neighborhood and degree of neighborhood disadvantage) and individual-level variables (store choice, purchasing habits) on select dietary outcomes of both SNAP participants and SNAP eligible (but nonparticipating) households. The first aim of this project is to study the determinants of store use, hypothesized as being associated with neighborhood disadvantage (including SNAP participation and eligibility), while also controlling for food resources, number and types of food venues within neighborhoods, and other factors. Fractional multinomial econometric models will be used to achieve this aim. The second aim, using multilevel analysis, is to study the determinants of purchasing habits for key foods, hypothesized as being associated with store use—which itself is a function of neighborhood disadvantage and the food resource environment—as well as to study household characteristics among SNAP and SNAP-eligible households.
The results of this study will allow key stakeholders to assess how best to improve food resources within neighborhoods where low-income residents live.
A Spatial Interaction Model of Food Access and Its Relationship to Food Insecurity
- Barbara Laraia, University of California, Berkeley (PI)
- Hilary Hoynes, University of California, Berkeley (Co-PI)
- Janelle Downing, University of California, Berkeley (Graduate Student Researcher)
The direct and indirect effects of place on health have been a source of great debate and interest in the last decade. A number of studies have examined the relationship between access to healthy food and food insecurity, poor diet, and obesity. Policymakers have been concerned that healthy foods might be less accessible to the poor if supermarkets are located far from their homes, and transportation and time costs are prohibitive. The purpose of this project is to develop measures of food access and to use them to assess food insecurity of low-income Americans.
This study draws on Reilly’s law of retail gravitation and spatial interaction theory to describe a shopper’s interaction with his or her food environment as a function of the location, product prices, and store attributes of the closest store and competing stores. The relationship between food access and food insecurity will be estimated at the household level using a multinomial logit model. In the course of the study, the investigators will create a conceptually-friendly measure of access that can be added to the ERS Food Access Research Atlas to be used by policymakers and community advocates.
This project will enhance policymakers' understanding of how economic and spatial characteristics of the environment interact to influence food insecurity at the household level. It will provide insight on how geospatially-derived, theoretical measures of access might be used to contextualize food insecurity risk across a nationally representative population of low-income Americans.
Household Composition and the Calorie Crunch: The Importance of Non-Unitary Models of the Family for Budgeting Over Time
- Michael Kuhn, University of Oregon (PI)
Many researchers have identified that expenditures on and consumption of food throughout the monthly SNAP (food-stamp) cycle fluctuate considerably. Such fluctuations may be related to declining food intake and reduce food security over the course of the SNAP month.
FoodAPS data offer a unique opportunity to simultaneously measure both the expenditure and consumption profiles of a household, map the correspondence between the two, and ask how that mapping is affected by household structure, food availability, and food prices.
- Is declining food intake especially severe in areas with low food availability? Making one very large shopping trip at the beginning of the month may save on convenience costs but result in a situation where budgeting food-on-hand to last an entire month is difficult.
- Is declining intake driven by price fluctuations? Is declining intake due to the shopping or eating patterns of certain types of individuals within a household?
- Could parents be over-appeasing the demands of children early in the month or racing to consume the available resources before the end of the month?
Answering these questions speaks directly to how SNAP policy can be adjusted to mitigate end-of-month hunger.
If You Build It Will They Come? Store Choice Determinants among SNAP and Low-Income Households
- Sofia Berto Villas-Boas, University of California, Berkeley (PI)
- Rebecca Taylor, University of California, Berkeley (Co-Investigator)
Despite the growing body of research on food deserts and food insecurity, there is still little evidence supporting the assumption that improved access will alleviate dietary problems. In light of this uncertainty, it is important to understand current determinants of store choice among low-income households—as well as the types and chains of retail food stores that are influencing customers to make healthier food purchases—before implementing policies that incentivize retailers to do business in food deserts.
With this concern in mind, this research asks: 1) What are the determinants of store choice among SNAP participants and, in particular, low-income households? 2) What influences store choice both among and within different types of store categories? 3) What are the effects of both prices and in-store services on store choice? and 4) How does store choice influence the purchasing decisions for all types of fruits and vegetables (such as frozen, canned, juiced, and dried) and not just fresh produce? To tackle these four questions, the research design employs a mixed logit demand model common in the discrete choice literature and relies heavily on the FoodAPS datasets, as well as data on in-store services.
The results of this research will have policy implications regarding the improvement of food access for low-income households. For instance, results may imply that policymakers should incentivize not only the building of certain types of food stores in food deserts (supermarket versus convenient store), but also retail outlets with specific in-store services (such as bakeries, pharmacies, and banks). Moreover, if the results are heterogeneous across socio-demographic groups, Healthy Food Financing Initiative (HFFI) incentives potentially should be designed to fit the demographic composition of each identified food desert in question.