a= CENTER for

RETIREMENT
RESEARCH

at BOSTON COLLEGE

THE RELATIONSHIP BETWEEN DISABILITY INSURANCE RECEIPT
AND FOOD INSECURITY

Barbara A. Butrica, Stipica Mudrazija, and Jonathan Schwabish

CRR WP 2021-16
November 2021

Center for Retirement Research at Boston College
Hovey House
140 Commonwealth Avenue
Chestnut Hill, MA 02467
Tel: 617-552-1762 Fax: 617-552-0191
https://crr.bc.edu

All of the authors are with the Urban Institute. Barbara A. Butrica is a senior fellow, Stipica
Mudrazija is a senior research associate, and Jonathan Schwabish is a senior fellow. The
research reported herein was pursuant to a grant from the U.S. Social Security Administration
(SSA) funded as part of the Retirement and Disability Research Consortium (RDRC). The
findings and conclusions expressed are solely those of the authors and do not represent the views
of SSA, any agency of the federal government, the Urban Institute, or Boston College. Neither
the United States Government nor any agency thereof, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or responsibility for the accuracy,
completeness, or usefulness of the contents of this report. Reference herein to any specific
commercial product, process or service by trade name, trademark, manufacturer, or otherwise
does not necessarily constitute or imply endorsement, recommendation or favoring by the United
States Government or any agency thereof. The authors thank Craig Gundersen, Adam Dewey,
and Feeding America for making available and extending the food price and food insecurity data.
They further thank Benjamin Docter and John Walsh for excellent research assistance. The
authors also thank Craig Gundersen and Elaine Waxman for their review and comments.

© 2021, Barbara A. Butrica, Stipica Mudrazija, and Jonathan Schwabish. All rights reserved.
Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission
provided that full credit, including © notice, is given to the source.
About the Center for Retirement Research

The Center for Retirement Research at Boston College, part of a consortium that includes
parallel centers at the National Bureau of Economic Research, the University of Michigan, and
the University of Wisconsin-Madison, was established in 1998 through a grant from the U.S.
Social Security Administration. The Center's mission is to produce first-class research and forge
a strong link between the academic community and decision-makers in the public and private
sectors around an issue of critical importance to the nation's future. To achieve this mission, the
Center conducts a wide variety of research projects, transmits new findings to a broad audience,
trains new scholars, and broadens access to valuable data sources.

Center for Retirement Research at Boston College
Hovey House
140 Commonwealth Ave
Chestnut Hill, MA 02467
Tel: 617-552-1762 Fax: 617-552-0191

https://crr.bc.edu/

Affiliated Institutions:

The Brookings Institution
Mathematica - Center for Studying Disability Policy
Syracuse University
Urban Institute
Abstract

This paper examines the relationship between rates of food insecurity, people with
disabilities, and awards for the Social Security Disability Insurance (DI) program from 2010 to
2019. It uses unique data on food establishment locations combined with estimates of food
insecurity and food prices from Feeding America's Map the Meal Gap project, DI award rates
from the Social Security Administration, as well as demographic and socioeconomic information
from the American Community Survey. The paper documents the geographic variation in rates
of disability, DI awards, and food insecurity using maps and regression analyses to explore the
relationship between them. By controlling for local access to food (operationalized as the
number or share of food establishments), the paper accounts more accurately for the association
between disability and food insecurity. The findings are descriptive and should not be

interpreted causally.

The paper found that:

e Food insecurity rates are higher in counties with high rates of disability and DI receipt
than in those with lower rates of disability and DI receipt. The paper reports that counties
with high rates of disability and DI receipt have lower average food costs, but more
limited access to food sources than counties with lower rates of disability and DI receipt.

e The paper also finds that counties with high rates of disability and DI receipts have fewer
food establishments. Furthermore, counties with high rates of disability and DI receipt
also have a larger supply of unhealthy food options.

The policy implications of the findings are:

e Considering the dual challenges of people with disabilities and food insecurity is
important to providing services and programs to people and communities in need.

e Understanding how and why people apply for benefits through the DI program (as well as
nutrition programs) is likely impacted by food insecurity and access to healthy food.

e There are additional factors to consider when examining the relationship between
disability and food insecurity including local policies and structures, other service
providers such as nonprofits and community-based organizations, and how different

programs are administered and run.
Introduction

Despite a wealth of research into the determinants of food insecurity, and another strand
of research on how disability is associated with economic outcomes, our knowledge of the
relationship between disability and food insecurity remains limited. Much of the existing
research that does explore this relationship relies on publicly available data sources! that lack
measures of access to food sources, such as grocery stores and restaurants, their geographic
location, and their numbers and densities. Those studies that do consider the geographic location
of food sources often lack the nongeographic information necessary to understand fully the
relationship between disability and food insecurity.

We seek to fill this gap in the literature by using geographic data on food establishments
combined with other relevant demographic and economic information. Unlike in previous
studies, these data allow us to control for access to food-by measuring the availability and
accessibility of food establishments-thereby accounting more accurately for the association
between disability and food insecurity.

In general, we find that counties having a large share of residents with disabilities or high
rates of DI beneficiaries are also limited in their availability and accessibility to food
establishments (respectively defined as the number of food establishments per 1,000 residents
and the number of food establishments per square mile). Digging deeper, we find that these
same counties also have a larger share of food establishments that likely provide unhealthy food
options, which include small grocery stores, convenience stores, limited-service restaurants,
pharmacies, and gas stations. We find a strong correlation between many of these types of food
establishments and disability and DI rates. Importantly, we find a strong correlation between
disability and food insecurity even after controlling for access to food.

Understanding the link between disability and food insecurity has potential implications
for multiple public assistance programs including the Supplemental Nutrition Assistance
Program (SNAP), the National School Lunch and School Breakfast programs, and the Social

Security Disability Income and Supplemental Security Income programs, not to mention the

 

' For example, related studies have used the Survey of Income and Program Participation (Sonik et al. 2016; She and
Livermore 2007), the Current Population Survey (Coleman-Jensen 2020), Panel Study of Income Dynamics (Huang,
Guo, and Kim 2010), and the National Health and Nutrition Examination Survey (Heflin, Altman, and Rodriguez
2018). Guo, Huang, and Porterfield (2018) use the restricted-access National Health Interview Survey to examine
the transition to adulthood among people with disabilities.
multitude of nonprofits, food banks, and other community-based organizations that provide
assistance and support to people and families who are food insecure or have disabilities. We
view this analysis as an early step in better understanding the relationship between disability

status and food insecurity.

Existing Literature

The existing literature on the relationship between disabilities and food insecurity appears
to have four main findings. First, households with at least one person with a disability have rates
of food insecurity that are substantially higher than households in which no one has a disability
(see, for example, Brucker et al. 2015; Brucker and Coleman-Jensen 2017; Heflin, Altman, and
Rodriguez 2019; She and Livermore 2007). Brucker and Nord (2016), for example, find that
young adults with intellectual or developmental disabilities have significantly higher levels of
food insecurity, even when controlling for poverty. Coleman-Jensen and Nord (2013) find that
food insecurity rates are highest among households with working-age adults who are disabled.
And more generally, Gundersen and Ziliak (2015) report a negative association between food
insecurity and health status.

Other research has found that adults with disabilities may be food insecure because they
have limited financial resources that force trade-offs between buying food and paying for
healthcare and other needs (see, for example, Huang, Guo, and Kim 2010; She and Livermore
2007; Nord and Kantor 2006; and the scoping review by Schwartz, Buliung, and Wilson 2019).
Gettens and Henry (2019) find that DI beneficiaries mostly support their consumption with their
DI payments, which results in low levels of consumption.

Second, the relationship between disability status and food insecurity holds not only for
people from low-income households. There are a non-trivial number of people with disabilities
who are food insecure and also live in moderate-income households. Because people with
disabilities face higher expenses (relating to health care and adaptive equipment, for example)
than people without disabilities, they also require higher incomes to meet their basic needs.
Coleman-Jensen and Nord (2013) find that 13 percent of households that included an adult not in
the labor force because of a disability had incomes that were at least three times the Federal

poverty line but were also considered food insecure.
Third, mental health disabilities have larger impacts than physical disabilities on rates of
food insecurity. Maynard et al. (2018) find that most of the 39 articles in their meta-analysis of
the literature showed associations between depression and food insecurity. They also noted that
longitudinal analyses suggest that there may be a bi-directional relationship between mental
health and food insecurity-with food insecurity increasing the risk of depression symptoms or
diagnosis, as well as depression predicting food insecurity.

And fourth, the effect of disability on food insecurity varies by severity and, to a lesser
extent, the type of the disability. Balistreri (2019), for example, finds that children with more
complex healthcare needs have about twice the rate of food insufficiency compared with children
with no special healthcare needs or children with less complex healthcare needs. Gregory and
Coleman-Jensen (2017) find that rates of food insecurity vary across ten major chronic diseases
including hypertension/high blood pressure, diabetes, and chronic obstructive pulmonary disease
(see also Tarasuk et al. 2013).

Although the existing research has explored the relationship between disability status and
food insecurity, there is much less work on the relationship between receipt of DI benefits and
food insecurity. Brucker (2016) is one of the few exceptions in the literature. Like other studies,
the author finds that young adults with disabilities are more likely to live in households that are
food insecure than their counterparts without disabilities (see also Brucker and Nord 2016). The
author also finds that even young adults receiving DI were more likely to live in food insecure
households. How this translates to the intersection of disability status and food insecurity in
adulthood (not to mention the reliance on different kinds of support programs), to our
knowledge, has not yet been explored. Improving our understanding of how DI receipt intersects
with food insecurity can help us better understand how the DI program supports beneficiaries
and the future financial picture of the Social Security system.

We seek to expand the existing research on the relationship between disability and food
insecurity by considering both disability and DI receipt. In addition, we account for access to
food in this relationship, which is largely missing from previous studies. Research on the link
between food insecurity and food access-such as the finding from Bonanno and Li (2015) that
shows an increase in the density of large food stores reduces food insecurity-may be
particularly important for people with disabilities. In a report for the Morrison Institute for
Public Policy at Arizona State University, Lee and McFadden (2016) note "[t]ransportation
issues, increased risk of poverty, and a lack of support to shop for and prepare healthy meals
place people with disabilities at a greater disadvantage of living in food deserts."

Importantly, different from most studies, we consider whether food insecurity and access
to food is different for those with and without disabilities or DI benefits. Previous work, by
comparison, has explored the impact of disability (or DI beneficiary status) on food insecurity-
as we are careful to emphasize throughout, our results should not be viewed through a causal
lens but instead through a perspective of the correlation between food insecurity and disability
status or DI award receipt. Because close to two in five households with very low food security
include an adult with a disability (Coleman-Jensen and Nord 2013), we expect to find
differences. In the next section, we explain our access to unique data that enables us to shed

light on this important question.

Data

To understand better the relationship between disability status, DI receipt, and food
insecurity, we combine data from multiple sources. In this section, we describe those data, our
merging method, and additional considerations about those data that readers should be aware of

as they consider the findings in the next section.

InfoGroup Business Data (IBD)

The Urban Institute has access to the IBD, which offers a rare opportunity to explore
access to food stores among people with disabilities. The IBD is a national database of
approximately 25 million business establishments and includes the address, industry, number of
employees, and sales volume for each business in every year from 1998 through 2019. The IBD
categorizes all businesses in the US by the North American Industry Classification System
(NAICS) and the Standard Industrial Classification (SIC) codes. Using the NAICS, we
differentiate between different types of food establishments-grocery stores, other grocery
(including specialty food) stores, convenience stores, warehouse clubs, full-service restaurants,
limited-service restaurants (such as takeout eateries and fast-food restaurants), other restaurants
(including cafeterias, buffets, and snack bars), department stores, pharmacies, and gas stations.
Using the SIC description, we further limit department stores, pharmacies, and gas stations to

include only those whose description (entered in text fields in the data) includes "food markets,"
"grocers-retail," "convenience store," "food products-retail," or "foods-carry out." We also use
the IBD location name to ensure that pharmacies include all CVSs, Walgreens, and Rite-Aids,
which are known to sell food. Finally, the IBD data contains the number of employees at each
site, which enables us to differentiate between smaller grocery stores (less than 50 employees)
and larger ones (50 or more employees)." Larger grocery stores are more likely to provide
healthy food options (Treuhaft and Karpyn 2010) and a large share of food expenditures by
households at risk of food insecurity are spent at these kinds of establishments (Gregory,
Mancino, and Coleman-Jensen et al. 2019). Appendix Table A1 includes a crosswalk that maps

our food store categories to their NAICS codes.

American Community Survey (ACS)

We use the ACS for population and county-level place characteristics. These include
standard demographic and economic controls for age, educational attainment, race and ethnicity,
median household income, median house value, and total population, as well as the percent of
people in each county who reported having some kind of disability-one of our key variables of

interest.

Social Security Administration (SSA)

We obtained the number of people participating in the DI program from SSA from their
annual reports (Social Security Administration 2020). We converted these raw numbers to
percent of the county population by using population counts from the ACS data.

Map the Meal Gap

We are grateful to Feeding America for providing us with county-level rates of food
insecurity and thrifty food prices from the Map the Meal Gap project
(https://map.feedingamerica.org/). The Map the Meal Gap project is an annual study conducted

 

by the nonprofit group Feeding America since 2011 using the Census Bureau's Current

Population Survey (CPS) to estimate food insecurity. The study combines data from various

 

? These thresholds are based on our inspection of the data and observation that many of the grocery stores we know
to be large report at least 50 employees.
sources to provide county-level estimates of food insecurity (rates and numbers)*, food budget
shortfalls, and a relative cost-of-food price index.4 See Feeding America (2021) and Gundersen

et al. (2021) for more detail.

Local Area Unemployment Statistics (LAUS)
We use the LAUS data from the BLS for measures of the county unemployment rate.

National Center for Health Statistics (NCHS)
We use the NCHS 2013 urban-rural classification scheme, which distinguishes between

metropolitan, micropolitan*, and non-core areas.

Land Area

We obtain county-level land area (in square meters) from the Census Bureau.

We merged all datasets together at the county-year level. Our final data set includes
3,133 counties for each year 2010 through 2019 for a total of 31,330 county-year observations.

Methods

We estimate a series of regressions to examine the joint relationships between these
variables. At the county-level, these regressions capture the association between access to food
(measured by the number and density of food establishments) and food insecurity (measured in
the Map the Meal Gap project) and the share of the population that has a disability (from the
ACS data) or is receiving DI benefits (from the SSA), controlling for other economic and

demographic characteristics. We emphasize that we are examining the association between

 

3 In short, the Map the Meal Gap project uses state-level CPS data to estimate the relationship between food
insecurity and its determinants (e.g., poverty, unemployment, median income, etc.). The model results are then
applied to the county level.

4 The food price values were provided by Feeding America and reflect the local equivalent of the national average
amount that food secure individuals report spending on a single meal. To localize this average, they use Nielsen
pre-tax prices for a market basket that reflects the Thrifty Food Plan (TFP) in every county as well as county and
state sales taxes on groceries. The meal cost values can be used as a relative price index to calculate a TFP at a local
level, which, for our purposes, is at the county-level. The food insecurity literature has documented an important
impact of food prices on food insecurity (Gregory and Coleman-Jensen 2013; Gundersen and Ziliak 2018).

5 From the CDC reference guide, the concept of a micropolitan statistical area closely parallels that of the MSA, but
a micropolitan statistical area comprises nonmetropolitan counties and has a smaller nucleus (CDC 2014).
these factors and not ascribing a causal relationship between any of these factors and rates of
disability or DI awards. We are also not estimating a joint relationship between these correlates
even though there is surely some interaction and overlap between them. That is, we are not
looking at food affordability and accessibility of healthy foods simultaneously-we simply do
not know enough about the food establishments in our data to know for certain, for example, that
grocery stores carry healthier foods than convenience stores. Furthermore, grocery stores and
restaurants are fundamentally different businesses with respect to affordability, and future
research could explore those differences in greater detail.

We also recognize that additional factors such as federal, state, and local policies may
have important impacts on these relationships. For example, SSA policy around placement and
utilization of intake offices has been shown to have an important impact on applications for DI
benefits (Deshpande and Li 2019). Federal and state guidelines set forth by the US Department
of Agriculture determine access, eligibility, and benefits for a variety of nutrition programs such
as SNAP. Finally, there are a host of other factors that vary locally, including health behaviors
(including obesity and smoking; see, e.g., Dwyer-Lindregn et al. 2014 and Schimmel Hyde et al.
2021), environmental factors (such as air quality and lead exposure; see, e.g., Environmental
Protection Agency 2020), and the characteristics of an area (such as walkability and availability
of public transit; see, e.g., Conderino et al. 2021 and Cooksey-Stowers et al. 2017).

Our baseline regression model takes the following form:

Disabilityg, = Bo + 6, FoodEstab,;; + B,FoodInsecurity,, + B2FoodPrice,; + B3Xot

+ 6¢ +14 + Ege

where Disability,, measures the share of people with disabilities or the share receiving DI
benefits in county g in year t. FoodEstab,;; captures the number and density of various food
establishments 7 in county g in year t. FoodInsecurity,; is the food insecurity rate and
FoodPriceg, is the average price per meal, both from the Map the Meal Gap project, and
Xgt includes time-varying county-level demographic and socioeconomic characteristics. These
characteristics are age, educational attainment, race and ethnicity, homeownership, median
household income, median house value, the unemployment rate, and total population. Fixed

effects 6, captures unobserved characteristics that vary by year but not county (i.e., year fixed
effects), and 7, captures unobserved characteristics that vary by county but not year (i.e., county
fixed effects).

While we believe that it is important to approach this analysis using both the share of
people with disabilities and the share of people receiving DI benefits, there are real and
important differences between them. First, the measures come from different sources (Census
Bureau and SSA) and are collected in different ways (household survey and administrative
records). Second, the two measures are presumably (though not necessarily) related since those
receiving DI are likely a subset of the population reporting disabilities. Differences between the
rates can occur because some people's disabilities are not severe enough to warrant receiving DI
benefits or some people who might qualify for DI don't know about the program or are unable to
apply for the program. Further research could explore the differences between these two
populations and how those differences relate to (and interact with) access to food and food
insecurity.

We normalize FoodEstab,,; in two different ways to represent access to food using the
taxonomy conceptualized by Penchansky and Thomas (1981).° One set of models uses the
number of food establishments per 1,000 residents to capture the availability or supply of food
sources. Another set of models uses the number of food establishments per square mile to
capture the accessibility or location of food sources. Of course, these measures are limited
because availability and accessibility aggregated to the county-level will not reflect disparities in
access to food within counties. Finally, we also estimate the models separately for "rural" and
urban/micropolitan areas to further understand how density impacts access to food. We

designate counties as "rural" if their NCHS classification is "non-core."

Results
In this section, we present our main results, both descriptive and visual, as well as our

estimated regressions. We discuss each set of findings in turn.

 

6 The authors' taxonomy includes availability, accessibility, accommodation, affordability, and acceptability. Our
analyses capture availability, accessibility, and affordability.
Descriptive Results

Table 1 presents summary statistics of food costs, food insecurity rates, and food
establishments in 2019 for counties with and without high rates of disability or DI receipt.
Throughout the paper, we describe counties whose disability or DI receipt rates are at or above
the 75 percentile for all counties in a given year as having "high" disability or DI receipt rates.
We describe those whose rates are below the 75" percentile as having "lower" disability or DI
receipt rates. Consistent with other studies, we find that food insecurity rates are higher (around
4.5 percentage points) in counties with high rates of disability and DI receipt than those with
lower rates of disability and DI receipt. We also find that counties with high rates of disability
and DI receipt have lower average food costs, but more limited access to food sources than
counties with lower rates of disability and DI receipt.'

Map 1 shows the geographic patterns of food insecurity in counties above and below the
75 percentile of the DI receipt distribution). Mirroring other findings®, high-DI receipt counties
(shown in shades of blue)}-and the associated high rates of food insecurity-tend to be clustered
in the southeastern part of the United States and in the Appalachia region in southern West
Virginia and eastern Kentucky. We also see that most counties in Maine are considered high-DI
receipt counties, as are many counties in Michigan and New Mexico. Counties in yellow are
those with lower-DI receipt (i.e., rates of DI receipt below the 75" percentile). They are
scattered throughout the country and include several counties with high rates of food insecurity
located along the Arizona-New Mexico border and in South Dakota.

On average, there are only 2.69 food establishments per 1,000 residents in counties with
high rates of DI receipt, but 2.99 food establishments per 1,000 residents in counties with lower
rates of DI receipt (Table 1). Differences in access to food establishments are more pronounced
when considering the number per square mile. On average, there are only 0.24 food
establishments per square mile in counties with high rates of DI receipt, but 1.19 food
establishment per square mile in counties with lower rates of DI receipt. High DI areas have

fewer food options to support the size of their population and their food options are more spread

 

7 Of course, average food costs are not the same as relative food costs. Although the food price data included in the
model takes national meal costs and local area tax rates into account, two counties with the same average food costs
will have different relative costs if one county is wealthier than the other.

8 See Schimmel Hyde et al. (2020, 2021).
out than in lower DI areas. This may be especially challenging if those with disabilities have
limited mobility and limited transportation options.

We look at the total number of food establishments in each county in Map 2,
distinguishing between high- and lower-DI receipt counties. Here, high-DI receipt counties
(again in shades of blue) have fewer total food establishments per person and vary slightly across
the country. Most counties (599 or 76 percent of all these counties) have between 2 and 4 total
food establishments per 1,000 people. Lower-DI receipt counties have more food establishments
overall as well as more counties with more establishments-there are 217 counties (or 9 percent
of the total) that have between 4 and 6 establishments per 1,000 people.

In addition to having more limited access to food establishments, Table 1 shows that
counties with high rates of disability and DI receipt have a different mix of food establishments
than other counties. They have more convenience stores and gas stations and fewer full-service
restaurants per 1,000 residents than counties with lower rates of DI receipt. Counties with high
rates of disability and DI receipt also have more small grocery stores (fewer than 50 employees),
which are likely to have more limited food options, than counties with lower rates of DI receipt.
This disparity in healthy food options is more evident in Table 2, which shows the distribution of
food establishments. In counties with high rates of disability and DI receipt, food establishments
are more likely to be small grocery stores, convenience stores, and gas stations, and less likely to
be larger grocery stores and full-service restaurants. Next, we compare the share of all food
establishments that are likely to provide more healthy food options, which we classify as larger
grocery stores, warehouse clubs, full-service restaurants, and department stores, with all those
that are likely to have more unhealthy food options, including smaller grocery stores,
convenience stores, limited-service restaurants, pharmacies, and gas stations. Among all food
establishments, the bottom panel of Table 2 shows that 59.1 percent of establishments fall in the
"likely unhealthy" group and 32.3 percent fall in the "likely healthy" group in counties with a
high percentage of DI beneficiaries.' In contrast, 45.5 percent are "likely unhealthy", and 40.5

percent are "likely healthy" in counties with a lower percentage of DI beneficiaries.

 

9 We believe our definition of "likely healthy" and "likely unhealthy" food establishments follows conventional
wisdom and there is some evidence in the literature (e.g., Bonanno and Li (2015) and the citations within Treuhaft,
Karpyn (2010)) to support this taxonomy. Of course, changing these classifications-or having even better data on
the exact types of foods provided in each type of establishment-might affect our conclusions. Small grocery stores
may be the exception to this classification because they often offer a wider variety of foods than gas stations but
may still not be able to offer a full array of healthy foods such as fruits, vegetables, and lean proteins.

10
In Map 3, we take a closer look at these "likely unhealthy" food establishments around
the country, again distinguishing between the two DI density categories. Here, we see an
especially high density of these establishments in the high-DI counties (the darkest blue colors),
particularly in the Appalachia region of the country, as well as in parts of Arkansas, Mississippi,
and Alabama. There is a scattering of counties with a high concentration of unhealthy
establishments across the country, especially in pockets in Ohio and Alaska, and running
vertically through the center of the country.

These same patterns generally hold even accounting for differences in the urbanization of
counties. With one exception, food availability and accessibility are more limited in both "rural"
and urban/micropolitan counties with high disability and DI receipt than in those with lower
disability and DI receipt (Table 3). The exception is that there are significantly more total food
establishments per square mile in "rural" counties with high rates of DI receipt than in "rural"
counties with lower DI rates. The largest difference between these two types of counties is in
full-service restaurants-we find there are 1.26 full-service restaurants per 1,000 residents in
"rural" counties compared with 0.88 full-service restaurants per 1,000 residents in
urban/micropolitan counties. Also note that differences in the availability (1.e., per capita) of
food establishments by high and lower disability and DI rates are larger in "rural" counties than
in urban/micropolitan counties, while differences in the accessibility (i.e., per square mile) of
food establishments by disability and DI rates are larger in urban/micropolitan counties than in
"rural" counties.

Table 4 shows the distribution of food establishments for counties with high and lower
rates of disability and DI receipt and how the distribution varies for "rural" and
urban/micropolitan counties. As before, we find that counties with a high prevalence of
disability and DI receipt have a larger percentage of food establishments that are likely unhealthy
and a smaller percentage of food establishments that are likely healthy. However, these
differences are significantly more pronounced in urban/micropolitan counties than in "rural"
counties. In urban/micropolitan counties, for example, food establishments that are likely
unhealthy account for 58.8 percent of all food establishments in counties with high rates of DI
receipt and 45.3 percent of all food establishments in counties with lower rates of DI receipt-a
difference of 13.5 percentage points. In "rural" counties, food establishments that are likely

unhealthy account for 59.8 percent of all food establishments in counties with high rates of DI

11
receipt and 50.5 percent of all food establishments in counties with lower rates of DI receipt-a
difference of 9.4 percentage points.

What we do not know from these tabulations and the forthcoming regression analysis, is
where people eat food or do their food shopping. While we can identify the specific addresses of
the food establishments in our data, we do not know the specific addresses of people with
disabilities (or receiving DI), so we cannot see exactly where people need to travel to make their
food purchases. This is potentially important because we do not know if people are crossing
county lines to do their food shopping and whether that behavior is more or less likely in rural or
urban areas. Again, this is an area ripe for further research-by combining SSA administrative
data on DI receipt, which presumably has address information of DI beneficiaries, with our IDB
data, we could at least get a better sense of specific access and availability to food

establishments.

Regression Results

Previous literature has confirmed the relationship between disability and food insecurity.
Our descriptive results also suggest a relationship between disability or DI receipt, and food
access. Next, we consider the relationship between disability and food insecurity, controlling for
access to food and other factors. Table 5 presents the coefficients from fixed effects models of
the county-level disability rate for the primary variables of interest (see Appendix Table A2 for
the full results). Models 1-3 use the number of food establishments per 1,000 residents and
models 4-6 use the number of food establishments per square mile. As in previous studies, we
find a positive correlation between food insecurity and disability. Regarding the availability
(supply) of food establishments, Model 1 shows that small grocery stores and department stores
are positively and significantly correlated, and warehouse clubs are negatively and significantly
correlated with disability rates (shaded cells). However, "rural" counties appear to drive these
results (Model 2). For urban/micropolitan counties, in contrast, convenience stores, limited-
service restaurants, and gas stations are positively correlated with disability rates (Model 3).!°

Regarding the accessibility (location) of food establishments, Model 4 shows a positive

association with disability rates for convenience stores and limited-service restaurants, and a

 

10 We also estimated the main regressions (Table 5) by dropping the bottom and top 1 percent of counties ranked by
the population in 2019. Those results were largely similar to the ones reported here.

12
negative association with disability rates for warehouse clubs and full-service restaurants.
Somewhat surprising is that large grocery stores are also positively and significantly correlated
with disability rates. This finding could be due to any number of factors including access to
transportation and the location of these establishments; additional research could explore specific
areas in greater detail to see how transportation access and availability may affect these results.
Again, there are some notable differences between "rural" and urban/micropolitan counties.
Accessibility to convenience stores and gas stations is positively correlated with disability rates
in "rural" counties, but not in urban/micropolitan counties. In urban/micropolitan counties,
limited-service restaurants are positively correlated with disability and full-service restaurants
are negatively correlated with disability. In contrast, restaurant location is not correlated with
disability in "rural" counties.

Table 6 presents the coefficients from fixed effects models of DI receipt (see Appendix
Table A3 for the full results). Again, models 1-3 use the number of food establishments per
1,000 residents and models 4-6 use the number of food establishments per square mile. It is
interesting to note that food insecurity is correlated with DI participation overall and in
urban/micropolitan counties, but it is uncorrelated with DI participation in "rural" counties.
Turning to the coefficients on food establishments, we generally find smaller point estimates and
fewer that are statistically significant than in the models of disability rates (table 5); however,
most of the signs on the coefficients are the same between the models. As noted previously,
there is no reason to expect the results to be the same because the disability and DI populations
and the underlying data sources differ.

Model 1 shows that the availability of small grocery stores, convenience stores, other
restaurants (such as buffets), and gas stations is positively and significantly associated with the
county-level DI beneficiary rate, while the supply of large grocery stores is negatively and
significantly associated with rates of DI receipt. "Rural" counties seem to be the main driver of
these results as is evidenced by Model 2. In contrast, small grocery stores and other restaurants
are the only food establishments that are correlated (positively) with the share of DI beneficiaries
in urban/micropolitan counties (Model 3).

Regarding accessibility, small grocery stores and gas stations are the only food
establishments associated (positively) with the share of DI beneficiaries (Model 4). As with

disability (table 5), we find large differences between "rural" and urban/micropolitan counties.

13
Notably, we find that access to food establishments is more highly correlated with the DI
beneficiary rate in "rural" counties than in urban/micropolitan counties. In "rural" counties,
small grocery stores and gas stations have a significant positive relationship with rates of DI
receipt, while large grocery stores, specialty grocery stores, and full-service restaurants have a
significant negative relationship with rates of DI receipt. Unexpected is that pharmacies also

have a statistically significant negative relationship with rates of DI receipt in "rural" counties.

Select County Profiles

Table 7 profiles the disability, DI receipt, and food establishments of several illustrative
counties across the country in 2019. They range in population and area from Los Angeles, CA,
which is large in both population and area to Billings, ND, which is small in population, and to
Hickman, KY, which is small in area. We explore these counties only to demonstrate the
variation in the different components of this analysis and how a more detailed look at specific
areas of the country (requiring, of course, more geographically detailed disability and DI data)
could provide greater insight into the challenges people with disabilities face accessing healthy
foods.

Among the counties profiled, Hickman, KY has the highest percentage of residents with
disabilities (24.3 percent) and the highest percentage of DI beneficiaries (6.2 percent). It also has
the highest food insecurity rate (14.3 percent) and the lowest average cost per meal ($2.79).
Consistent with our previous findings, it also has a relatively limited number of food
establishments-only 2.44 food establishments per 1,000 residents and 0.05 food establishments
per square mile. Not only are the availability and accessibility of food establishments limited,
but more than half (54.5 percent) of them likely provide mostly unhealthy food options. Of the
11 food establishments we identify in this county, more than half are small grocery stores,
conveniences stores, limited-service restaurants, and gas stations. Again, we do not know what
kind of food these establishments sell; however, because of their size or the nature of their
primary business (e.g., selling gasoline), it seems likely that their food selection is fairly limited
and primarily convenient-qualities often associated with unhealthy foods.

In contrast, Billings, ND has a lower-than-average disability rate (13.8 percent) and very
low DI rate (0.5 percent). Its food insecurity rate is the lowest of the counties profiled (5.6

percent) and its average cost per meal is the highest ($3.70). It has the largest supply of food

14
establishments (13.02 per 1,000 residents) but is one of the least food accessible (0.01 food
establishments per square mile). Yet only a quarter (25.0 percent) of its food establishments are
likely unhealthy.

Los Angeles, CA and Providence, RI-the two largest counties we include-sit
somewhere in the middle of these other two counties. About 10 percent of people in Los
Angeles have a disability but only 1.5 percent are receiving DI benefits. The percentage point
gap between the share of people with disabilities and the DI award rate are lower in these two
counties than across the country or in the other two counties we identified. Compared with the
country as a whole, both counties have more food establishments per capita (3.26 and 3.15,
compared with 2.92) and per square mile (8.11 and 4.89 compared with 0.95). While the share
of likely unhealthy food establishments in Providence is on par with the nation (39.8 versus 38.9
percent), it is significantly lower in Los Angeles (30.6 percent).

Figure 1 shows the positive association between rates of food insecurity and DI

beneficiaries for all counties in the country with these selected counties highlighted in yellow.

Discussion

With nearly 10 million people receiving DI benefits, 35 million Americans estimated to
be food insecure, and 34 million people in poverty in 2019, there is a clear and obvious need to
understand better how these factors are linked together. In this paper, we capture more
accurately the relationship between disability and food insecurity by accounting for access to
food using two measures and a unique data set of food establishments that is more up to date
than other publicly available data sources. We find that counties with a large share of residents
with disabilities and high DI participation are also limited in their availability and accessibility to
food establishments. These same counties also have a larger share of food establishments that
likely provide mostly unhealthy food options, based on the assumption that certain categories of
food outlets have limited availability of fruits, vegetables, lean proteins, and dairy and a greater
abundance of snack foods and sugar-sweetened beverages. Importantly, we find a strong
correlation between disability and food insecurity even after controlling for access to food. A 1
percentage point increase in the food insecurity rate is correlated with a 0.961 percentage point
(6.1 percent) increase in the disability rate and a 0.012 percentage point (0.3 percent) increase in

the DI beneficiary rate.

15
Understanding the areas in which high rates of disability and access to food or food
insecurity are concentrated may highlight the potential for targeted interventions that could assist
workers, families, and children. SSA, as well as the USDA and various nonprofit and other
stakeholder groups, could use this information to target both pre- and post-entitlement supports
for affected people and communities.

Our multivariate analysis sheds some light on the relationship between disability and
food insecurity. The models incorporate predictors that are readily available from public sources
as well as the restricted InfoGroup data that contains location-specific information on food
establishments across the country. It is important to note that we assign a correlation
interpretation to our results and do not make any claims about causality in the outcomes that we
observe.

There are caveats to this analysis that are worth considering. First, our measures of food
access, food insecurity, and food prices are incomplete. We are not able to look consistently at
affordability and accessibility of healthy foods. In some cases, such as large grocery stores, we
assume that they are better able to carry healthier foods, but we are unable to say much about
their affordability with our data. In other cases, such as fast-food restaurants, we assume that
they carry both less healthy and more affordable foods. Second, any comparisons along the
metric of affordability are incomplete-comparing affordability of restaurants to grocery stores,
for example, is more complicated than a simple price comparison. Third, there are a slew of
unmeasured factors that we have not accounted for in our modeling including, for example,
administrative policies in the DI program, or factors that might be difficult to measure
systematically such as social networks or access to food banks.

As such, it may be important to extend the work conducted here along several
dimensions. First, better, more consistent data measured at a more granular geographic level
(e.g., administrative SSA data with specific addresses for DI beneficiaries) could generate more
accurate accessibility information. Second, data on different types of disabilities could be used
to improve our understanding of who is applying for DI benefits and how those types of
disabilities (and health status more generally) are correlated with measures of food insecurity.
Third, other measures of access to food (especially healthy food) such as access to food banks
and other nonprofits, as well as rates of program uptake, would broaden the perspective on the

nutrition safety net for needy families. And finally, incorporating qualitative information would

16
add richness to the analysis-speaking with disability and nutrition advocates, state and local
program administrators, and, of course, people with disabilities, and people and families who are

food insecure.

Conclusion

In this paper, we account for access to food-by measuring the availability and
accessibility of food stores and restaurants within each county-in the relationship between rates
of disability and DI beneficiaries, and food insecurity. In general, we find that counties having a
large share of residents with disabilities or high rates of DI beneficiaries are also limited in their
availability and accessibility to food establishments. Digging deeper, we find that these same
counties also have a larger share of food establishments that likely provide unhealthy food
options. These include small grocery stores, convenience stores, limited-service restaurants,
pharmacies, and gas stations. We find a strong correlation between many of these types of food
establishments and disability and DI beneficiary rates. Importantly, we find a strong correlation
between disability and food insecurity even after controlling for access to food. Our results
should not be interpreted through a causal lens, however, as they only indicate a correlation
between these factors. Linked data, more geographically-specific data, and more information on

specific types of disabilities would all be ways to extend this line of research.

17
References

Balistreri, Kelly Stamper. 2019. "Food Insufficiency and Children with Special Healthcare
Needs." Public Health 167: 55-61.

Bonanno, Alessandro and Jing Li. 2015. "Food Insecurity and Food Access in US Metropolitan
Areas." Applied Economic Perspectives and Policy 37(2): 177-204.

Brucker, Debra L. and Alisha Coleman-Jensen. 2017. "Food Insecurity Across the Adult Life
Span for Persons with Disabilities." Journal of Disability Policy Studies 28(2): 109-118.

Brucker, Debra L. 2016. "Food Insecurity Among Young Adults with Disabilities in the United
States: Findings from the National Health Interview Survey." Disability and Health
Journal 9: 298-305.

Brucker Debra L. and Derek Nord. 2016. "Food Insecurity Among Young Adults with
Intellectual and Developmental Disabilities in the United States: Evidence from the
National Health Interview Survey." AJIDD-American Journal on Intellectual and
Developmental Disabilities 121(6): 520-532.

Brucker, Debra L., Sophie Mitra, Navena Chaitoo, and Joseph Mauro. 2015. "More Likely to be
Poor Whatever the Measure: Working-Age Persons with Disabilities in the United
States." Social Science Quarterly 96(1): 273-296.

Caspi, Catilin E., Glorian Sorensen, $.V. Subramanian, and Ichiro Kawachi. 2012. "The Local
Food Environment and Diet: A Systematic Review." Health Place 18(5): 1172-1187.

Centers for Disease Control and Prevention (CDC). 2014. "2013 NCHS Urban-Rural
Classification Scheme for Counties." Vital and Health Statistics 2(166). Atlanta, GA.
Available at: https://www.cdc.gov/nchs/data/series/sr_02/sr02_166.pdf

Coleman-Jensen, A. 2020. "US Food Insecurity and Population Trends with a Focus on Adults
with Disabilities." Physiology & Behavior 220.

Coleman-Jensen, Alisha and Mark Nord. 2013. Food Insecurity Among Households with
Working-Age Adults with Disabilities. ERR-144. Washington, DC: U.S. Department of
Agriculture, Economic Research Service.

Conderino, Sarah E., Justin M. Feldman, Benjamin Spoer, Marc N. Gourevitch, and Lorna E.
Thorpe. 2021 (forthcoming). "Social and Economic Differences in Neighborhood
Walkability Across 500 U.S. Cities." American Journal of Preventive Medicine.

Cooksey-Stowers, Kristen, Marlene B. Schwartz, and Kelly D. Brownell. 2017. "Food Swamps

Predict Obesity Rates Better Than Food Deserts in the United States." International
Journal of Environmental Research and Public Health 14(11): 1366.

18
Deshpande, Manasi and Yue Li. 2019. "Who Is Screened Out? Application Costs and the
Targeting of Disability Programs." American Economic Journal: Economic Policy 11(4):
213-248.

Dwyer-Lindgren, L., A. H. Mokdad, T. Srebotnjak, A. D. Flaxman, G. M. Hansen, and C. J. L.
Murray. 2014. "Cigarette Smoking Prevalence in US Counties: 1996-2012." Population
Health Metrics 12(1).

Environmental Protection Agency. 2020. "Environmental Quality Index - Technical Report
(2006-2010)." Washington, DC: Available at:
https://cfpub.epa.gov/ncea/risk/recordisplay.cfm?deid=350545

Feeding America. 2021. "Map the Meal Gap 2021: Technical Appendix." Chicago, IL. Available
at: https://www.feedingamerica.org/sites/default/files/2021-
05/Map%20the%20Meal%20Gap%202021%20Technical%20Brief.pdf

Gettens, Jack and Alexis Henry. 2019. "Making Ends Meet: How Low-Income DI Beneficiaries
Meet Their Needs." DRC Working Paper 2019-02. Chicago, IL. Mathematica.

Gregory, Christian A., Lisa Mancino, and Alisha Coleman-Jensen. 2019. "Food Security and
Food Purchase Quality Among Low-Income Households: Findings from the National
Household Food Acquisition and Purchase Survey (Foodaps)." Economic Research
Report Number 269. Washington, DC: U.S. Department of Agriculture Economic
Research Service.

Gregory, Christian A. and Alisha Coleman-Jensen. 2017. "Food Insecurity, Chronic Disease, and
Health Among Working-Age Adults." Economic Research Report Number 235.
Washington, DC: U.S. Department of Agriculture, Economic Research Service.

Gregory, C. and A. Coleman-Jensen. 2013. "Do High Food Prices Increase Food Insecurity in
the United States?" Applied Economic Perspectives and Policy 35: 679-707.

Gundersen, Craig and James P. Ziliak. 2018. "Food Insecurity Research in the United States:
Where We Have Been and Where We Need to Go?" Applied Economic Perspectives and
Policy 40(1): 119-135.

Gundersen, Craig and James P. Ziliak. 2015. "Food Insecurity and Health Outcomes." Health
Affairs 34(11): 1830-1839.

Gundersen, Craig, Monica Hake, Adam Dewey, and Emily Engelhard. 2021. "Food Insecurity
during COVID-19." Applied Economic Perspectives and Policy 43(1): 153-161.

Guo, B., Huang, J. and S. L. Porterfield. 2018. "Transition to Adulthood: Dynamics of

Disability, Food Security, Health, and SNAP Participation." Discussion Paper Series
DP2018-01. Lexington, KY: University of Kentucky, Center for Poverty Research.

19
Heflin, Colleen M., Claire E. Altman, and Laura L. Rodriguez. 2019. "Food Insecurity and
Disability in the United States." Disability and Health Journal 12: 220-226.

Huang, Jin, Baorong Guo, and Youngmi Kim. 2010. "Food Insecurity and Disability: Do
Economic Resources Matter?" Social Science Research 39: 111-124.

Lee, Sang Enu and Erica S. McFadden. 2016. "Rethinking 'Food Deserts' and People with
Disabilities." Phoenix, AZ: Arizona State University, Morrison Institute for Public
Policy. Available at:

https://addpc.az.gov/sites/default/files/media/Nutrition'o20Brief_1.pdf

Maynard, Merryn, Lesley Andrade, Sara Packull-McCormick, Christopher M. Perlman, Cesar
Leos-Toro, and Sharon I. Kirkpatrick. 2018. "Food Insecurity and Mental Health among
Females in High-income Countries." International Journal of Environmental Research
and Public Health 15(7): 1424.

Nord, Mark and Linda S. Kantor. 2006. "Seasonal Variation in Food Insecurity Is Associated
with Heating and Cooling Costs among Low-Income Elderly Americans." The Journal of
Nutrition 136(11): 2939-2944.

Penchansky, Roy and J. William Thomas. 1981. "The Concept of Access: Definition and
Relationship to Consumer Satisfaction." Medical Care 19(2): 127-140.

Schimmel Hyde, Jody, Jonathan Schwabish, Paul O' Leary and Dara Lee Luca. 2021a
(forthcoming). "Understanding the Local-Level Predictors of Disability Program Flows:
New Adult Awards and Beneficiary Work Activity." Working Paper. Chestnut Hill, MA:
Center for Retirement Research at Boston College.

Schimmel Hyde, Jody, Anna Hill, Jonathan Schwabish, and Aaron R. Williams. 2021b
(forthcoming). "Are There 'Hot Spots' of Primary Impairments Among New SSDI
Awardees-and Do We Know Why?" Working Paper. Chestnut Hill, MA: Center for
Retirement Research at Boston College.

Schwartz, Naomi, Ron Buliung, and Kathi Wilson. 2019. "Disability and Food Access and
Insecurity: A Scoping Review of the Literature." Health & Place 57: 107-121.

She, Peiyun and Gina A. Livermore. 2007. "Material Hardship, Poverty, and Disability Among
Working-Age Adults." Social Science Quarterly 88(4): 970-989.

Sonik, Rajan, Susan L. Parish, Subharati Ghosh, and Leah Igdalsky. 2016. "Food Insecurity in
US Households That Include Children with Disabilities." Exceptional Children 83(1): 42-
57.

Tarasuk, V., A. Mitchell, L. McLaren, and L. McIntyre. 2013. "Chronic Physical and Mental

Health Conditions among Adults May Increase Vulnerability to Household Food
Insecurity." The Journal of Nutrition 143(11): 1785-1793.

20
Treuhaft, Sarah and Allison Karpyn. 2010. "The Grocery Gap: Who Has Access to Healthy Food
and Why It Matters." Oakland, CA: PolicyLink. Available at:
http://thefoodtrust.org/uploads/media_items/grocerygap.original.pdf

U.S. Social Security Administration. 2020. "OASDI Beneficiaries by State and County, 2019."

Washington, DC. Available at:
https://www.ssa.gov/policy/docs/statcomps/oasdi_sc/2019/index.html

21
Maps

Map 1. Percentage of County-Level Food Insecure Residents in 2019, by Areas with Low and
High Rates of DI Receipt

 

< 75% a 7st
percentile percentile
30% 30%
20% 20%
10% 10%
0% 0%

Note: Percentiles based on the distribution of rates of disability or DI receipt awards across the country.
Source: Authors' calculations using merged data. See text for details.

22
Map 2. Number of Food Establishments (per 1,000 people) in 2019, by Areas with Low and High
Rates of DI Receipt

 

percentile percentile

1,797

 

 

599
273
19 137 | | 42 4 7
0-2 0-2 6-8 8+
Number of food establishments Number of food establishments

Note: Percentiles based on the distribution of rates of disability or DI receipt awards across the country.
Source: Authors' calculations using merged data. See text for details.

23
Map 3. Percentage of Food Establishments That Are "Likely Unhealthy" in 2019, by Areas with

Low and High Rates of DI Receipt

 

<= 75°
percentile
100%
75%
50%
25%
0%

> 7 5th
percentile
100%
75%
50%
25%
0%

Note: Percentiles based on the distribution of rates of disability or DI receipt awards across the country. "Likely
Unhealthy" establishments are defined as grocery stores with fewer than 50 employees; convenience stores; limited-

service restaurants; pharmacies; and gas stations.

Source: Authors' calculations using merged data. See text for details.

24
Figure

Figure 1. DI Beneficiary Rate and Food Insecurity is Positively Associated

© 2500000 ©» 5000000 oy 7500000 3 10000000
Percent food insecure
30%

20%

 

Hickman, KY
Los Angeles, CA
e .
10% Providence, RI
Billings, ND
T T T T
0% 5% 10% 15%

DI beneficiaries (percent)

Note: Size of circle represents population in each county; yellow circles represent those counties listed in Table 7.
Source: Author's calculations using data from the U.S. Social Security Administration and Feeding America.

25
Tables

Table 1. County-Level Food Statistics in 2019, by Areas with Low and High Rates of
Disability and DI Receipt

 

 

% Disabled % DI Beneficiaries
< 75th > 75th < 75th > 75th
Percentile Percentile Percentile Percentile
% Food Insecure 12.0% 16.6% 12.0% 16.5%
Mean Cost Per Meal $3.22 $3.08 $3.22 $3.07
Number Per 1,000
Residents
Grocery 0.31 0.38 0.32 0.35
< 50 Employees 0.27 0.34 0.28 0.31
50+ Employees 0.05 0.04 0.05 0.04
Missing 0.00 0.00 0.00 0.00
Grocery Other 0.15 0.11 0.16 0.10
Convenience Store 0.32 0.38 0.33 0.37
Warehouse Club 0.00 0.00 0.00 0.00
Restaurant Full Service 1.05 0.93 1.08 0.84
Restaurant Limited
Service 0.71 0.65 0.70 0.68
Restaurant Other 0.18 0.12 0.19 0.10
Department Store 0.01 0.01 0.01 0.01
Pharmacy 0.02 0.02 0.02 0.02
Gas Station 0.19 0.25 0.19 0.23
Total Food
Establishments 2.94 2.84 2.99 2.69
Number Per Square Mile
Grocery 0.11 0.02 0.10 0.03
< 50 Employees 0.09 0.02 0.09 0.02
50+ Employees 0.02 0.00 0.02 0.00
Missing 0.00 0.00 0.00 0.00
Grocery Other 0.06 0.01 0.06 0.01
Convenience Store 0.07 0.02 0.07 0.03
Warehouse Club 0.00 0.00 0.00 0.00
Restaurant Full Service 0.53 0.05 0.52 0.07
Restaurant Limited
Service 0.31 0.05 0.30 0.08
Restaurant Other 0.11 0.01 0.10 0.01
Department Store 0.00 0.00 0.00 0.00
Pharmacy 0.01 0.00 0.01 0.00
Gas Station 0.02 0.01 0.02 0.02
Total Food
Kstablishments 1.22 0.16 1.19 0.24

 

Notes: Percentiles defined on the distribution of DI awards across the country. Shaded cells are those
for which counties > 75th percentile are greater than for counties < 75th percentile.
Source: Authors' calculations using merged data. See text for details.

26
Table 2. Distribution of Food Establishments in 2019, by Areas with Low and
High Rates of Disability and DI Receipt

 

 

 

 

 

 

 

% Disabled % DI Beneficiaries
< 75th > 75th < 75th > 75th
Percentile Percentile Percentile Percentile
Grocery 8.5% 11.3% 8.4% 11.3%
< 50 Employees 6.6 9.7 6.6 9.6
50+ Employees 1.8 1.6 1.8 1.6
Missing 0.0 0.0 0.0 0.0
Grocery Other 5.3 4.2 5.3 4.1
Convenience Store 73 12.9 7.2 13.0
Warehouse Club 0.2 0.0 0.2 0.1
Restaurant Full Service 38.1 31.6 38.3 30.3
Restaurant Limited
Service 27.8 26.8 27.7 28.4
Restaurant Other 8.6 45 8.6 4.6
Department Store 0.2 0.3 0.2 0.3
Pharmacy 1.2 0.9 1.2 1.0
Gas Station 2.9 7A 2.8 7.1
Total Food
Establishments 100.0 100.0 100.0 100.0
% Disabled % DI Beneficiaries
< 75th > 75th < 75th > 75th
Percentile Percentile Percentile Percentile
Likely Healthy 40.3% 33.5% 40.5% 32.3%
Likely Unhealthy 45.8% 57.7% 45.5% 59.1%
Unknown 13.9% 8.8% 14.0% 8.6%

 

Notes: Percentiles based on the distribution of rates of disability or DI receipt across the country.
"Likely Healthy" establishments are defined as grocery stores with 50 or more employees;
warehouse clubs; and full-service restaurants. "Likely Unhealthy" establishments are defined as
grocery stores with fewer than 50 employees; convenience stores; limited-service restaurants;
pharmacies; and gas stations. Shaded cells are those for which counties > 75th percentile are
greater than for counties < 75th percentile.

Source: Authors' calculations using merged data. See text for details.

27
 

09°¢
070
c0'0
10°0
oro
€L'0
6L°0
00°0
se0
60°0
00°0
¥0°0
Lo0
1¢0

LO'E$
8S

LL?
vI'0
£00
10'0
020
LL''0
96°0
00°0
870
STO
00°0
$00
610
v0

v7 E$
"8 IT

9L7 sere
970 870
c0°0 100
10°0 00°0
60°0 LY0
£90 8¢'0
88°0 9T'T
00°0 00°0
6£°0 Iv'0
ITO 810
00°0 00°0
€0°0 v0'0
seo evo
8£°0 Lv0
LO'e$ 07 ES
%0°L1 STI

cot
1c0
c0'0
10°0
ITO
0L'0
c8°0
00°0
veo
oro
00°0
v0°0
970
0c0

OT'E$
AHCI

9L'T
vI'0
£00
10°0
610
840
660
00°0
87'°0
vI'0
00°0
$00
610
v0

eT e$
%6' IT

LOT LTE
80 L70
10°0 10°0
10°0 10°0
c10 910
19°0 85°0
66°0 COT
00°0 00°0
Iv'0 oro
clo LTO
00°0 00°0
£0°0 700
6¢°0 Ir'0
cro sv'0
LO'€$ 07 ES
%6 91 %T TI

SJUSMAYST[Qvsy Poo [201
UOTRIS sey
AdeuIeYd
9101S Juowedaq
IYO JueINVsoyY
SOIAIOS poy] yueINE}soY
STAINS [Ny UeINEsoY
qnyD ssnoyare A.
910]§ SOUSTUSATIO"Z
IOYIO AIS90IN
SUISSIJ
sooXoidurgq +0¢
sooAoldurq og >
AIQ001£)

SJUIPISeY 00O'T 19d J9quiny

[R28 Jog S00
aINdasu] pooj%

 

qyuedieg 9HUI10g

qyuUsteg syuasd10g

quedieg 9U8010g

qyueaieg 9yued19g

 

 

WSL <= WSL > WSL = WSL > WSL= WSL > WSL= WSL >
ney odors Aueqay) lB, mej odors A/ueqay ulBINY,,
soBpyoueg 10 % Pages %

UDIYOdOADIPY/UDGA) PUD ,, JOANN,
Aq pun jdiasay Iq puv Ayiqvsiq fo sajoy y3ify puv samo yum svasy Ag '607 Ul SOUSIDIS poo. JaaaT-AqunoZ *¢ 9[GQe I]
6¢

"S[Iejop OJ 1X9} 99G "eJEp poSIoul BuIsn suOTE[NoTLO ,sIONINY "aaunog

'g]QuUa0IAd USL > SoUNOD JO} UETY

Joyeord ore o[TUsOIEd ING, < SOTJUNOD YOIYA JOF Oso} are s[[90 pepeys "AyUNOS oY} ssosoe spleMe JC JO VONNGLSIp oy} UO poulfep soTI}Usoleg :So}ON,

 

 

 

9¢°0 88'T st0 L0°0 0c0 €3'T 60°0 oro SJUITIYS Ges poo, [vO],
70°0 £00 10°0 00°0 70°0 £00 10°0 10°0 TOHeIS SBD
00°0 70°0 00°0 00°0 00°0 700 00°0 00°0 Anvueyd
00°0 00°0 00°0 00°0 00°0 00°0 00°0 00°0 a1o}g yusuedaq
700 LT0 00°0 00°0 10°0 9T°0 00°0 00°0 JOYIO JuLINe\sy
ITO Lv'0 v0°0 Z0°0 60°0 970 z0°0 £0°0 SOTAIOS POPU] JUCINEISOY
ITO 780 v0°0 £0°0 60°0 6L°0 £0°0 v0°0 BDIAIOG [NJ JUeMesoy
00°0 00°0 00°0 00°0 00°0 00°0 00°0 00°0 qnyD ssnoyore A
¢0'0 O10 c0'0 T0°0 v0°0 Oro 100 10°0 S10]g SoUSTUSAO,)
T0°0 oro 00°0 00°0 T0°0 60°0 00°0 00°0 ToqIO AIB001N)
00°0 00°0 00°0 00°0 00°0 00°0 00°0 00°0 BUISST
TO°0 £00 00°0 00°0 10°0 700 00°0 00°0 soohojdurg +0¢
£00 vI0 c0'0 T0°0 70°0 vI'0 10°0 10°0 sookordurg 0¢ >
+0°0 L10 c0'0 10°0 £00 9T°0 10°0 10°0 AI990IL)
aA eenbs 19g 19quNny
qyuI.19g «aWUI010g qpus19g 3QUID10g quaa1g = 9yN9919g qgusd.19g = 3HU99.10g
WSL< WSL> WSL< WsL> WSL< WsL> Wsl< WsL>
ues odo.1syA/ueqiy win, Uy odo. TA/ueg.sy lean,

(panui1juod) UDJ1JOAOADIA/UDGAL) PUD ,, JDANY,,
Aq pup jdiazay Iq pup Ayypiqosiq fo sajoy ysif] pup samoTT yim spaay Aq '6 [QZ Ul SOUSIIDIG poo. JaaaT-Kuno *¢ s[Qe J
0€

"S[IBJop OJ 1X9} 99g "eJep poBIou Bursn suoTE[NoTeo ,sIOWINY "aounog

'gruaoiad WIS /, > SOUND IO} URY) JoyBOId ore opMUDOIEd YIC¢/

= SATJUNOS YOIYM JO} SSOY} ov S{[99 papeys 'suoneys sed pus 'soroevueyd 'syuviNe}sal SITAIBS-Po}IUUT] [SoJ0}S DOUSTUSAUIOD 'sooAOTdus YS WRU} IOMay YIM
So10}s AIODOIS sv PSUTJop ore syuoWYsTqviso ,AUITVOyUL) AOA!'],, "S}UBINYISAI SOTAIOS-]]NY pus 'sqnjd ssnoyarem 'sooAO]AuIe SIOU IO YG YIM Sarloys AIO00I38
SB pouljap ore syuouTYsi[quise ,,Aqyeop Apr], 'AUNod oy] ssoloe spreMme diodel Iq] JO Aj[Iqesip JO sayel JO VOLNGLYSIp ay} UO poseq so]USAIOg :So]ON,

 

 

 

 

 

 

 

 

%1°6 %YIyT MCL YT %Il"6 %OrI %7B %Z Or umouyuy)
%QBs MES %8'6S %SOS %COLS %ISy %0'8S YES Ayppeoquy, ATOxT
MITE %I'OP LTE Wp' BE %MEEE %y' OP %YEE %ELE Ayyeoy APOVT

qusd.19g 9 099.19g qHus19g «=: BUNI0g yHU19g «8 = ayuad.g yHU19g «=. 9UaI.19g
WSL< WSL > WSL = WSL > WSL< WSL > WSL <= WSL >
uv odo1sTyA/uegin ulTBINy,, ue Todo.1dTA/Ueg.1/) ulBany,,
SILIBIDIJIUIG Id % PIrqesiqd %
00T 001 OOT OOT 001 '00T '00T "001 syuOUTysT[quisy
poo [20],
79 LZ L'8 TL L9 LZ ¢°8 EL WOTIRIS SED
TI 71 L0 50 I'l ZI 90 $0 Aoeuueyd
z0 70 70 70 70 z0 €0 70 aloyg yuouredaq
rs 3°8 €€ ws ¢ L'8 gE 67 JOYIO JuBINE}soYy
T'0€ "SZ Cv IZ 162 "SZ VEZ C IZ SOIAIOg
pour] jueINe\soyY
'OE €8E 'TE 9°9€ Ele T'8¢ UZ V'9€ SOIAING [NJ JUeIMEsOY
'0 Z0 00 00 10 z0 00 00 qn{D asnoysre MA
STI OL Tr UZ TTI TL 6€I TTI a10}§ SOUSTSAUOD
Y es ar 6S I'v ES vy 6's I3q}IQ Al99015)
00 00 00 00 00 00 00 00 SUISSI
ST 6'T v'T oT Lt 6T rT oT soohojdurg +0¢
9°8 v9 611 L6 $8 ¢9 Cl 66 soohojdury 0¢ >
~%r'OI MER MEET My TT %T7OI %E"R YETI MOTT AI990ID)
alHusd19g ayMad19g qyua19g «=. aUADA10g HuIIIg «= ayuads19g yHuaI9g «=: ayUaIIIg
WSL< WSL > WSL < WSL > WSL< WSL > WSL < WSL >
DeyodorsA/uEqiy uly, uepodossTA/ueqi wlBany,,
soLIBPousg Id % PaqeSIG %

UDIIJOdOANY/UDGA]) PUD ,, [DANY,,

Aq pun sdiaoay Tq puv 4yiqosiq, fo savy ySipy pun samoT yy svaay Aq '6107 UE SJuamYsquisy poo, fo uoyNqLAsIC "y QPL.
Table 5. Fixed Effects Regression Results of the County-Level Disability Rate

 

 

Uses Number of Food Uses Number of Food Establishments
Establishments Per 1,000 Residents Per Square Mile
Urban/ Urban/
All "Rural" Micropolitan All "Rural" Micropolitan
(1) (2) (3) (4) (5) (6)
Grocery Size < 50 Employees 0.003* 0.003 0.005 -0.001 -0.021 0.000
(0.002) (0.002) (0.004) (0.001) (0.021) (0.001)
Grocery Size 50+ Employees -0.001 -0.002 -0.000 0.009** -0.047 0.011**
(0.005) (0.007) (0.008) (0.004) (0.040) (0.004)
Grocery Size Missing 0.069 0.025 0.141* 0.181*** 3.230 0.137**
(0.046) (0.041) (0.081) (0.060) (3.437) (0.058)
Grocery Other 0.002 0.003 -0.002 -0.002 0.012 -0.002**
(0.002) (0.002) (0.004) (0.001) (0.009) (0.001)
Convenience Store 0.000 -0.001 0.005** 0.006* 0.035** 0.002
(0.002) (0.002) (0.002) (0.003) (0.017) (0.002)
Warehouse Club -0.082*** - -0.073** -0.086 -0.081** - -2.763*** -0.070*
(0.028) (0.031) (0.055) (0.038) (0.748) (0.039)
Restaurant Full Service 0.000 0.000 0.001 -0.000** 0.003 -0.000***
(0.001) (0.001) (0.002) (0.000) (0.003) (0.000)
Restaurant Limited Service -0.001 -0.004 0.004* 0.002** 0.001 0.002**
(0.002) (0.002) (0.002) (0.001) (0.013) (0.001)
Restaurant Other -0.002 -0.002 -0.002 0.002 -0.009 0.001
(0.002) (0.003) (0.003) (0.003) (0.014) (0.002)
Department Store 0.024***  0.026*** 0.019 0.013 0.775 0.000
(0.007) (0.008) (0.017) (0.019) (0.569) (0.018)
Pharmacy 0.017 0.003 0.020 -0.004 0.073* -0.004*
(0.010) (0.017) (0.012) (0.003) (0.038) (0.002)
Gas Station 0.002 0.001 0.006* 0.006 0.075*** -0.002
(0.002) (0.002) (0.003) (0.009) (0.028) (0.007)
% Food Insecure 0.961*** 0.935*** 1.005*** 0.963*** 0.941 *** 1.001***
(0.034) (0.047) (0.041) (0.035) (0.048) (0.041)
Cost Per Meal -0.000 -0.002 0.001 -0.000 -0.001 0.001
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
R-Squared 0.284 0.260 0.340 0.282 0.260 0.336
N 31,330 13,290 18,040 31,330 13,290 18,040

 

Notes: All regressions include demographic and socioeconomic controls, year fixed effects, and county fixed effects.
Shaded cells are statistically significant results for the food establishment variables. See Appendix Table A3 for full
regression results. (*p < 0.10, **p < 0.05, ***p < 0.01) Shaded cells are statistically significant results for the food

establishment, food insecurity, and food cost variables.
Source: Authors' calculations using merged data. See text for details.
Table 6. Fixed Effects Regression Results of the County-Level Rate of DI Receipt

 

 

Uses Number of Food Uses Number of Food Establishments
Establishments Per 1,000 Residents Per Square Mile
Urban/ Urban/
All "Rural" Micropolitan All "Rural" Micropolitan
(1) (2) (3) (4) (5) (6)

Grocery Size < 50 Employees 0.001*** 0.001***  0.003*** 0.001*  0.020*** 0.000
(0.000) (0.000) (0.001) (0.000) (0.005) (0.000)

Grocery Size 50+ Employees -0.002* = -0.002* -0.001 -0.001 = -0.016*** -0.001
(0.001) (0.001) (0.002) (0.001) (0.003) (0.001)

Grocery Size Missing 0.001 0.002 -0.000 -0.011 0.204 -0.004
(0.004) (0.004) (0.011) (0.016) (0.220) (0.016)

Grocery Other 0.000 0.000 0.000 -0.000 = -0.010*** -0.000
(0.000) (0.000) (0.001) (0.000) (0.004) (0.000)

Convenience Store 0.001**  0.001** 0.000 -0.000 0.007 -0.000
(0.000) (0.000) (0.001) (0.001) (0.004) (0.001)

Warehouse Club -0.005 -0.007 -0.006 -0.006 -0.188 -0.008
(0.008) (0.013) (0.007) (0.006) (0.425) (0.006)

Restaurant Full Service 0.000 0.000 -0.000 -0.000 = -0.004*** -0.000
(0.000) (0.000) (0.000) (0.000) (0.001) (0.000)

Restaurant Limited Service -0.000 -0.000 -0.001 -0.000 0.001 -0.000
(0.000) (0.000) (0.001) (0.000) (0.003) (0.000)

Restaurant Other 0.002*** 0.002***  0.003*** -0.000 0.001 -0.000
(0.000) (0.000) (0.001) (0.000) (0.003) (0.000)

Department Store -0.003 -0.004 -0.007 -0.009 -0.262 -0.007
(0.003) (0.002) (0.009) (0.013) (0.168) (0.014)

Pharmacy 0.001 -0.000 0.003 0.000 -0.016** 0.001
(0.002) (0.003) (0.002) (0.001) (0.007) (0.001)

Gas Station 0.001*** 0.001*** 0.001 0.003* 0.013** 0.002
(0.000) (0.000) (0.001) (0.002) (0.007) (0.002)
% Food Insecure 0.012** 0.003 0.027*** 0.012** 0.004 0.027***
(0.005) (0.007) (0.007) (0.005) (0.007) (0.007)

Cost Per Meal 0.001*** 0.001***  0.000*** 0.001***  0.001*** 0.000**
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

R-Squared 0.266 0.186 0.414 0.260 0.181 0.405
N 31,330 13,290 18,040 31,330 13,290 18,040

 

Notes: All regressions include demographic and socioeconomic controls, year fixed effects, and county fixed
effects. Shaded cells are statistically significant results for the food establishment, food insecurity, and food cost
variables. See Appendix Table A3 for full regression results. (*p < 0.10, **p < 0.05, ***p < 0.01).

Source: Authors' calculations using merged data. See text for details.

32
Table 7. Place Characteristics of Illustrative Counties in 2019

Average Over Los Angeles,

 

All Counties CA Hickman, KY Billings, ND Providence, RI

FIPS code 6037 21105 38007 44007
Population 103,630 10,100,000 4,510 922 635,737
Area in Square Miles 1,118 4,059 242 1,148 409
% Disabled 16.0% 9.9% 24.3% 13.8% 13.8%
% DI Beneficiaries 4.0% 1.5% 6.2% 0.5% 4.2%
Percentage point difference 12.0 8.4 18.1 13.3 9.6
% Food Insecure 13.2% 10.9% 14.3% 5.6% 11.1%
Mean Cost Per Meal $3.18 $3.57 $2.79 $3.70 $3.63
Food Establishments

Total Per 1,000 Residents 2.92 3.26 2.44 13.02 3.15

Total Per Square Mile 0.95 8.11 0.05 0.01 4.89

% Total Likely Healthy 38.1% 46.2% 36.4% 58.3% 44.3%

% Total Likely Unhealthy 38.9% 30.6% 54.5% 25.0% 39.8%

% Total Unknown 8.3% 10.2% 9.1% 16.7% 15.9%

 

Note: "Likely Unhealthy" establishments are defined as grocery stores with fewer than 50 employees; convenience stores;
limited service restaurants; pharmacies; and gas stations.
Source: Authors' calculations using merged data. See text for details.

33
Appendix Table Al. Crosswalk Between NAICS Code and Food Establishment Categories

 

 

Analyzed

NAICS Code _ NAICS Description Study Category
445 Food and Beverage Stores

4451 Grocery Stores

445110 Supermarkets and Other Grocery (except Convenience) Stores Grocery

445120 Convenience Stores Convenience Store
4452 Specialty Food Stores Grocery Other

722 Food Services and Drinking Places

722511 Full-Service Restaurants Restaurant Full
722513 Limited-Service Restaurants Restaurant Limited
722514 Cafeterias, Grill Buffets, and Buffets Restaurant Other
722515 Snack and Nonalcoholic Beverage Bars Restaurant Other
452 General Merchandise Stores

4522 Department Stores

452210 Department Stores Department Store
4523 General Merchandise Stores, including Warehouse Clubs and Supercenters

452311 Warehouse Clubs and Supercenters Warehouse Club
447 Gasoline Stations

4471 Gasoline Stations

447110 Gasoline Stations with Convenience Stores

447190 Other Gasoline Stations Gas Station

446 Health and Personal Care Stores

4461 Health and Personal Care Stores

446110 Pharmacies and Drug Stores Pharmacy

 
Appendix Table A2. Fixed Effects Regression Results of the County-Level Disability Rate

Uses Number of Food Establishments

Uses Number of Food Establishments

 

 

Per 1,000 Residents Per Square Mile
Urban/ Urban/
All "Rural" Micropolitan All "Rural" Micropolitan
(1) (2) (3) (4) 6) (6)
Grocery Size < 50 Employees 0.003* 0.003 0.005 -0.001 -0.021 0.000
(0.002) (0.002) (0.004) (0.001) (0.021) (0.001)
Grocery Size 50+ Employees -0.001 -0.002 -0.000 0.009** -0.047 0.011**
(0.005) (0.007) (0.008) (0.004) (0.040) (0.004)
Grocery Size Missing 0.069 0.025 0.141* 0.181*** 3.230 0.137**
(0.046) (0.041) (0.081) (0.060) (3.437) (0.058)
Grocery Other 0.002 0.003 -0.002 -0.002 0.012 -0.002**
(0.002) (0.002) (0.004) (0.001) (0.009) (0.001)
Convenience Store 0.000 -0.001 0.005** 0.006* 0.035** 0.002
(0.002) (0.002) (0.002) (0.003) (0.017) (0.002)
Warehouse Club -0.082*** = -0.073** -0.086 -0.081** = -2.763*** -0.070*
(0.028) (0.031) (0.055) (0.038) (0.748) (0.039)
Restaurant Full Service 0.000 0.000 0.001 -0.000** 0.003 -0.000***
(0.001) (0.001) (0.002) (0.000) (0.003) (0.000)
Restaurant Limited Service -0.001 -0.004 0.004* 0.002** 0.001 0.002**
(0.002) (0.002) (0.002) (0.001) (0.013) (0.001)
Restaurant Other -0.002 -0.002 -0.002 0.002 -0.009 0.001
(0.002) (0.003) (0.003) (0.003) (0.014) (0.002)
Department Store 0.024***  -0.026*** 0.019 0.013 0.775 0.000
(0.007) (0.008) (0.017) (0.019) (0.569) (0.018)
Pharmacy 0.017 0.003 0.020 -0.004 0.073* -0.004*
(0.010) (0.017) (0.012) (0.003) (0.038) (0.002)
Gas Station 0.002 0.001 0.006* 0.006 0.075*** -0.002
(0.002) (0.002) (0.003) (0.009) (0.028) (0.007)
% Food Insecure 0.961*** = 0.935*** 1.005*** 0.963*** 0.941*** 1.001***
(0.034) (0.047) (0.041) (0.035) (0.048) (0.041)
Cost Per Meal -0.000 -0.002 0.001 -0.000 -0.001 0.001
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
% Ages 18-64 0.101*** = 0.093*** 0,124*** 0,104*** = 0,094*** 0.126***
(0.025) (0.028) (0.048) (0.025) (0.030) (0.048)
% Ages 65+ 0.282*** = 0.304*** 0.224*** 0.285*** = 0.306*** 0.227***
(0.031) (0.039) (0.041) (0.032) (0.041) (0.042)
% HS Diploma -0.066*** = -0.047** -0.096*** -0.067*** = -0.049** -0.097***
(0.015) (0.020) (0.019) (0.015) (0.020) (0.019)
% Some College -0.031* -0.028 -0.031 -0.033** -0.030 -0.032
(0.016) (0.022) (0.021) (0.016) (0.022) (0.020)
% College Graduates -0.082*** = -0.075*** -0.092*** -0.083*** = -0.076*** -0.095***
(0.018) (0.024) (0.024) (0.018) (0.024) (0.024)
% Black 0.041* 0.044 0.025 0.041 0.044 0.026
(0.025) (0.040) (0.023) (0.025) (0.039) (0.023)
% Hispanic 0.053** 0.063* 0.006 0.054** 0.064* 0.015

35
(0.027) (0.035) (0.030) (0.027) (0.035) (0.031)

% Homeowners 0.073***  0.074***  --0,069*** 0.072*** 0.074*** -0,068**#
(0.011) (0.014) (0.013) (0.011) (0.014) (0.014)

% Unemployed -0.492*** - -0.474*#* 0.5264 ** -0.494*** .0.476** = .0,5254**
(0.024) (0.037) (0.027) (0.025) (0.037) (0.027)

Median Income 0.0444**  0.042*** 0,048 0.045***  09.042*** = 0.047##*
(0.004) (0.006) (0.006) (0.005) (0.006) (0.006)

Median Home Value -0.007#** -- -0.007 -0.008**# -0.007** -- -0.006 -0.008***
(0.003) (0.005) (0.003) (0.003) (0.005) (0.003)
Total Population 0.000*** 0.012 0.000** 0.000*** 0.011 0.000*
(0.000) (0.008) (0.000) (0.000) (0.008) (0.000)

Constant -0.172*** - -0,178*** = -0,183##* -0.174*** -.0.180*** --0.176***
(0.030) (0.039) (0.041) (0.031) (0.041) (0.041)
R-Squared 0.284 0.260 0.340 0.282 0.260 0.336
N 31,330 13,290 18,040 31,330 13,290 18,040

 

Notes: All regressions include year fixed effects and county fixed effects. Shaded cells are statistically significant results

for the food establishment, food insecurity, and food cost variables. (*p < 0.10, **p < 0.05, ***p < 0.01).

Source: Authors' calculations using merged data. See text for details.

36
Appendix Table A3. Fixed Effects Regression Results of the County-Level Rate of DI Receipt

Uses Number of Food Establishments

Uses Number of Food

 

 

Per 1,000 Residents Establishments Per Square Mile
Urban/ Urban/
Micropolita Micropolit
All "Rural" n All "Rural" an
(1) (2) (3) (4) 6) (6)
Grocery Size< 50 Employees 0.001*** --0.001*** 0.003*** 0.001* 0.020*** 0.000
(0.000) (0.000) (0.001) (0.000) (0.005) (0.000)
Grocery Size 50+ Employees -0.002* -0.002* -0.001 -0.001 -0.016*** -0.001
(0.001) (0.001) (0.002) (0.001) (0.003) (0.001)
Grocery Size Missing 0.001 0.002 -0.000 -0.011 0.204 -0.004
(0.004) (0.004) (0.011) (0.016) (0.220) (0.016)
Grocery Other 0.000 0.000 0.000 -0.000 -0.010*** -0.000
(0.000) (0.000) (0.001) (0.000) (0.004) (0.000)
Convenience Store 0.001** 0.001** 0.000 -0.000 0.007 -0.000
(0.000) (0.000) (0.001) (0.001) (0.004) (0.001)
Warehouse Club -0.005 -0.007 -0.006 -0.006 -0.188 -0.008
(0.008) (0.013) (0.007) (0.006) (0.425) (0.006)
Restaurant Full Service 0.000 0.000 -0.000 -0.000 -0.004*** -0.000
(0.000) (0.000) (0.000) (0.000) (0.001) (0.000)
Restaurant Limited Service -0.000 -0.000 -0.001 -0.000 0.001 -0.000
(0.000) (0.000) (0.001) (0.000) (0.003) (0.000)
Restaurant Other 0.002***  -0.002*** 0.003*** -0.000 0.001 -0.000
(0.000) (0.000) (0.001) (0.000) (0.003) (0.000)
Department Store -0.003 -0.004 -0.007 -0.009 -0.262 -0.007
(0.003) (0.002) (0.009) (0.013) (0.168) (0.014)
Pharmacy 0.001 -0.000 0.003 0.000 -0.016** 0.001
(0.002) (0.003) (0.002) (0.001) (0.007) (0.001)
Gas Station 0.001***  -0.001*** 0.001 0.003* 0.013** 0.002
(0.000) (0.000) (0.001) (0.002) (0.007) (0.002)
% Food Insecure 0.012** 0.003 0.027*** 0.012** 0.004 0.027***
(0.005) (0.007) (0.007) (0.005) (0.007) (0.007)
Cost Per Meal 0.001***  -0.001*** 0.000*** 0.001***  -0.001*** 0.000**
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
% Ages 18-64 0.017** 0.012 0.036*** 0.018** 0.014* 0.036***
(0.007) (0.008) (0.011) (0.007) (0.008) (0.012)
% Ages 65+ -0.005 -0.011 0.021** -0.004 -0.009 0.019**
(0.009) (0.012) (0.008) (0.009) (0.012) (0.009)
% HS Diploma -0.008** -0.008* -0.009** -0.008** -0.008* -0.011***
(0.004) (0.004) (0.004) (0.004) (0.004) (0.004)
% Some College -0.010*** = -0.015*** -0.005 -0.011*** = -0.014*** -0.007*
(0.004) (0.005) (0.004) (0.004) (0.005) (0.004)
% College Graduates -0.003 -0.003 -0.004 -0.002 -0.002 -0.005
(0.003) (0.004) (0.004) (0.004) (0.005) (0.004)
% Black 0.000 -0.010 0.012* 0.001 -0.009 0.013*
(0.005) (0.006) (0.007) (0.005) (0.006) (0.007)

37
% Hispanic -0.027** ---0,028 -0.018* -0.025* -0.026 -0.018
(0.013) (0.017) (0.010) (0.013) (0.017) (0.012)
% Homeowners -0.002 -0.002 -0.000 -0.001 -0.002 -0.000
(0.002) (0.003) (0.003) (0.002) (0.003) (0.003)
% Unemployed -0.002 0.001 -0.009* -0.002 0.001 -0.008
(0.005) (0.007) (0.005) (0.005) (0.007) (0.005)
Median Income -0.000 0.001 -0.003** -0.000 0.001 -0.003**
(0.001) (0.002) (0.001) (0.001) (0.002) (0.001)
Median Home Value -0.006*** -0,005*** -_-0.006**# -0.006*** - -0.005*** - -0,006*#*
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Total Population -0.000*** ---0,004 -0.000*** -0,.000*** 0,004 -0.000**
(0.000) (0.002) (0.000) (0.000) (0.002) (0.000)
Constant 0.045*** 0,053"** --0,032*** 0.045*** 9,052"  - 0,033#*
(0.005) (0.007) (0.010) (0.005) (0.007) (0.011)
R-Squared 0.266 0.186 0.414 0.260 0.181 0.405
N 31,330 13,290 18,040 31,330 13,290 18,040

 

Notes: All regressions include year fixed effects and county fixed effects. Shaded cells are statistically significant
results for the food establishment, food insecurity, and food cost variables. (*p < 0.10, **p < 0.05, ***p < 0.01).
Source: Authors' calculations using merged data. See text for details.

38
RECENT WORKING PAPERS FROM THE
CENTER FOR RETIREMENT RESEARCH AT BOSTON COLLEGE

How to Increase Usage of SSA's Online Tools
Jean-Pierre Aubry and Kevin Wandrei, November 2021

Work-Related Overpayment and Benefit Suspension Experiences of Federal Disability
Beneficiaries
Marisa Shenk and Gina Livermore, November 2021

How Has COVID-19 Affected the Labor Force Participation of Older Workers?
Laura D. Quinby, Matthew S. Rutledge, and Gal Wettstein, October 2021

The Impact of Claimant Representation Fee Schedules on the Disability Applicant Process
and Recipient Outcomes
Cody Tuttle and Riley Wilson, September 2021

Does Media Coverage of the Social Security Trust Fund Affect Claiming, Saving, and
Benefit Expectations?
Laura D. Quinby and Gal Wettstein, September 2021

Does Social Security Serve as an Economic Stabilizer?
Laura D. Quinby, Robert Siliciano, and Gal Wettstein, July 2021

Are Older Workers Capable of Working Longer?
Laura D. Quinby and Gal Wettstein, June 2021

Do Stronger Employment Discrimination Protections Decrease Reliance on Social Security
Disability Insurance? Evidence from The U.S. Social Security Reforms
Patrick Button, Mashfiqur R. Khan, and Mary Penn, June 2021

Trends in Opioid Use among Social Security Disability Insurance Applicants
April Yanyuan Wu, Denise Hoffman, and Paul O'Leary, March 2021

The Value of Annuities
Gal Wettstein, Alicia H. Munnell, Wenliang Hou, and Nilufer Gok, March 2021

Will Women Catch Up to Their Fertility Expectations?
Angi Chen and Nilufer Gok, February 2021

The Consequences of Current Benefit Adjustments for Early and Delayed Claiming
Andrew G. Biggs, Angi Chen, and Alicia H. Munnell, January 2021

All working papers are available on the Center for Retirement Research website
(hitps://crr.bc.edu) and can be requested by e-mail (crr@bc.edu) or phone (617-552-1762).