USDA 09 - Access To Affordable & Nutritious Food
USDA 09 - Access To Affordable & Nutritious Food
USDA 09 - Access To Affordable & Nutritious Food
Department
of Agriculture
Report to Congress
This report was prepared by the Economic Research Service (ERS), the Food
and Nutrition Service (FNS), and the Cooperative State Research, Education,
and Extension Service (CSREES) of the U.S. Department of Agriculture under
the direction of Michele Ver Ploeg of ERS. Contributors include Vince Breneman,
Tracey Farrigan, Karen Hamrick, David Hopkins, Phil Kaufman, Biing-Hwan Lin,
Mark Nord, Travis Smith, and Ryan Williams of ERS; Kelly Kinnison, Carol Olander,
and Anita Singh of FNS; Elizabeth Tuckermanty of CSREES; Rachel Krantz-Kent
and Curtis Polen of the Bureau of Labor Statistics; and Howard McGowan and
Stella Kim of the U.S. Census Bureau.
Report to Congress
United States
Access to Affordable and
Department
of Agriculture Nutritious Food
Measuring and Understanding Food
Deserts and Their Consequences
June 2009
Abstract
The Food, Conservation, and Energy Act of 2008 directed the U.S. Department of
Agriculture to conduct a 1-year study to assess the extent of areas with limited access
to affordable and nutritious food, identify characteristics and causes of such areas,
consider how limited access affects local populations, and outline recommendations
to address the problem. This report presents the findings of the study, which include
results from two conferences of national and international authorities on food deserts and
a set of commissioned research studies done in cooperation with the National Poverty
Center at the University of Michigan. It also includes reviews of existing literature, a
national-level assessment of access to supermarkets and large grocery stores, analysis
of the economic and public health effects of limited access, and a discussion of existing
policy interventions. The study uses a variety of analytical methods and data to assess
the extent of limited access to affordable and nutritious food and characteristics of areas
with limited access. Overall, findings show that a small percentage of consumers are
constrained in their ability to access affordable nutritious food because they live far
from a supermarket or large grocery store and do not have easy access to transportation.
Urban core areas with limited food access are characterized by higher levels of racial
segregation and greater income inequality. In small-town and rural areas with limited
food access, the lack of transportation infrastructure is the most defining characteristic.
Existing data and research are insufficient to conclusively determine whether areas with
limited access have inadequate access.
Acknowledgments
This report was prepared by the Economic Research Service (ERS), the Food and
Nutrition Service (FNS), and the Cooperative State Research, Education, and Extension
Service (CSREES) of the U.S. Department of Agriculture under the direction of Michele
Ver Ploeg of ERS. Contributors include Vince Breneman, Tracey Farrigan, Karen
Hamrick, David Hopkins, Phil Kaufman, Biing-Hwan Lin, Mark Nord, Travis Smith,
and Ryan Williams of ERS; Kelly Kinnison, Carol Olander, and Anita Singh of FNS;
Elizabeth Tuckermanty of CSREES; Rachel Krantz-Kent and Curtis Polen of the Bureau
of Labor Statistics; and Howard McGowan and Stella Kim of the U.S. Census Bureau.
Contents
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
ii
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Summary
Increases in obesity and diet-related diseases are major public health
problems. These problems may be worse in some U.S. communities because
access to affordable and nutritious foods is difficult. Previous studies suggest
that some areas and households have easier access to fast food restaurants
and convenience stores but limited access to supermarkets. Limited access
to nutritious food and relatively easier access to less nutritious food may
be linked to poor diets and, ultimately, to obesity and diet-related diseases.
Congress, in the Food, Conservation, and Energy Act of 2008, directed the
U.S. Department of Agriculture (USDA) to conduct a 1-year study to assess
the extent of the problem of limited access, identify characteristics and
causes, consider the effects of limited access on local populations, and outline
recommendations to address the problem.
This report presents the findings of the study, which include results from two
conferences of national and international authorities on food deserts and a
set of commissioned research studies done in cooperation with the National
Poverty Center at the University of Michigan. It also includes reviews of
existing literature, a national-level assessment of access to supermarkets and
large grocery stores, analysis of the economic and public health effects of
limited access, and a discussion of existing policy interventions. A variety of
analytical methods and data are used to assess the extent of limited access to
affordable and nutritious food and characteristics of areas with limited access.
Findings
• Of all households in the United States, 2.3 million, or 2.2 percent, live
more than a mile from a supermarket and do not have access to a vehicle.
An additional 3.4 million households, or 3.2 percent of all households,
live between one-half to 1 mile and do not have access to a vehicle.
• Data on time use and travel mode show that people living in low-income
areas with limited access spend significantly more time (19.5 minutes)
traveling to a grocery store than the national average (15 minutes).
However, 93 percent of those who live in low-income areas with
iii
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
limited access traveled to the grocery store in a vehicle they or another
household member drove.
Urban core areas with limited food access are characterized by higher levels
of racial segregation and greater income inequality. In small-town and rural
areas with limited food access, the lack of transportation infrastructure is the
most defining characteristic.
Supermarkets and large grocery stores have lower prices than smaller
stores. A key concern for people who live in areas with limited access is
that they rely on small grocery or convenience stores that may not carry
all the foods needed for a healthy diet and that may offer these foods and
other food at higher prices. This report examines whether prices of similar
foods vary across retail outlet types and whether the prices actually paid
by consumers vary across income levels. These analyses use proprietary
household-level data that contain information on food items purchased by
approximately 40,000 demographically representative households across the
United States. Results from these analyses show that when consumers shop
at convenience stores, prices paid for similar goods are, on average, higher
than at supermarkets.
Low-income households shop where food prices are lower, when they
can. Findings also show that food purchases at convenience stores make up
a small portion of total food expenditures (2 to 3 percent) for low-income
consumers. Low- and middle-income households are more likely to purchase
food at supercenters, where prices are lower. Administrative data on SNAP
benefit redemptions from 2008 show that 86 percent of SNAP benefits were
redeemed at supermarkets or large grocery stores. Research that considers
the prices paid for the same food across household income levels indicates
that while some of the very poorest households—those earning less than
$8,000 per year—may pay between 0.5 percent and 1.3 percent more for
their groceries than households earning slightly more, households earning
between $8,000 and $30,000 tend to pay the lowest prices for groceries,
whereas higher income households pay significantly higher prices.
The study also examined food shopping behavior and the types of food
purchased for SNAP participants and other low-income households. Data
from the 1996/1997 NFSPS show that SNAP participants were, on average,
1.8 miles from the nearest supermarket. However, the average number of
miles both SNAP participants and eligible nonparticipants traveled to the
store most often used was 4.9 miles. These same data show that SNAP
iv
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
participants who did not shop at supermarkets purchased less noncanned
fruit, noncanned vegetables, and milk than SNAP participants who shopped
frequently at a supermarket.
Easy access to all food, rather than lack of access to specific healthy foods,
may be a more important factor in explaining increases in obesity. Many
studies find a correlation between limited food access and lower intake of
nutritious foods. Data and methods used in these studies, however, are not
sufficiently robust to establish a causal link between access and nutritional
outcomes. That is, other explanations cannot be eliminated as the cause of
lower intake. A few studies have examined food intake before and after
healthy food options become available (either within existing stores or
because new stores opened). The findings are mixed—some show a small
but positive increase in consumption of fruits and vegetables, while others
show no effect.
Food has been used as a tool for community development. Projects such
as farmers’ markets, community gardens, promotion of culturally specific
foods for ethnic minorities and Native Americans, local food production
and promotion, youth agricultural and culinary training programs, and many
other types of programs have all been implemented in a variety of settings,
both urban and rural. USDA’s Community Food Projects Competitive Grant
program has much experience in funding and nurturing such programs.
v
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The current state of research is insufficient to conclusively determine
whether some areas with limited access have inadequate access. Future
research should consider improved methods to measure access levels,
availability, and prices of foods faced by individuals and areas. More
research is needed to understand how access, availability and price affect
the shopping and consumption behaviors of consumers. Data linking
information on the types of foods consumers purchase and eat with measures
of consumers’ levels of access and the prices they face could help explain the
economic consequences of food access. Studies that use improved methods
and data to determine how food access affects diet, obesity, and other health
outcomes are also needed to help explain the health consequences of food
access.
Methods
vi
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
consequences of food deserts, and programs and policies to mitigate the
adverse effects of food deserts.
vii
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 1
Introduction
Increases in rates of obesity and related chronic diseases that may be linked
with poor diets, such as diabetes and heart disease, are major public health
concerns. Some advocates, community leaders, and researchers are worried
that these problems, and poor diets in general, may be more severe in certain
poor and rural American communities because these areas have limited
access to affordable and nutritious foods. A primary concern is that some
poor or rural areas do not have access to supermarkets, grocery stores, or
other food retailers that offer the large variety of foods needed for a healthy
diet (for example, fresh fruits and vegetables, whole grains, fresh dairy and
meat products). Instead, individuals in these areas may be more reliant on
food retailers or fast food restaurants that only offer more limited varieties
of foods. It is hypothesized that the relative lack of access to full-service
grocery stores and the easier access to fast and convenience foods may be
linked to poor diets and, ultimately, to obesity and other diet-related diseases.
It was this concern that led Congress, in the Food, Conservation, and Energy
Act of 2008, (hereafter referred to as the 2008 Farm Bill) to direct the U.S.
Department of Agriculture (USDA) to conduct a 1-year study of areas
with limited access to affordable and nutritious food. The 2008 Farm Bill
directed USDA to assess the extent of the problem of limited access, identify
characteristics and causes of limited access and the effects limited access
has on local populations, and outline recommendations for addressing the
causes and effects of limited access. The USDA study was conducted by
a team of researchers, policy analysts, and program leaders from USDA’s
Food and Nutrition Service (FNS), Cooperative State Research, Education,
and Extension Service (CSREES), and the Economic Research Service 1One additional activity that is not
(ERS), which served as the lead agency. This report provides the analysis further described in this report is the
and findings of the USDA study. A number of information-gathering and Workshop on Access to Affordable and
Nutritious Foods: Understanding Food
data-analysis activities were conducted as part of the study. Each of these
Deserts held on October 9, 2008, in
activities and their purposes is described later in this introduction.1 Washington, DC. An agenda for this
workshop is included in appendix A.
Definitions, Concepts, and Background Literature
2We note that this study focuses
TThe language in the 2008 Farm Bill defined a food desert as an “area in on the ease at which households and
the United States with limited access to affordable and nutritious food, individuals can get to stores that sell the
particularly such an area composed of predominantly lower income foods they want at affordable prices.
We do not focus on the related concept
neighborhoods and communities” (Title VI, Sec. 7527). In order to consider of food security, which measures
the extent of such areas, the following questions first need to be answered: whether households or individuals have
access to enough food for an active,
• What is affordable food and nutritious food? healthy life. The concepts are clearly
related, but, in general, food security
• What does it mean to have (or not have) access to such food? measures focus less on physical access
and more on whether a household
• Do individuals or do areas lack access? can afford food. For example, some
individuals or households may have
The concern over food deserts is that some consumers have difficulty low food security but may live only
one block from a large supermarket, so
accessing food retailers that offer affordable and nutritious food.2 The
that physical access to food is less of
ease or difficulty in getting to a food retailer depends on the location of the a problem than whether the family can
store in relationship to the consumer and the consumer’s travel patterns, the afford to buy the food.
1
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
consumer’s individual characteristics (e.g., income, car ownership, disability
status), and neighborhood characteristics (e.g., the availability of public
transportation, availability of sidewalks, and crime patterns in the area).
Before considering how many people and places may be affected by limited
access to affordable and nutritious food, one must first determine what is
meant by “nutritious food” and “affordable,” and how access to affordable
and nutritious foods can be measured.
Affordability of food refers to the price of a particular food and the relative
price of alternative or substitute foods. Affordability of food is also impacted
by the budget constraints faced by consumers, who must consider not only
the prices of different foods to meet their food needs, but also the prices
of other necessities (e.g., housing, clothing, and transportation). USDA
provides guidance on national standards for nutritious diets at various
costs levels—the Thrifty, Low-cost, Moderate-cost and Liberal Food Plans
(Carlson et al., 2007a; Carlson et al., 2007b). Within each plan is a market
basket of foods in quantities that reflect current dietary recommendations,
food composition data, food prices, and actual consumption patterns.
According to the Low-cost Food Plan, a family of four with two adults
(age 19 to 50) and two children (ages 6 to 8 and 9 to 11) could consume a
nutritious diet for $175.60 per week (USDA, 2009).
Studies of food access have also measured the availability and prices of
foods in USDA’s Thrifty Food Plan (TFP) in stores as a standardized way
to compare availability and affordability of foods in geographic areas (for
example, Block and Kouba, 2005; Hendrickson et al., 2006; Mantovani et al.,
1997; Rose et al., 2009). Such uses of the TFP provide an absolute measure
of availability and price (as opposed to a relative measure) and allow
aggregation across store types (Bitler and Haider, 2009).
Once the availability and price of food has been measured, studies of food
access typically then measure how easy it is for consumers to access the
4Appendix
food. The ease or difficulty of food access has been measured many ways.4 table B.1 summarizes the
measures of access to health foods used
One common method is to measure distance from consumers’ residences
by over 30 studies. Figure B.1 shows
to the nearest food retailer that offers healthy and affordable foods (often to the locations of these studies.
supermarkets or large grocery stores).5 Distances in sparsely populated areas
are often not directly comparable to distance in densely populated areas. As 5Distance is usually measured from
a result, many studies consider access in rural areas separately from access the centroid of an area (e.g., ZIP Code,
in suburban and urban areas. “Walkable” distance measures have often been census tract, or block) to the nearest
used to characterize access in urban areas. The definition of such a distance supermarket.
is often 1 kilometer or about a half mile (app. table B.1). Similar concepts
for less densely populated suburban and rural areas have not been applied,
but often a distance is designated to distinguish access limitations. For
example, areas more than 10 miles from a supermarket have been called food
deserts (Blanchard and Lyson, 2006; Morton and Blanchard, 2007). These
designations of what may be considered an “acceptable” distance to a food
source in less densely populated areas are somewhat arbitrary, especially
considering that without a car, any distance of more than a mile or so could
be considered unacceptably far.
3
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Distance is almost always measured as distance from a residential area
to a store, assuming home to store travel is the way most people access
supermarkets. But people do not just travel from home to store. They travel
to work, school, church, and beyond and often purchase food on the way.
Using an access measure that only considers distance from home is likely to
underestimate the options available for food shopping.
A problem with both distance and density measures, however, is that they
6This helpful distinction is made by
only measure “potential access,” and not “realized access.”6 Potential access
Sharkey and Horel, 2009.
shows where consumers could possibly shop, while realized access shows
where consumers actually shop. A consumer that does not care to eat at
fast food restaurants or convenience stores may have high access to these
stores but may pass by them on the way to a supermarket that is farther
away. And even if the concentration of convenience stores is higher in some
neighborhoods, most of the food shopping could be conducted at larger
supermarkets. For example, Broda et al. (forthcoming) find that compared
with higher income families, low-income families spend slightly more of
their food budget at convenience stores, which offer prices that are, on
average, greater than those in traditional grocery stores. However, the study
7Most studies consider only areas
also found that compared with higher income families, low-income families
with high concentrations of poor peo-
spend a greater share of total expenditures at supercenters, where lower ple. Some also consider areas with low
prices almost completely offset the higher prices at convenience stores. To vehicle ownership rates, high concen-
further illustrate this point, data show that, on average, SNAP participants trations of elderly, and the availability
lived 1.8 miles from the nearest supermarket but traveled 4.9 miles to the of public transportation. See Necker-
foodstore they most often used (Cole, 1997). (More details on both of these man et al., 2009, for more details.
8A separate concern is for those who
findings are provided in chapter 5).
are too poor to buy food regardless of
how accessible it is. USDA’s House-
Area-based versus individual based concepts of access hold Food Security in the United States
series reports the percent of Americans
Studies that use area-based measures of access, either distance or density, who do not have access to enough food
usually focus only on areas with high concentrations of vulnerable for an active, healthy life for all house-
populations.7 Examining only areas in which a relatively high proportion of hold members. In 2007, 11.1 percent
poor people live, for example, will miss many poor people who live in less of households were food insecure at
some time during the year and about
poor areas, but who may also have limited access.8 Further, not everyone 4.1 percent of all households had very
who lives in low-income areas is poor. There will certainly be people who low food security some time during the
live in a low-income area with limited access but who themselves have year (Nord et al., 2008).
4
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
adequate resources to travel to a supermarket regularly. Ownership of,
or easy access to, a motorized vehicle may be the best marker of access
regardless of whether someone lives in a poor area or not. The majority of
U.S. households own cars (89.7 percent).9 Vehicle ownership rates among 9See chapter 3 for details on house-
those living in rural areas (94.6 percent) are higher than among those living hold access to vehicles.
in urban areas (87.8 percent). Those with low incomes are less likely to own
a vehicle, but time use and travel mode data reported in Chapter 2 show that
most people, even low-income people, take their own vehicles or drive with
someone else to do their grocery shopping.
The bulk of studies of food access find relative differences across areas
10 Larson et al. (2009) (Not in refer-
in access to some types of food retailers and foods.10 Researchers have
ences) reviewed 54 studies of dispari-
documented the inequality of access to supermarkets in urban inner city ties in access to healthy food.
areas (Donohue, 1997), while others have focused on differences in access
to supermarkets in poor versus nonpoor areas (O’Conner and Abell, 1992;
Cotterill and Franklin, 1995; Pike, 2000). Moore and Diez Roux (2006)
investigated racial disparities in the number and variety of grocery stores
in neighborhoods. Zenk et al. (2005) compared distances to the nearest
supermarket among poor White and poor non-White households. Extending
that approach, Gallagher (2007 and 2006) compared differences in the
ratio of supermarkets to other foodstores in a neighborhood in Detroit and
Chicago. More recently, Neckerman et al. (2009) examined the retail food
environment in New York City. They considered the characteristics of
households, such as race, income, and forms of available transportation,
including vehicle ownership or access to mass transit, as factors affecting a
household’s foodstore access. Findings show lower access to supermarkets
and other healthy food stores for neighborhoods composed primarily of
African-Americans, where populations were heavily reliant on mass transit
for transportation.
Report Outline
A variety of data and methods was used to assess the extent of limited access
to affordable and nutritious food, including both individual measures of
access and area-based measures of access. Responses to a national-level
household survey of food adequacy and access were analyzed, as were
data estimating differences among households in the time spent traveling
to grocery stores and the travel mode used. To determine the extent of
areas with limited access, a comprehensive database was developed to
identify the location of supermarkets and large grocery stores within the
continental United States. Food access was estimated as the distance to
the nearest supermarket or large grocery store, which is used as a proxy for
the availability of affordable and nutritious food. The analysis specifically
considered distance to the nearest supermarket for low-income populations
and for households without access to a vehicle. Differences in rural, urban,
and areas in between were considered. These results are presented in
Chapter 2. Chapter 3 uses the same data on the location of supermarkets to
analyze the household and neighborhood characteristics that distinguish areas
with low access from areas with better access. The novel contribution of
these analyses is that they are national in scope and combine two databases
on supermarket and grocery store location.
Supermarkets are not the only sources of healthy and affordable foods.
Many smaller scale sources may be used by those who are underserved by
supermarkets. However, a complete assessment of the food environment of
every area in the United States is an enormous task that is beyond the scope
of this study. Instead, USDA cooperated with the National Poverty Center
(NPC) at the University of Michigan to commission six studies of the food
environment at more localized levels. These studies provide more detail on
the food environment in New York City; Indianapolis, Indiana; New Orleans,
Louisiana; Salt Lake County, Utah; the Brazos Valley in rural Texas; and
Portland, Oregon. Methods and findings from these studies, along with the
national level analyses, are discussed in Chapters 2 and 3.11 11Drafts of these papers and an
Populations that live in areas with limited access to affordable and nutritious
food may adjust their food shopping behaviors and diets based on the food
environment in their area. These adjustments could be due to the lack of
availability of some foods or to the relative prices of different foods offered
from different food retailers. Chapter 5 considers how food access relates
to food choice—that is, whether consumers in areas with limited access
face higher prices for similar goods and whether they have different food
purchasing behaviors. Comparisons of the prices that consumers paid for
similar foods (milk, ready-to-eat cereal, and bread) purchased at different
retail outlets (supermarkets and grocery stores vs. convenience stores)
are made using hedonic price models. Differences in the prices offered
at different retail outlets could lead consumers to adjust where they shop
and what they purchase. The chapter also considers shopping behavior for
populations with limited access, which can further the understanding of the
adjustments that consumers make to different prices and retail availability.
A summary of a body of work conducted by FNS on the shopping patterns
of participants of the Supplemental Nutrition Assistance Program (SNAP—
formerly called the Food Stamp Program) is provided in the chapter. ERS
also analyzed how SNAP participants’ expenditures on foods in several food
groups (e.g., canned and noncanned fruits and vegetables) varied by self-
reported measures of access to supermarkets. Spending on these food groups
by people with relatively easy access to supermarkets is compared with
spending by those with less access to supermarkets. Finally, findings from
a study conducted by ERS and external researchers on whether poor people
pay more for similar foods relative to higher income people are integrated.
7
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
localities, as well as describes several other policy options that may be
considered to reduce the effects of limited access.
Not all of the questions about the extent, causes, and consequences of food
deserts will be answered in this report. The final chapter, Chapter 9, outlines
an agenda for further research on the causes and consequences of areas with
limited access to affordable and nutritious food.
References
Apparicio, P., M.S. Cloutier, and R. Shearmur (2007). “The Case of
Montréal’s Missing Food Deserts: Evaluation of Accessibility to Food
Supermarkets,” International Journal of Health Geographics 6(4).
Bitler, Marianne, and Steven J. Haider (2009). “An Economic View of Food
Deserts in the United States,” National Poverty Center Working Paper, www.
npc.umich.edu/news/events/food-access/index.php
Blanchard, T.C., and T. Lyson (2006). Food Availability and Food Deserts
in the Nonmetropolitan South, Assistance Needs of the South’s Vulnerable
Populations, Number 12.
Carlson, A., M. Lino, W.Y. Juan, K. Hanson, and P.P. Basiotis (2007a).
Thrifty Food Plan, 2006, CNPP-19, U.S. Department of Agriculture, Center
for Nutrition Policy and Promotion.
Cotterill, R., and A. Franklin (1995). “The Urban Grocery Store Gap,” Food
Marketing Policy Center, University of Connecticut.
8
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Gallagher, M. (2006). Examining the Impact of Food Deserts on Public
Health in Chicago, Mari Gallagher Research & Consulting Group.
Morton, L.W., and T.C. Blanchard (2007). “Starved for Access: Life in
Rural America’s Food Deserts,” Rural Realities 1(4): 1-10.
Pike, J. (2000). “Boston’s Poor Pay More For Food, Theft and Other
Problems Increase Business Costs, Prices,” Massachusetts News, August 2.
Rose, D., J.N. Bodor, C.M. Swalm, J.C. Rice, T.A. Farley, and P.L.
Hutchinson (2009). “Deserts in New Orleans? Illustrations of Urban Food
Access and Implications for Policy,” National Poverty Center Working
Paper, http://www.npc.umich.edu/news/events/food-access/index.php
9
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Sharkey, J.R., and S. Horel (2009). “Characteristics of Potential Spatial
Access to a Variety of Fruits and Vegetables in a Large Rural Area,”
National Poverty Center Working Paper, www.npc.umich.edu/news/events/
food-access/index.php
Zenk S, N., Schulz A, J., Israel B, A., James S, A., et al. (2005).
“Neighborhood Racial Composition, Neighborhood Poverty, and the Spatial
Accessibility of Supermarkets in Metropolitan Detroit,” American Journal of
Public Health, 95(4):660.
10
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 2
11
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
In addition to asking food security questions, the CPS-FSS, until 2001, asked
a general question about whether the household had enough of the kinds of
foods it wanted and needed. Those households who responded that they
did not have enough of the kinds of foods they wanted were asked followup
questions about why they did not have enough food. Respondents could
answer by choosing from among several options, including options directly
related to store access (see box, “CPS-FSS Questions on Food Access”).
Table 2.1 provides the population weighted responses to these questions
and provides a direct measure of the percent of households that do not
always have enough of the foods they want because of access limitations.
Eighty-one percent of households always had the kinds of foods they wanted
to eat. Sixteen percent always had enough food to eat but did not always
SS1 Which of these statements best describes the food eaten in your household-
-enough of the kinds of food we want to eat, enough but not always the kinds
of food we want to eat, sometimes not enough to eat, or often not enough to
eat?
Those who gave response #2, “enough but not always the kinds of foods we want to
eat” were asked SS1B:
SS1B Here are some reasons why people don’t always have the kinds of food they
want. For each one, please tell me if that is a reason why YOU don’t always
have the kinds of food you want to eat.
Those who responded to question SS1 with response #3 or #4, “sometimes” or “often
not enough to eat” were asked SS1C:
SS1C Here are some reasons why people don’t always have enough to eat. For
each one, please tell me if that is a reason why YOU might not always have
enough to eat.
Notes: Question SS1 is still asked in the CPS-FSS. Up until 2001, questions SS1B
and SS1C were also asked as follow up questions. For questions SS1B and SS1C,
multiple responses were accepted.
12
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.1
National estimates of the percent of households who do not
have enough of the kinds of foods they want because of food
access limitations
Percent of
all households
Households that always had the kinds of foods
they wanted to eat 81.0
Households that had enough to eat but did not
always have the kinds of foods they wanted to eat 16.0
Households that sometimes or often did not
have enough to eat 3.1
Reported reasons for not always having the kinds of foods or enough food:
Households that always had enough to eat, but did not always have
the kinds of foods they wanted to eat because it was too hard to get
to the store or the kinds of foods they wanted were not available 5.1
Households that sometimes or often did not have enough
to eat because it was too hard to get to the store 0.6
Total with food access limitations 5.7
Notes: Some of those who reported it was “too hard to get to the store” may be elderly or
disabled. Of the 5.7 percent who reported access limitations, more than half (3.0 percent of all
households) also cited that they did not have enough money for food.
Source: USDA, ERS calculations based on 2001 CPS-FSS survey data.
have the kinds of foods they wanted to eat. Another 3 percent sometimes or
often did not have enough food to eat.
Respondents who reported they had enough to eat but did not always have
the kinds of foods they wanted were asked why they did not have the kinds
of foods they wanted. Among all households, 5.1 percent reported that they
did not have the kinds of foods they wanted because it was either too hard to
get to the store or the foods they wanted were not available. Respondents
who reported that they sometimes or often did not have enough food to eat
were also asked why. A total of 0.6 percent of all households said they did
not always have enough to eat and that it was because it was too hard to get
to the store. Thus, based on these questions, 5.7 percent of all households
reported they did not always have the food they want or need because of
access limitations.
Not all of these access limitations reflect a lack of a nearby store with
adequate food. Some who report that it is too hard to get to the store may
be disabled or elderly and frail. This group may very well have food access
problems, but it is not necessarily because they do not have nutritious food
options nearby. Further, of the 5.7 percent who cited access problems, more
than half, or 3.0 percent, also cited that they did not have enough money
for food. Another reason these responses may not indicate access problems
related to nearby availability of food is that the condition “enough but not
always the kinds of foods we want to eat” does not necessarily indicate
whether the food available was nutritionally adequate or not. Despite
these caveats, these estimates from the CPS-FSS could be considered an
estimate of the number of households that face food access limitations.
13
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area-Based Measures of Access to Affordable and
Nutritious Food
Individual measures of access provide one estimate of the number of people
affected by limited access. The primary intent of the congressional mandate
was to focus on area-based measures of access. Area-based measures are
important because characteristics of the areas where people live, work, or
travel may affect access to healthy and affordable food, which may affect
diet and health.
This section examines the extent of areas in the U.S. that have low access to
supermarkets, a reliable source of nutritious and affordable foods. A national
supermarket directory is first developed and geocoded. Data on population,
income, and other household characteristics from the 2000 Census are
aggregated to square kilometer grids that cover the continental United States.
The distances are measured from the center of these 1-kilometer grids to the
nearest supermarket for the entire U.S. population, for low-income areas and
higher income areas, and by characteristics of individuals or households.
Access to supermarkets is described using these distance measures first
for the entire U.S. population, then separately for Census Urbanized Area
designations. The descriptions use simple population-weighted median
distance to stores overall and across Urbanized Area and population
characteristics. Each area is assigned to one of three categories of access
based on whether the distance to the nearest supermarket is within a range
of “walkable” distances. For rural areas, a “drivable” distance measure is
considered.
An ERS review of studies of food prices found that supermarket prices are
10 percent lower, on average, than those of smaller foodstores, in part, due
to lower per unit costs resulting in lower margins over cost of goods sold
(Kaufman et al., 1997). Neckerman, et al. (2009) cite a number of audit
surveys of food prices, finding that store type is highly associated with
price and that supermarkets, larger chain stores, or discount stores such as
supercenters, tend to offer lower prices.
14
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Montovani et al. (1997) examined characteristics and services of a nationally
representative sample of 2,400 stores authorized to receive benefits from the
Supplemental Nutrition Assistance Program (SNAP). Price, quality, and
variety of store foods were assessed in terms of the market basket of goods
13Market basket quality was
that reflect the Thrifty and Low Cost Food Plans.13 This analysis focused on
measured in terms of availability of
product availability and cost in areas with different concentrations of poverty. acceptable items as guided by a USDA
In urban areas, market basket costs in supermarkets and large grocers were publication on buying quality food
nearly equivalent across levels of poverty. Prices were less at “other” (1975).
stores located in high-poverty areas versus those in lower poverty areas.
Availability of market basket items did not vary by poverty level among
supermarkets in urban areas. Variety did vary by poverty level for large
grocers. Fresh produce and fresh seafood were less available in large grocers
located in high-poverty areas. Fresh meat was more available, however,
at large grocers in these locations. In rural areas, market basket costs were
consistently similar in higher and lower poverty areas. With the exception
of fresh seafood, a similar proportion of market basket items was available in
supermarkets and large grocery stores, regardless of the area’s poverty level.
Food quality was similar across different store types and poverty levels in
rural areas. Results from this analysis confirm that, on average, supermarkets
and large grocery stores offer lower prices and more variety than other store
types. Large grocers were more similar to supermarkets than other store
types, especially in rural zip codes.
The analysis uses supermarkets and large grocery stores (hereafter defined
simply as “supermarkets”) as proxies for food retailers that offer a variety
of nutritious, affordable retail foods. The industry-standardized definition
requires that to be considered a supermarket, a retailer must have annual
sales of at least $2 million and contain all the major food departments found
in a traditional supermarket, including fresh meat and poultry, produce, dairy,
14The $2 million annual sales
dry and packaged foods, and frozen foods.14, 15
requirement has been used by the
retail food industry since at least
Two separate national-level directories of foodstores from the year 2006 were 1980. If adjusted for annual inflation,
used to develop a comprehensive list of supermarkets in the U.S. The first the equivalent in 2008 dollars is
directory is a list of authorized stores that accept SNAP benefits. More than approximately $4.5 million. By
166,000 outlets were authorized in 2006, but only approximately 34,000 met using the unadjusted annual sales,
we potentially include medium-sized
the supermarket definition criteria. In addition to the store name and address,
grocery stores in both the industry and
SNAP data include a store type classification, the most recent authorization SNAP store directories.
year’s total sales and total food sales, and total SNAP redemptions. The 15Supercenters are included in our
SNAP data were augmented with additional supermarket data from Trade definition of supermarkets. However,
Dimensions TDLinx (a Nielsen company), a proprietary source of individual warehouse club stores, also known
supermarket store listings also for the year 2006.16 This data set includes as wholesale club stores, were not
the name and address of supermarkets, the type of supermarket, annual sales included in this study for two reasons.
First, warehouse/wholesale club stores
volume range, and other supermarket characteristics. Details on how these
are not considered to be supermarkets
data were merged and cleaned can be found in Appendix C. The combined by the industry, and second, few of
list of supermarkets was converted into a GIS-useable format by geocoding these stores have applied to accept
the street addresses into store point locations. The final combined data set SNAP benefits.
included locations for 40,108 supermarkets and supercenters nationwide. 16TDLinx data only include
Food is also sold in restaurants, fast food outlets, and related foodservice
establishments. In fact, nearly half of all food spending is on food away
from home (Martinez and Kaufman, 2008). In considering the effect of
food access on diet and health, access to restaurants and other foodservice
establishments is important because food from these sources accounts for
a significant part of the total diet. However, this national-level analysis
does not consider access to restaurants and other food service outlets.
Compared with foodstores and other retail food outlets, the monetary costs
of an equal quantity of food purchased in a restaurant are higher than the
costs at supermarkets or grocery stores. The cost of food sold in restaurants
represents about one-third of the price at which it is sold, so that two-thirds
of the cost of restaurant food is the premium of having someone prepare
and serve it to the customer. In a grocery store, the cost of food accounts
for about three-fourths of the retail price, on average. Thus, while eating at
a restaurant may diminish the time costs of buying and preparing food for
a consumer, those costs are eventually paid for by the consumer. For these
reasons, food eaten at restaurants is less affordable due to its higher per unit
cost relative to foodstores and other retail food outlets.
16
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Measuring access from the geographical unit to the foodstore
This study uses distance to the nearest supermarket as a measure of access.18 18Chapter 4 also uses a variety mea-
For each grid cell, the distance is calculated from its geographic center to sure of distance, which is the distance
to three different supermarkets.
the nearest supermarket. Median distances to the nearest supermarket are
calculated for the Nation as a whole and across different subpopulations.
Based on the grid measure of distance to the nearest supermarket, three
categories of access (high, medium, and low) are created for two types of
access—walking access and driving access. Walking access measures a
range of distances for which it is feasible to walk to a supermarket, while
drivable access measures a range of distances for which it is feasible to
drive to a supermarket. A time-based distance measure equivalent for both
walking and driving is developed. The walkability range is categorized
as either 1) high, if a supermarket is within a half mile; 2) medium, if a
supermarket is between ½ and 1 mile; and 3) low, if the nearest supermarket
is more than a mile away. For rural areas, a drivability range is also
measured. Drivability is categorized as either 1) high, if a supermarket is
within 10 miles; 2) medium, if a supermarket is between 10 and 20 miles;
and 3) low, if a supermarket is greater than 20 miles away.
17
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Finally, because it is difficult to use the same measures to characterize access
in densely populated urban areas compared with less populated suburban and
rural areas, a separate analysis is conducted using Census Urbanized Area
definitions. The three definitions are as follows: Urban Areas, densely settled
that contain 50,000 or more people, such as a core city and surrounding
suburbs; Urban Clusters, densely settled local areas that have at least 2,500
people but fewer than 50,000 people, such as smaller cities and towns; and
Rural Areas, low-density areas with populations of less than 2,500, including
all areas not classified as either Urban Areas or Urban Clusters.
Table 2-3
Household vehicle access and supermarket access
Households without access to a vehicle
Between 1/2 to 1 mile More than 1 mile
from a supermarket from a supermarket
Geographic area Total households1 Number Percent Number Percent
Millions Millions Millions
Total U.S. 104.9 3.4 3.3 2.4 2.3
Low-income areas 25.1 4.5 17.8 0.9 3.8
19
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
households with medium or low access (those more than one-half mile from
a supermarket) since those who have high access can walk to a supermarket
that, at most, is one-half mile away.
Only 2.4 million households, or 2.3 percent of all 104.9 million households
in the U.S., live more than a mile from a supermarket and do not have access
to a vehicle. An additional 3.4 million households, or 3.2 percent of all
households, do not have access to a vehicle and are between one-half to 1
mile from a supermarket. Thus, for the total U.S. population, between 2.3
and 5.7 percent of all households may be outside of a walking distance to a
supermarket and lack access to a vehicle.
Table 2.3 also presents the number of households without access to vehicles
and distance to supermarkets by urbanicity. These estimates show that a
greater percentage of all rural households lacks access to a vehicle and lives
more than 1 mile from a supermarket (4.4 percent of all rural households and
7.4 percent of all rural households in low-income areas). It is not surprising
that people in rural areas live farther from the nearest supermarkets. But it is
perhaps unexpected that a greater percentage lack access to a vehicle. Urban
areas have the smallest percentages of households without access to a vehicle
that are more than a mile from a supermarket. For urban areas, a greater
percentage of households without vehicles are between one-half to 1 mile
from the nearest supermarket (12.2 percent of all urban households without
vehicles or 22 percent of all households without vehicles in low-income
areas).
The analysis now turns specifically to supermarket access for areas with high
concentrations of low-income people. Map 2.1 shows low-income areas in
the U.S., which are 1-kilometer grid cells where more than 40 percent of the
total population has income less than or equal to 200 percent of the Federal
poverty level. The map shows the dispersion of low-income areas across the
country, but the map also shows greater concentration of low-income areas in
the South, Southwest, and Upper Plains States. Rural low-income areas are
better reflected on the map than urban low-income areas, which are difficult
to see on the national level view the map provides.
Table 2.4 focuses on these low-income areas. The right half of the table
shows the number (and percent) of people in low-income areas by access
level. It also shows the percent of the total U.S. population represented in
these low-income areas. The left half of the table focuses only on those
people with incomes below 200 percent of Federal poverty guidelines. It is
worth noting here, and will be supported later in the chapter, that low-income
people who live outside of low-income areas are, in general, farther from
supermarkets than low-income people who live in low-income areas.
20
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2-4
Supermarket access for people in low-income and higher-income areas (walking distances)
Low-income areas1
All people in low-income areas All low-income people in low-income areas
Percent of total
Access level2 Total number Percent of people in U.S. population Percent of low- Percent of total
(walking) (millions) low-income areas (millions) Total number income people U.S. population
1Low-income and non-low-income areas defined according to ERS criteria. See text for details.
2Highaccess defined as less than or equal to 1/2 mile of a supermarket. Medium access defined as more than 1/2 mile but less than or equal to
one mile from a supermarket. Low access defined as more than one mile from a supermarket.
Source: USDA, ERS analysis based on data from Census of Population, 2000 and the ERS-compiled supermarket directory for the contiguous
U.S. in 2006.
Map 2.1
Low income areas of the contiguous 48 United States (1 km grids in which 40 percent
of population have incomes below 200 percent of the Federal poverty level)
21
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The first point to note in table 2.4 is that only 33 percent of people in
low-income areas live in areas with low access. In contrast, 32 percent of
people in low-income areas live in high-access areas and 35 percent live
in medium-access areas. Thus, one estimate of the number of people who
live in low-income areas with low access to supermarkets is 23.5 million,
which is 8.4 percent of the total U.S. population (out of a total of 279.6
million people in 2000). If those with medium-access levels who live in
low-income areas are included, then 48.4 million, or 17 percent of the total
U.S. population, is more than half a mile from a supermarket.
Not all people in low-income areas, however, have low income. In fact,
only about half have income less than 200 percent of the Federal poverty
thresholds (36 million out of a total of 71 million). It is likely that those who
are not low income but live in low-income areas have adequate resources to
access supermarkets even if they are more than a mile away. If the analysis
excludes those who live in low-income areas, but who are not themselves,
low income, then only 11.5 million, or 4.1 percent of the total U.S.
population, has low access to supermarkets. Including those low-income
people in low-income areas with medium access yields a total of 24 million
people, or 8.6 percent of the total population.
Thus far, the study has considered urban, rural, and all areas in between
similarly. But distance does not mean the same thing in rural areas as
in urban areas. The focus next turns to analysis of access by urbanicity.
Each grid area is assigned to one of three Census Urbanized Areas: Urban
Areas, Urban Clusters, and Rural Areas. Together, these three urban types
characterize the range of urbanicity found in the U.S. population. A separate
analysis of access is conducted for each urban type at the national level.
Populations within an urban type are assumed to have similar levels of
density, measured as population per unit of area, such as per square mile or
square kilometer. Areas having similar population densities are more likely
to have similar levels of the built environment—the infrastructure (e.g.,
networks of roads, transportation services, utilities, communication networks,
and government services) and businesses, manufacturing plants, and retail
stores. Analysis by urban type results in measurement of access within
similar built environments, resulting in greater comparability across similar
populations regardless of location.
22
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.5
Urban area access to supermarkets--overall and for income and demographic
subpopulations (walking distance)
Distance to nearest supermarket
High access Medium access Low access
(0.5 miles or less) (Between 0.5-1 mile) (More than 1 mile)
Income Sub-
level of Number Total population Median Number Number Number
Population area (millions) percent percent (miles) (millions) Percent (millions) Percent (millions) Percent
Low
Total population income 45.3 100.0 24.4 0.57 19.2 42.5 19.0 42.1 7.0 15.4
of urban areas Higher
income 140.6 100.0 75.6 0.71 42.1 30.0 57.9 41.2 40.6 28.9
Total 185.9 100.0 100.0 61.4 33.0 76.9 41.4 47.6 25.6
Subpopulations
Low
income 23.5 51.9 46.4 0.56 10.2 43.3 9.8 41.5 3.6 15.1
Population with
Higher
low income
income 27.1 19.3 53.6 0.65 9.0 33.1 11.5 42.3 6.7 24.6
Total 50.6 27.2 100.0 19.1 37.8 21.2 41.9 10.2 20.2
Low
Households income 3.4 22.0 40.5 0.50 1.7 50.2 1.3 38.4 0.4 11.3
without access Higher
to a vehicle income 5.1 9.3 59.5 0.42 2.8 56.1 1.6 30.8 0.7 13.1
Total 8.5 12.2 100.0 4.6 53.7 2.9 33.9 1.1 12.4
Low
income 30.8 68.1 44.2 0.55 13.7 44.5 12.6 41.0 4.5 14.5
Non-White
Higher
population
income 38.9 27.7 55.8 0.60 15.3 39.4 15.6 40.1 8.0 20.5
Total 69.8 37.5 100.0 29.1 41.6 28.2 40.5 12.5 17.9
Low
income 4.5 10.0 20.3 0.58 1.8 40.8 2.0 43.7 0.7 15.6
Elderly
Higher
population
income 17.7 12.6 79.7 0.69 5.4 30.5 7.5 42.5 4.8 27.0
Total 22.2 11.9 100.0 7.2 32.6 9.5 42.7 5.5 24.6
Source: Source: USDA, ERS analysis based on data from Census of Population, 2000 and the ERS-compiled supermarket directory for the
contiguous U.S. in 2006.
Table 2.5 first presents data on supermarket access for urban areas. A greater
share of low-income individuals lives outside of low-income areas (53.6
percent) than in low-income areas (46.4 percent). Further, low-income
individuals who live in higher income areas live farther from supermarkets
than those who live in low-income urban areas. About 15 percent of those in
low-income urban areas are more than a mile from a supermarket, compared
with 29 percent for those in higher income areas. Median distances to
supermarkets reflect this as well.
The next rows focus specifically on access for low-income individuals, which
make up 27.2 percent of all urban dwellers. Results show that 43.3 percent
of low-income individuals who live in low-income areas are within one-half
mile of a supermarket and another 41.5 percent of are between half a mile to
one mile from the nearest supermarket. The remaining 15.1 percent of the
23
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
low-income individuals who live in low-income areas resided more than a
mile from the nearest supermarket.
Maps 2.2 and 2.3 illustrate these measures of access for two urban areas, the
Washington, DC, and St. Louis, Missouri, metro areas. For Washington, DC,
there were two low-income areas, outlined in black, that are largely contained
within the city’s boundaries. The St. Louis area has a large low-income area
that spans a good portion of the City of St. Louis in Missouri and across
the Mississippi River into Illinois. In both maps, circles shaded light green
indicate areas that are within a 1-mile radius of a supermarket. The color
shadings of the areas indicate population density where the darker shading
indicates grids with more people and the lighter shading indicates grids
with fewer people. This study focuses particular attention on areas that are
Map 2.2
Washington, DC Urban Area
24
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
outside of the one-mile radius of a supermarket and in the neighborhoods
with low income.
Map 2.2 shows that there are some low-income areas within DC and on the
border with Prince Georges County, Maryland, which are outside of 1 mile
from a supermarket. But for most of these areas, the population density
is low to moderate. The situation looks more severe in the St. Louis area.
While the most densely populated grids are within one mile of a supermarket,
there are several low-income grids with moderate to high densities that are
more than a mile from a supermarket, especially in the central and northern
part of St. Louis, Missouri. The low-income area just east of the Mississippi
River in Illinois has few supermarkets, but there are few grids outside of one
mile that have moderate to high population densities. The Illinois side of the
river has many grids with low population densities that are more than a mile
from a supermarket.
Table 2.5 also shows supermarket access for households without access to
vehicles, overall, and then separately by whether or not the households live in
low-income or higher income areas. About 12.2 percent of urban households
Map 2.3
St. Louis, Missouri Urban Area
25
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
do not have access to a vehicle; however, this number is almost twice as
large—22 percent—for households in low-income areas. Those who live
in higher income areas are much more likely to have access to a vehicle—
only 9.3 percent do not have access to a vehicle. Just over 50 percent of
low-income-area households without access to a vehicle had a high level of
access. Another 38.4 percent of this population had medium access. The
remaining 11.3 percent of households living in low-income areas without
access to vehicles were more than 1 mile from the nearest supermarket and
had low access to supermarkets. Among households that lived outside of
low-income areas and did not have access to a vehicle, a greater percentage
had high access than similar households that lived in low-income areas, but
fewer had medium access and more had low access.
A greater share of the elderly population in low-income areas had high access
(40.8 percent) than those living outside low-income areas (30.5 percent).
While both groups have about the same percentage of the population of
elderly with medium access, the greatest difference occurred among the
low-access group. Almost 16 percent of the elderly who lived in low-income
areas were outside of 1 mile from the nearest supermarket, but 27 percent
of those living outside low-income areas were more than a mile from a
supermarket.
The population in urban clusters (areas with at least 2,500 people but fewer
than 50,000 people) makes up 9.1 percent of the total U.S. population. Table
2.6 presents accessibility measures for vulnerable populations in low-income
and higher income areas within urban clusters.
26
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.6
Urban cluster access to supermarkets--overall and for income and demographic
subpopulations (walking distance)
Distance to nearest supermarket
High access Medium access Low access
(0.5 miles or less) (Between 0.5-1 mile) (More than 1 mile)
Income Sub-
level of Number Total population Median Number Number Number
Population area (millions) percent percent (miles) (millions) Percent (millions) Percent (millions) Percent
Low
Total population income 9.9 100.0 38.8 0.72 2.8 28.7 4.1 41.8 2.9 29.6
of urban clusters Higher
income 15.6 100.0 61.2 0.82 3.6 23.4 5.9 38.0 6.0 38.6
Total 25.5 100.0 100.0 6.5 25.4 10.1 39.5 8.9 35.1
Subpopulations
Low
income 4.8 48.6 54.4 0.71 1.4 28.7 2.0 42.0 1.4 29.3
Population with
Higher
low income
income 4.0 25.8 45.6 0.77 1.0 25.5 1.6 39.5 1.4 35.0
Total 8.8 34.6 100.0 2.4 27.3 3.6 40.9 2.8 31.9
Low
Households income 0.5 13.5 52.3 0.66 0.2 32.4 0.2 43.1 0.1 24.5
without access Higher
to a vehicle income 0.4 7.3 47.6 0.69 0.1 30.5 0.2 41.5 0.1 28.0
Total 0.9 9.6 99.9 0.3 31.5 0.4 42.3 0.2 26.1
Low
income 4.0 40.8 66.7 0.75 1.1 26.4 1.7 41.5 1.3 32.1
Non-White
Higher
population
income 2.0 12.9 33.3 0.85 0.4 22.3 0.7 36.3 0.8 41.5
Total 6.0 23.7 100.0 1.5 25.0 2.4 39.7 2.1 35.3
Low
income 1.4 13.8 35.2 0.68 1.5 30.7 0.6 42.9 0.4 26.4
Elderly
Higher
population
income 2.5 16.1 64.8 0.78 1.5 25.0 1.0 39.5 0.9 35.6
Total 3.9 15.2 100.0 3.0 27.0 1.6 40.7 1.3 32.3
Source: USDA, ERS analysis based on data from Census of Population, 2000 and the ERS-compiled supermarket directory
for the contiguous U.S. in 2006.
low-income areas had slightly better access to supermarkets than those who
lived outside of low-income areas, but the distributions are very similar.
areas (93 percent) had better access than those who lived in low-income
areas (85 percent). This is in contrast to low-income populations in urban
areas and urban clusters where low-income individuals who lived in
low-income areas were closer to supermarkets than low-income individuals
who lived in higher income areas. For rural populations, supermarket access
typically involves driving to an urban area or urban cluster, where higher
population densities are more likely to contain larger stores. Map 2.4 shows
supermarket access of South Dakota, a largely rural state. This map uses
light blue shading to indicate the location of low-income areas. Circles
with the 10 and 20 mile radii around a supermarket indicate which areas are
within each of these distances of a supermarket. Finally, the brown shading
scheme (white to dark brown) indicates population density (low to high).
The map shows that there are several small towns in low-income portions
of the State that are outside of a driving range from a supermarket (more
28
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Map 2.4
Supermarket access in South Dakota
29
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
population certainly has very limited access to supermarkets based on driving
distances.
Table 2.8 shows the average time spent in travel to grocery shopping on an
average day by level of access to the nearest supermarket (as defined above).
The table shows the average minutes spent traveling to grocery stores for
shoppers who live in low-income areas with low, medium, and high access
to supermarkets.22 These averages are compared with the national average. 22Survey respondents did not report
Overall, the national average time spent traveling, one-way, to the grocery which type of “grocery” store they vis-
store was almost 15 minutes, and about 14 percent of the population traveled ited, only that they reported the activity
of grocery shopping.
to the grocery store on an average day.
Time spent traveling to the grocery store was greater in low-income areas
with low-access. The average time spent traveling to the grocery store for
those who lived in these areas, 19.5 minutes, was significantly greater than
the average time spent traveling to the grocery store for those in low-income
areas with high access (15.5 minutes) and for those in low-income areas
30
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.8
Average time spent in travel to grocery shopping on an average day by access to grocery stores
Average time is one-way, not total travel time (based on the shortest one-way time). Pooled 2003-2007 American Time Use Survey data
Average minutes per day Average % engaged Average Average Average
of travel related to grocery in travel related to minutes, minutes, Average percent,
shopping, for those who grocery shopping 90% CI 90% CI percent, 90% CI
grocery shopped (on an average day) min max 90 CI min max
Minutes Percent
Total population, age 15+, 2003-07 15.0 14.0 14.7 15.3 13.7 14.3
Low-income areas
Low access 19.5 12.1 18.1 20.9 11.1 13.1
Medium access 14.1 13.5 13.0 15.1 12.5 14.5
High access 15.5 12.3 14.3 16.7 11.3 13.4
Not-low-income areas
Low access 15.9 14.4 15.2 16.5 13.7 15.1
Medium access 12.5 14.7 12.1 12.9 14.1 15.3
High access 13.3 16.3 12.6 14.1 15.4 17.3
Income, 2003-07
Household Income <= 200%
poverty threshold 15.8 13.6 15.3 16.4 13.0 14.2
Low-income areas
Low access 19.3 13.7 17.3 21.3 12.2 15.1
Medium access 14.2 13.4 13.1 15.3 12.0 14.7
High access 16.4 12.5 14.5 18.2 11.0 14.0
Not-low-income areas
Low access 16.3 14.7 15.0 17.6 13.2 16.3
Medium access 13.6 13.3 12.6 14.7 12.0 14.6
High access 2.3 16.7 11.1 13.6 14.6 18.8
Household Income > 200% poverty
threshold 14.2 14.2 13.9 14.6 13.8 14.7
Low-income areas
Low access 20.5 11.3 18.3 22.6 9.8 12.8
Medium access 12.1 14.1 10.9 13.4 12.3 15.8
High access 13.5 12.6 11.9 15.2 10.9 14.4
Not-low-income areas
Low access 15.6 14.3 14.8 16.4 13.4 15.2
Medium access 11.8 15.0 11.4 12.3 14.2 15.9
High access 13.4 16.3 12.3 14.4 15.1 17.6
31
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.8
Average time spent in travel to grocery shopping on an average day by access to grocery stores (continued)
Average time is one-way, not total travel time (based on the shortest one-way time). Pooled 2003-2007 American Time Use Survey data
Average minutes per day Average % engaged Average Average Average
of travel related to grocery in travel related to minutes, minutes, Average percent,
shopping, for those who grocery shopping 90% CI 90% CI percent, 90% CI
grocery shopped (on an average day) min max 90 CI min max
Minutes Percent
Household Income missing 16.3 13.9 15.4 17.2 13.1 14.7
Low-income areas
Low access 17.7 9.7 14.4 21.1 7.5 12.0
Medium access 19.4 12.3 14.5 24.5 9.6 14.9
High access 17.0 11.2 12.8 21.1 8.8 13.5
Not-low-income areas
Low access 16.5 14.4 14.9 18.0 12.5 16.3
Medium access 14.4 15.0 12.9 15.8 13.1 16.9
High access 14.4 15.9 12.4 16.4 13.5 18.3
-- indicates that estimate is suppressed due to small cell size.
Source: 2003-2007 American Time Use Survey data, Current Population Survey sampling frame from Census Bureau.
Store access levels are tract-level classifications based on categories of access used in Chapter 2.
Data with missing tract-level classification were included in Total population, age 15+ estimates.
with medium access (14.1 minutes). These differences are large and
statistically significant. In addition, those in low-access areas shopped less
frequently—on average once every 8 days versus a national average of once
every 7 days.
The difference in average time spent traveling to the grocery store by access
level may not be surprising given that this study’s definition of access is
based on distance, and that, all else equal, it is expected that those who live
more than 1 mile from a supermarket would spend more time traveling to
the grocery store than those who live less than half a mile or less than a mile
from the supermarket. To put these averages into context, table 2.8 also
reports average time spent traveling to grocery stores by households in higher
income areas separately by their access levels. As expected, those with low
access spend the most time traveling to the grocery store (15.8 minutes)
compared with those who are closer. But the average of those in higher
income areas that are more than a mile from a store is still almost 4 minutes
shorter than the 19.5 minute average of those in low-income areas who are
more than a mile from a grocery store.
Table 2.8 also considers average time spent traveling to get groceries for
sample members living in metropolitan (metro) areas compared with those
23There are not enough households
living in nonmetropolitan (nonmetro) areas.23 Not surprisingly, those who
in urban clusters or rural areas in the
live in nonmetro areas spent more time, on average (16.9 minutes), traveling
sample to examine average minutes
to the grocery store than those living in metro areas (14.2 minutes), and a spent traveling to supermarkets across
smaller percent of nonmetro residents shopped on an average day than metro the same urbanicity categories used
residents. Surprisingly, those in nonmetro areas with low access spend about above, so here we focus on metro and
the same amount of time traveling to supermarkets as those in metro areas nonmetro statistical areas.
with low access.
The final set of averages shown in table 2.8 compares average time spent
traveling to grocery stores for those with household income below 200
percent of Federal poverty guidelines and for those with income above 200
percent of poverty. Also included are national estimates for the 13 percent of
the sample with missing income information. Those with low income who
32
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
live in low-income areas with low access spend about the same amount of
time traveling to grocery stores (19.3 minutes) as those who do not have low
income but who live in low-income areas with low access to grocery stores
(20.5 minutes). In contrast, those with low income who live in low-income
areas with medium or high access take more time to get to the grocery store
(14.2 minutes and 16.4 minutes, respectively) than those who do not have
low income but who live in low-income areas with medium (12.1 minutes)
or high access (13.5 minutes). It is possible that these higher income
individuals in low-income areas have access to their own vehicles for grocery
shopping and choose to shop outside their neighborhoods.
Grocery shoppers from low-access areas were more likely to have been
accompanied by children on their trips to the grocery store than others—29.1
percent versus a national average of 22.8 percent. Having children along
on the trip is likely to make the trip more cumbersome, making travel and
grocery shopping more difficult for these low-access shoppers.
The last rows in table 2.9 show whether grocery shoppers shop from home
25Appendix
or from work, or their trip chaining patterns.25 For the majority of shoppers, C contains information on
the time distance from the grocery store to home is shorter than the time how trip chaining was classified.
distance from the grocery store to work (about 92 percent). But for about 8
percent, the time distance from work to the grocery store was shorter than
the time distance from home to the grocery store (5.9 percent directly from
work to the grocery store and 2.1 percent clustered with other stops from
work to the grocery store). Interestingly, those in low-income areas with low
access were the most likely to access grocery stores directly, bunched with
other activities, or from work (7.7 percent directly from work and 3.6 percent
33
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 2.9
Characteristics of grocery shopping by level of access to supermarkets
Characteristics are of one-way shortest travel time to grocery store.
Pooled 2003-2007 American Time Use Survey data
Low-income areas Not-low-income areas
Low Medium High Low Medium High
Total access access access access access access
Percent
Mode of transportation
1. Car, truck, motorcycle (driver or passenger w/hh member) 90.2 93.3 87.1 65.3 96.7 92.3 83.9
2. Walking or bicycle 4.8 2.3 5.4 23.1 0.3 3.1 10.0
3. Public transportation (bus, subway/train) 0.3 0.1 0.9 1.9 0.0 0.3 0.3
4. Other (passenger w/nonhh member, boat/ferry,
taxi/limo, unspecified) 4.8 4.3 6.6 9.7 2.9 4.4 5.8
Total 100.0 100.0 100.00 100.0 100.0 100.0 100.00
With whom
Alone 48.8 40.0 39.0 40.8 49.9 52.7 53.6
With household members 42.1 49.3 49.9 46.1 41.1 39.5 37.9
With others, not household members 9.1 10.8 11.1 13.1 9.1 7.8 8.6
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0
With children (persons under 18 years old) 22.8 29.1 28.2 32.8 20.3 22.6 19.4
Note that “with whom” is for travel to grocery store, and not grocery shopping. The person or
persons with the respondent may only be present for part of the travel.
Trip chaining
Home to store, direct / Store to home direct 63.6 54.8 64.1 61.0 64.5 68.4 66.6
Home to store, clustered activities/ Store to home, clustered activities 28.4 33.9 31.1 34.9 26.5 25.8 26.3
Work to store, direct / Store to work direct 5.9 7.7 3.5 3.3 6.6 3.9 5.9
Work to store, clustered activities/ Store to work, clustered activities 2.1 3.6 1.4 0.8 2.5 1.9 1.3
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Source: 2003-2007 American Time Use Survey data, Current Population Survey sampling frame from Census Bureau. Store
access levels are tract-level classifications based on categories of access used in Chapter 2. Data with missing tract-level
classification were included in Total population estimates.
bunched with other activities from work). Those from low-income areas that
had medium or high levels of access were less likely to access grocery stores
from work. These data indicate that some of those who live in low income
areas with low access choose grocery stores closer to work than to home.
Data presented here are for the entire U.S. population. One study specifically
focused on a sample of low-income people. The National Food Stamp
Program Survey of 1996/1997 (NFSPS) surveyed a sample of participants
of the SNAP and eligible nonparticipants. Sample members were asked
about the modes of transportation and out-of-pocket costs used to travel
to stores where they shopped for food and about how much time it took to
travel to foodstores (Ohls et al., 1999). Close to 76 percent of participants
and 85 percent of eligible nonparticipants reported use of a car to shop.
Food stamp participants either drove (45 percent) or got a ride with family
or friends (31 percent). Among the 22 percent of participants who reported
some transportation expenses, the average cost per shopping trip was $6.54.
Average round trip travel time to the most frequently used store was 23-24
minutes for participants and eligible nonparticipants. The survey compared
these patterns across participants who lived in urban, mixed, and rural areas.
34
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Average distance, time, and out-of-pocket cost to the most often used store
were greater for those living in rural settings. Rural participants were less
likely than their urban area counterparts to report out-of-pocket expenses
associated with food shopping—perhaps because 94 percent either drove or
got a ride with others.
Time costs to travel to grocery stores are only part of the time costs involved
in healthy eating—preparing nutritious food can be more costly in terms
of time than prepared meals or restaurant meals. Previous ERS research
examined time spent in food preparation for women across income levels
and family composition. This study found that being a mother who worked
full-time or a single mother were more important in explaining differences in
time spent in food preparation than were either earnings or income (Mancino
and Newman, 2007). Specifically, food preparation time falls as mothers
spend more time working outside the home. Single women with children
spend less time preparing food than married women.
Summary
The number of people who have low access to healthy food depends upon
which measure is used. Direct questions from a nationally representative
sample of U.S. households in 2001 show that up to 5.7 percent of all U.S.
households did not always have the food they wanted or needed because of
access-related problems. Households that live far from a supermarket and
that do not have vehicles likely have limited access to nutritious food. Of
all households in the U.S., 2.3 million, or 2.2 percent, live more than a mile
from a supermarket and do not have access to a vehicle. An addition 3.4
million households, or 3.2 percent of all households, live between one-half to
1 mile and do not have access to a vehicle.
Within urban areas, 10.1 million low-income individuals (20.2 percent) were
more than 1 mile from the nearest supermarket. Of the total, 3.6 million
lived in low-income areas. Within urban clusters, 1.4 million persons,
or 29.3 percent of the low-income population, were more than 1 mile
from the nearest supermarket. Of the 16.1 million persons living in rural
low-income areas, 85.7 percent (13.7 million persons) were within 10 miles
of a supermarket. Another 11.7 percent (1.9 million persons) were between
10 and 20 miles distant, and only 2.6 percent were more than 20 miles from
a supermarket. These differences underscore the importance of owning a
vehicle or having access to affordable transportation in rural areas.
Data on time use and travel mode show that those who live in low-income
areas that are more than a mile from a supermarket spend more time (19.5
35
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
minutes) traveling to grocery stores than the national average (15 minutes).
However, 93 percent of those who live in low-income areas more than a mile
from a supermarket traveled to the grocery store in a vehicle they or another
household member drove.
While considerable efforts were made to develop these data, measures, and
methods, different assumptions and measures are likely to produce differing
outcomes and conclusions when applied to the same data. More detailed
information which is not currently available would likely result in more
precise findings. It is hoped that the methods and findings in this chapter
will stimulate new research to provide additional insights about the nature
and extent of low-income populations faced with low access to sources of
nutritious and affordable foods.
References
Algert, S., J. Aditya Agrawal, and D. Lewis (2006). “Disparities in Access
to Fresh Produce in Low-Income Neighborhoods in Los Angeles,” American
Journal of Preventive Medicine 30(5): 365-370.
Chung, C., and S.L. Myers (1999). “Do the Poor Pay More for Food? An
Analysis of Grocery Store Availability and Food Price Disparities,” Journal
of Consumer Affairs 33: 276-96.
Cotterill, R., and A. Franklin (1995). “The Urban Grocery Store Gap,” Food
Marketing Policy Center, University of Connecticut.
36
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Donohue, R.M. (1997). “Abandonment and Revitalization of Central City
Retailing: The Case of Grocery Stores,” The University of Michigan DAI-A
58/10.
Kaufman, P., et al. (1997). Do the Poor Pay More for Food? Item
Selection and Price Differences Affect Low-Income Household Food Costs,
Agricultural Economic Report No. 759, U.S. Department of Agriculture,
Economic Research Service.
Mancino, L., and C. Newman (2007). Who Has Time to Cook? How Family
Resources Influence Food Preparation, U.S. Department of Agriculture,
Economic Research Service, Economic Research Report No. 40, http://www.
ers.usda.gov/publications/ERR40/err40.pdf
Mantovani, R.E., Daft, L., Macaluso, T.F., & Welsh, J., & Hoffman, K.
(1997, February). Authorized Food Retailer Characteristics Study (Technical
Report IV): Authorized Food Retailers’ Characteristics and Access Study.
Report submitted to U.S. Department of Agriculture, Food and Nutrition
Service by Macro International, Inc.
Morton, L.W., and T.C. Blanchard (2007). “Starved for Access: Life in
Rural America’s Food Deserts,” Rural Realities 1(4): 1-10.
37
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Neckerman, Kathryn M., M. Bader, M. Purciel, and P. Yousefzadeh (2009).
“Measuring Food Access in Urban Areas,” National Poverty Center Working
Paper, http://www.npc.umich.edu/news/events/food-access/index.php
Pike, J. (2000). “Boston’s Poor Pay More For Food, Theft and Other
Problems Increase Business Costs, Prices,” Massachusetts News, August 2.
Zenk S.N., A.J. Schulz, B.A. Israel, S.A. James et al. (2005). “Neighborhood
Racial Composition, Neighborhood Poverty, and the Spatial Accessibility of
Supermarkets in Metropolitan Detroit,” American Journal of Public Health,
2005; 95(4):660.
38
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 3
The food deserts literature suggests that those who have better access
to supermarkets tend to have healthier diets and lower levels of obesity
and related diseases (Laraia et al., 2007; Larson et al., 2009). However,
the extent to which limited access to supermarkets and other differential
aspects of the food environment contribute to known economic and racial
health disparities remains unclear. That lack of clarity stems in part from
conflicting findings with regard to access to supermarkets for low-income,
minority, and racially mixed neighborhoods. Some researchers have
found economically disadvantaged populations and neighborhoods with
high concentrations of racial minorities to have better access than their
counterparts (e.g., Moore and Diez Roux, 2006), while others have found the
reverse (e.g., Burns and Inglis, 2007). There is substantial literature showing
that low-income and minority populations are disproportionately at risk
when it comes to major public health concerns (e.g., Diez Roux et al., 2001;
Odoms-Young et al., 2009). But critical gaps remain in the understanding of
the associations between the neighborhood food environment and indicators
of health. Investigators have made considerable gains in the development of
approaches to better characterize neighborhood food environments and thus
identify the mechanisms that underlie those associations between the food
environment and poor health outcomes (e.g., Moore et al., 2008; Sharkey,
2009), but doing so continues to be a major challenge.
39
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
to a food retail environment that offers too few choices for nutritious food
and/or too many options for less nutritious alternatives. Thus, the food
environments of low-income populations require special consideration due
to the vulnerability of the individuals as well as that of the unique social and
physical setting in which they live (Gittelsohn and Sharma, 2009).
40
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
distance between one-half and 1 mile for walking and between 10 and 20
miles for driving, and high-access defined by a distance less than one-half
mile for walking and 10 miles for driving). However, for this analysis, the
geographic area of interest is neighborhoods, for which Census tracts are
commonly used proxies. As such, the grid-defined distance measures used in
Chapter 2 were aggregated to the Census tract for this study.27 27Specifically, tract-level distance
41
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 3.1
Access Category Variables and Corresponding Percent Frequency: Metro Core,
Micropolitan / Small Town Core, and Rural Tracts
Percent of micropolitan
Access category variables Percent of metro core tracts or small-town core tracts Percent of rural tracts
Percent
Walking--1 store (proximity)
High access 31.5 12.4 1.3
Medium access 42.6 46.9 8.6
Low access 25.9 40.7 90.1
Driving--1 store (proximity)
High access 100.0 99.8 75.8
Medium access 0.0 0.1 17.8
Low access 0.0 0.1 6.4
Walking--3 stores (variety)
High access 0.0 0.0 0.0
Medium access 12.8 3.3 0.3
Low access 87.2 96.7 99.7
Driving--3 stores (variety)
High access 100.0 84.5 32.5
Medium access 0.0 11.0 45.5
Low access 0.0 4.5 22.0
Source: USDA, ERS estimations based on 2006 ERS-compiled directory of supermarkets and RAND’s Center for Population Health
and Health Disparities Data Care.
order to be evenly distributed across the county (or metro area) with
respect to other racial groups. A dissimilarity statistic value of 0 reflects
absolute integration while a value of 1 reflects absolute segregation.
42
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 3.2
Summary Descriptives; Factor and Predictor Variables: Metro Core,
Micropolitan / Small Town Core, and Rural Tracts
Rural core Micro/Small-town core Metro core
Std. Std. Std.
Variable Mean deviation Mean deviation Mean deviation
Median value of owner-occupied housing $72,056 $53,004 $85,896 $50,248 $153,295 $123,395
Percent Black or African-American population 5.5% 13.8% 10.4% 18.5% 15.6% 25.3%
Percent Hispanic population 4.8% 11.4% 8.6% 16.2% 13.5% 20.1%
Percent linguistically isolated households 5.8% 10.5% 5.7% 7.3% 11.2% 10.4%
Percent of tract poverty population who are
65+ years old 14.2% 8.4% 12.7% 8.1% 11.8% 12.2%
Percent of tract poverty population who are
children 31.4% 10.3% 32.2% 10.5% 29.7% 14.6%
Percent vacant housing units 20.2% 15.3% 10.5% 8.4% 6.6% 6.6%
Percent persistent poverty county 14.8% 35.5% 9.9% 29.9% 1.7% 13.0%
43
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Discussion of Results
MDA yields two types of output that are particularly useful for this
investigation. The first is the structure coefficients, which are useful for
determining the characteristics that contribute the most to group separation
(i.e., low-, medium-, and high-access groups). The second is the set of
standardized discriminant coefficients (similar to beta coefficients in
regression analysis), which are useful for identifying the characteristics that
best predict group membership and can thereby be thought of as indicators of
associated risk. That output for all 12 discriminant analyses, and, therefore,
the three geographies (metro core, micropolitan/small-town core, and rural
core), are summarized in two typology tables.
The second and third most powerful predictors for rural core neighborhoods
were found to be the index of disadvantage (-.311) and the percent of the
tract poverty population who are 65 years of age or older (.252). This finding
implies, for example, that limited access is associated with a combination of
indicators of socioeconomic disadvantage. This reinforces the need to give
greater consideration to the notion of deprivation amplification in access
and related health inequalities research. Further, the relevance of the elderly
poverty population highlights that some people may face multiple barriers
to access. Thus, continued efforts to improve measurement and advance
multivariate techniques are needed.
The discriminant analysis results presented in tables 3.3 and 3.4 offer
evidence in support of the hypothesis that segregation by race and income are
associated with limited access. Segregation by race and income inequality
are the dominant predictors among all neighborhood and household context
variables that predict low-, medium-, and high-access levels. This was
particularly true for access based on driving distances in rural core and
micro/small-town core neighborhoods. However, some variability in
44
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 3.3
Group Separation and Predictor Variables for Low-, Medium-, and High-Access to Supermarkets:
Walking Proximity and Variety for Metro Core, Micropolitan / Small Town Core, and Rural Neighborhoods
Walking proximity Walking variety
Rural Micro/Small- Metro Micro/Small- Metro
Access group separation core town core core Rural core town core core
Percent Black or African-American population -0.065 0.046 0.130 0.066 0.045 0.108
Percent Hispanic population 0.093 -0.079 0.379 0.068 -0.155 0.506
Percent linguistically isolated households -0.118 -0.020 0.489 0.089 -0.188 0.778
Percent of tract poverty population who are
65+ years old 0.201 -0.027 -0.062 0.258 0.091 -0.091
Percent of tract poverty population
who are children -0.129 0.039 -0.038 0.456 0.188 -0.026
Percent vacant housing units -0.356 0.184 -0.027 0.207 0.135 -0.046
SES index of disadvantage -0.234 0.159 -0.304 0.834 0.318 -0.327
Persistent poverty county -0.120 0.047 -0.070 0.071 0.022 -0.055
Percent linguistically isolated households -0.190 -0.004 0.390 0.139 -0.102 0.605
Percent of tract poverty population who
are 65+ years old 0.252 -0.097 -0.031 0.044 0.022 -0.089
Percent of tract poverty population
who are children 0.014 -0.013 -0.172 0.069 0.155 -0.173
Percent vacant housing units -0.054 0.121 -0.113 0.117 0.127 -0.108
SES index of disadvantage -0.511 0.153 -0.171 0.582 0.451 -0.241
Persistent poverty county -0.052 -0.045 -0.150 0.044 -0.071 -0.141
Source: ERS estimates based on 2006 ERS-compiled directory of supermarkets and RAND’s Center for Population Health
and Health Disparities Data Care.
45
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 3.4
Group Separation and Predictor Variables for Low-, Medium-, and High-Access to Supermarkets:
Driving Proximity and Variety for Metro Core, Micropolitan / Small Town Core, and Rural Neighborhoods
Driving proximity Driving variety
Rural Micro/Small- Metro Micro/Small- Metro
Access group separation core town core core Rural core town core core
Land area of tract in miles 0.725 0.900 n/a 0.640 0.584 0.693
Roadway connectivity (alpha) 0.334 -0.063 n/a 0.374 0.202 -0.154
Median value of owner-occupied housing -0.313 -0.033 n/a -0.266 -0.211 0.113
Percent rural population -0.005 0.105 n/a 0.099 0.071 0.483
Percent Black or African-American population -0.173 -0.041 n/a -0.178 -0.134 -0.060
Percent Hispanic population 0.191 0.055 n/a 0.147 0.307 -0.086
Percent linguistically isolated households 0.323 0.265 n/a 0.194 0.178 0.182
Percent of tract poverty population
who are 65+ years old -0.056 -0.025 n/a 0.035 0.081 0.122
Percent of tract poverty population
who are children 0.003 0.073 n/a 0.069 0.137 -0.080
Percent vacant housing units 0.182 0.133 n/a 0.179 0.257 0.413
SES index of disadvantage -0.077 -0.187 n/a 0.036 -0.092 0.040
Persistent poverty county 0.072 0.052 n/a 0.018 0.106 0.350
Percent linguistically isolated households 0.350 0.237 n/a 0.243 0.156 0.455
Percent of tract poverty population who are
65+ years old -0.052 0.111 n/a 0.058 0.187 0.088
Percent of tract poverty population who are
children -0.169 0.101 n/a -0.037 0.071 0.020
Percent vacant housing units 0.302 0.032 n/a 0.401 0.283 0.429
SES index of disadvantage -0.016 -0.385 n/a -0.145 0.017 -0.034
Persistent poverty county 0.104 0.016 n/a 0.046 0.043 0.327
Source: ERS estimates based on 2006 ERS-compiled directory of supermarkets and RAND’s Center for Population Health
and Health Disparities Data Care.
46
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
predictors was found; for example, in rural core areas segregation was not
among the top predictors of walking accessibility. This finding lends some
support to the need for situation-specific research and policy.
Summary
The findings of this study indicate that low-access to supermarkets is
most heavily influenced by characteristics of neighborhood and household
socioeconomic environments, such as the extent of income inequality, racial
segregation, transportation infrastructure, housing vacancies, household
deprivation, and rurality. This lends support to the notion that there is
indeed a socioeconomic “contextual effect” that should be considered when
designing food access policy. In that vein, there is growing evidence that
documents the success of non-health interventions that have had a positive
impact on health (e.g., improvements to road networks and investments in
public transportation options), particularly for those living in deprived areas
(e.g., Cassady and Mohan, 2004; Wrigley et al., 2003).
References
Apparicio, P., M.S. Cloutier, and Shearmur, R. (2007). “The Case of
Montreal’s Missing Food Deserts: Evaluation of Accessibility of Food
Supermarkets,” Int J Health Geogr, February 12, 6:4.
Franco, M., A.V. Diez Roux, T.A. Glass, B. Caballero, and F.L. Brancati
(2008). “Neighborhood Characteristics and Availability of Healthy Foods in
Baltimore,” Amer J Prev Med, December 35(6): 561-7.
47
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Gittelsohn, J., and S. Sharma (2009). “Physical, Consumer, and Social
Aspects of Measuring the Food Environment Among Diverse Low-Income
Populations,” Am J Prev Med, April, 36(4 Suppl): S161-5.
48
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Wrigley, N., D.L. Warm, and B.M. Margetts (2003). “Deprivation, Diet
and Food Retail Access: Findings From the Leeds ‘Food Deserts’ Study,”
Environment and Planning A., 35: 151-88.
49
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 4
Many factors contribute to an individual’s overall diet, body weight, and the
risk of developing diet-related diseases, such as diabetes or cardiovascular
disease. Individual factors can explain some but not all of the differences
in the rates in which different population groups experience these problems.
Focus on food access has increased as researchers try to better understand
the factors besides individual behaviors that may lead to differences in diet
and health outcomes (Diez-Roux, 2009). Interest in the relationship of food
access to diet and health is also rooted in a substantial body of literature
that shows disparities in many health outcomes across race, ethnicity, and
socioeconomic status (Institute of Medicine, 2003; National Research
Council, 2004). It is hypothesized that differences in food access across
race, ethnicity, and socioeconomic status may contribute to or reinforce these
health disparities (Diez-Roux, 2009).
Conceptual Framework
In a simple conceptual model, it is hypothesized that individual (and family)
characteristics as well as characteristics of the physical environment impact
dietary decisions. Individual characteristics include demographics (age, sex,
race/ethnicity), socioeconomic status (income, education, and employment),
family characteristics (family size and composition, presence of children),
and preferences for food and other goods. The physical environment
includes the food environment (accessibility to stores and restaurants) along
with characteristics of the built environment, such as parks, sidewalks,
availability of public transportation, air pollution, and noise. It is also likely
that the social environment faced by individuals and families (e.g., cultural
and social norms, social support, and safety and violence) affects diet. Diet
is a major determinant of BMI and obesity status, and it is also a factor in
risks of such diseases as diabetes and cardiovascular disease. Some of the
same individual factors and physical and social environments affect BMI and
diet-related diseases as well.
51
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
makeup, exercise habits, and working conditions. The directions of these
relationships are not necessarily one-way. Obese individuals may have
difficulty exercising or engaging in an active lifestyle. Obesity has also
been tied to lower wages among women, which could affect their family’s
economic situation (Averett and Korenman, 1996; Cawley, 2004). And
health conditions themselves can impact diet; for example, those with high
blood pressure or diabetes may need to adopt special diets.
Literature Review
In reviewing the literature on food access and diet and health outcomes,
one must first distinguish between studies that attempt to examine causal
links between food access and health outcomes and those that only consider
cross-sectional correlations. Also, most of the studies examine the effects
of food access on proximal outcomes, such as food shopping behavior and
food consumption, often focusing on particular foods such as fruits and
vegetables, whole grains, or low-fat milk. Other studies, however, focus
on outcomes such as high BMI, obesity, and heart disease, which are not
as proximate outcomes. The causes of these more distant outcomes are
much broader than just food access, and, thus, other factors besides lack of
access to some foods may help account for the rate of incidence of these
health outcomes. To get a sense of this, the analysis includes a review
of literature on the degree to which specific foods that may be lacking in
some neighborhoods (such as fresh fruits and vegetables, low fat milk,
and whole grains) are related to such health outcomes as obesity, diabetes,
cardiovascular disease, and cancer.
The majority of studies that have examined the relationship between store
access and dietary intake find that better access to a supermarket or large
grocery store is associated with healthier food intakes (Larson et al., 2009).
The relationship between the availability of restaurants (both fast food and
full-service) and dietary intake has also been studied. In general, these
studies have found that greater availability of fast food restaurants and lower
prices of fast food restaurant items are related to poorer diet. Access to full-
service restaurants shows either no relationship or a positive relationship with
healthy dietary intake.
Only a few studies have used longitudinal data to measure how changes
in access affect changes in diet. The few that exist focus on changes in
shopping behavior and changes in dietary intake, not more distant outcomes
such as obesity or other diet-related diseases. Two studies have examined
the impact of the opening of a large supermarket in underserved areas in
Leeds and Glasgow, UK (Wrigley et al., 2003; Cummins et al., 2005). The
Leeds study used a pre-post intervention design, with survey interviews
of participants about their shopping and food intake 5 months before and
52
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
7 months after a Tesco supermarket opened in the area. The Glasgow
study used a pre-post study design to assess change in shopping and food
intake behaviors surrounding a new store opening, but it also considered
a comparison area that had similar neighborhood characteristics but did
not have a new store open in the area. The comparison area was added to
determine if any changes in shopping or diet could be due to secular changes
in diet that were not due to a new store opening. Results of both studies
showed that shopping behavior was affected by the openings of new stores—
that is, a significant number of sampled individuals from the neighborhood
switched their shopping to the new store. Both studies also show that
average fruit and vegetable intake increased among surveyed individuals, but
that the average increase was small (just over one-third of a serving). The
average increase in fruit and vegetable intake among those who switched
their main food shopping to the new store was larger, but still under one-half
of a full serving size. The increase in fruit and vegetable intake in Leeds was
statistically significant, but the increase in Glasgow was not. The Glasgow
study, which used a control comparison area, shows that some of the increase
in fruit and vegetable intake among sampled individuals could be due to
overall increased consumption of these foods in both the control and study
area—not due to the better accessibility to the store in the study area. Also
noteworthy is that in both studies, respondents who switched to the new store
reported better self-reported psychological health.
These small store interventions show some promise, however, much of the
research on the effectiveness of these interventions is formative. The studies
are usually on very small and localized samples and often have very short
followup periods from which changes can be observed. Further, there has
also been little evaluative research to determine the cost effectiveness and
sustainability of the changes in the stores and in consumers’ diets.
Many studies have examined the link between store and restaurant access and
BMI and obesity (see Larson et al., 2009, for a recent summary). In general,
these studies find that better access to a supermarket is associated with
53
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
reduced risk of obesity and better access to convenience stores is associated
with increased risk of obesity. Results with respect to restaurants are mixed.
Some show that fast food availability is associated with increased risk of
obesity for adults and children, but others find no association (Larson et al.,
2009).
Currie et al. (2009) examine how school-level obesity rates among ninth
graders in California are related to the distance between the school and fast
food and full-service restaurants. The study also examines weight gain
during pregnancy for women in Michigan, New Jersey, and Texas using
Vital Statistics data and measuring distance from each woman’s home to fast
food and full-service restaurants. Results vary across the two samples. They
find very localized effects in the sample of ninth graders in schools—the
rate of obesity in the school increases 5.2 percent for schools located within
0.10 of a mile of a fast food restaurant (relative to schools that are within
0.25 of a mile). There is no relationship between the school obesity rate and
distances of a quarter or half mile from a fast food restaurant and no effect
of full-service food restaurant availability. For the sample of mothers, the
studies find that living within half a mile of a fast food restaurant increases
the probability of gaining more than 20 kilograms during pregnancy by 2.5
percent. The authors interpret the smaller effects on women as evidence that
they are less constrained by travel than the ninth graders.
54
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The Relationship Between Consumption of Specific
Foods, Obesity, and Diet-Related Diseases
Part of the goal of improving access to healthy and affordable food is
to reduce obesity and diet-related diseases among populations that are
adversely affected by these health conditions. With respect to obesity, it
may be counterintuitive to think that a lack of access to any food is related
to obesity—clearly the problem is too much food. The hypothesized causal
pathway between lack of access and body weight is that some populations
cannot get healthy food options and thus rely on energy-dense options that
may cause weight gain. If healthier food is as available and as inexpensive
as energy-dense food, it is hypothesized that consumers will substitute
away from energy-dense foods to healthier foods and reduce the risk of
obesity. As noted earlier, there is scant causal evidence to either support
or refute this hypothesis. There are, however, a number of studies that
explore the relationship between consumption of specific foods (fruits and
vegetables, whole grains, low-fat milk, and beverages) and obesity and diet-
related diseases. Since these specific foods are often the foods lacking in
underserved areas (with the exception of beverages), this research is relevant
to the question of how lack of access affects obesity and diet-related diseases.
It is hypothesized that because of their high fiber content and, in the case
of whole grains, their improved glycemic control, fruits, vegetables, and
whole grains could increase satiety so that consumers who increased their
intake of these foods would substitute away from other foods that may be
more energy-dense. Such a substitution would either stabilize total caloric
intake or possibly reduce it. With respect to the fat content of milk, it is
hypothesized that consumers who choose low-fat milk instead of milks with
higher fat content would have lower caloric intake and lower BMI.
There is only weak support for these hypothesized relationships in the 31This section draws heavily on a
literature.31 In the case of fruit and vegetable consumption, cross-sectional review of the literature presented by
data show that people who eat more fruits and vegetables have lower BMI. Dr. Richard Mattes, Purdue University,
at the IOM Workshop on the Public
But cross-sectional data cannot distinguish whether consuming more fruits
Health Effects of Food Deserts (Mattes,
and vegetables causes lower BMI since those who eat more fruits and 2009).
vegetables may be more health conscious, more likely to exercise, and more
likely to have lower BMI relative to those who do not. Intervention and
longitudinal studies have shown that increased fruit and vegetable intake may
lead to small decreases in BMI, but some studies even show that increases in
body weight can occur with increased consumption of fruits and vegetables
because total caloric intake increased.
The case of whole grains is similar to that of fruits and vegetables. Relative
to refined grains, greater intake of whole grains provides little or no benefit
for weight management.
The relationship between beverage consumption and obesity has also been
extensively studied. Calories consumed from beverages as a portion of
Americans’ total energy intake have almost doubled in the past 40 years, so
much so that in 2002, 21 percent of total energy intake is from beverages
(Duffey and Popkin, 2007). It is hypothesized that beverages provide
less satiety than solid foods. As a result, increased calories consumed as
beverages may not lead to reductions in calories from solid food and in fact
may be consumed in addition to whatever calories come from solid foods
leading to increased energy intake and weight gain. A summary of research
suggests that beverage consumption is associated with increased energy
intake, weight gain, and BMI. But there is less research from longitudinal
studies or randomized controlled trials to establish the causality of beverage
consumption and weight gain.
foods like fruits and vegetables, whole grains, and low-fat milk alone may term “food swamp” to characterize
not make a dent in the obesity problem. Many of the stores that carry areas with an abundance of less healthy
these nutritious foods at low prices also carry all the less healthy foods food options in contrast to “food
deserts” that lack healthy food options.
and beverages as well. Without also changing the dietary behaviors of
consumers, interventions aimed at increasing access to healthy foods may not
be successful in addressing obesity.
Not all of the relationships between specific healthy foods and diet-related
diseases like diabetes, cardiovascular disease and cancer are well-understood,
but some broad conclusions can be drawn..33 First, plant-based foods like 33This section draws heavily from a
fruits, vegetables, and nuts and whole grains are linked to reduced risk presentation by Dr. Frank Hu, Harvard
of cardiovascular disease. Diets high in saturated fat, trans fat, or refined University, at the IOM Workshop on
the Public Health Effects of Food
sugars are linked with higher risk of diabetes and cardiovascular disease. Deserts.
Sugar-sweetened beverages increase the risk of obesity, diabetes, and
cardiovascular disease. The evidence of how these foods relate to risks of
cancer are not as clear cut and varies across types of cancers. For example,
consumption of nonstarchy vegetables and fruits probably protect against
cancers of the mouth, pharynx, and larynx and of the esophagus and stomach
(see IOM, 2009, for a more thorough discussion).
The link between plant-based foods and whole grains to lower risk of CVD
is relevant to questions about food accessibility. The lack of these foods in
consumers’ diets due to lack of access could plausibly contribute to increased
risk of CVD, especially if other foods high in saturated and trans fat are
relatively more accessible and inexpensive.34 The link between sugar- 34Thisis a hypothesis that could be
sweetened beverages and increased risk of obesity, diabetes, and CVD may empirically tested.
be less of a question about food access since these beverages are almost
56
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
omnipresent—in supermarkets, corner stores, vending machines, and many
other food and nonfood retailers.
Summary
There is clear evidence that the food environment is associated with the kinds
of foods that people eat. But most studies are cross-sectional and cannot
make causal links. A few studies have examined food intake before and after
healthy options for food become available (either within existing stores or
because new stores open). These studies show mixed results. Some show a
small but positive increase in consumption of fruits and vegetables and other
nutritious food, while others show no effect.
Studies that go beyond correlation and try to map out causal relationships
between the food environment and diet and health outcomes are rare. In
order to disentangle these relationships and to ultimately improve the design
of interventions that may reduce the impact of access barriers, improvements
in research are needed. Better models that relate the food environment to
diet and health are needed to disentangle causal relationships and define
tests of which factors may be most important in explaining the relationships
(e.g., availability or price). Experimental studies that can isolate the effects
of changes in the food environment to diet and health outcomes could
help. Taking advantage of natural experiments or quasi-experiments where
naturally occurring comparison groups or areas can be used to uncover causal
pathways would be useful. Longitudinal data that can be used to determine
changes in diet and health over time are also needed to improve what is
known about the relationships between food environment and dietary health.
References
Averett, S., and S. Korenman (1996). “The Economic Reality of the Beauty
Myth,” Journal of Human Resources, 31(2): 304-330.
Chen, S.E., R.J.G.M. Florax, and S.D. Snyder, (2009). “Obesity, Fast Food,
and Grocery Stores: Evidence from Geo-referenced Micro Data,” National
Poverty Center Working Paper, http://www.npc.umich.edu/news/events/
food-access/index.php
Duffey, K.J., and B.M. Popkin (2007). “Shifts in Patterns and Consumption
of Beverages Between 1965 and 2002,” Obesity, 15: 2739-2747.
Hu, F.B. (2009). “Effects of Selected Dietary Factors on CVD and Cancer,
“Presentation at the IOM/NRC Workshop on the Public Health Effects of
Food Deserts, January 26, Washington, DC.
Jeffery, R.W., J. Baxter, M.McGuire, and J. Linde (2006). “Are Fast Food
Restaurants an Environmental Risk Factor for Obesity?” International
Journal of Behavioral Nutrition and Physical Activity, 3(2):
Katz, L.F., J.R. Kling, and J.B. Leibman (2007). “Experimental Analysis of
Neighborhood Effects,” Econometrica 75(1): 83-119.
58
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
National Research Council (2004). Eliminating Health Disparities:
Measurement and Data Needs, M. Ver Ploeg and E. Perrin (eds.), Panel
on DHHS Collection of Race and Ethnicity Data, Committee on National
Statistics, Division of Behavioral and Social Sciences and Education,
Washington, DC: The National Academies Press.
Rose, D., J.N. Bodor, C.M. Swalm, J.C. Rice, T.A. Farley, P.L. Hutchinson
(2009). “Deserts in New Orleans? Illustrations of Urban Food Access and
Implications for Policy”. National Poverty Center Working Paper. http://
www.npc.umich.edu/news/events/food-access/index.php.
Wrigley, N., D. Warm and B.Margetts (2003). “Deprivation, Diet, and Food-
Retail Access: Findings From the Leeds ‘Food Deserts’ Study,” Environment
and Planning A 35: 151-188.
59
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 5
Policymakers are concerned about people with limited access to healthy food
because they believe it may influence food shopping and spending behavior,
the prices of food faced by people in areas with limited access, and the types
of foods purchased and consumed. This chapter examines these economic
consequences of limited access. Food shopping behaviors for participants
35On October 1, 2008, the Food
of the Supplemental Nutrition Assistance Program are summarized.35 The
Stamp Program changed its name to
chapter also considers food spending behavior for SNAP participants with
the Supplemental Nutrition Assistance
different levels of access to supermarkets, examining the types of foods that Program (SNAP). Because all of the
SNAP participants purchase based on their access to supermarkets. Finally, research discussed in this section was
the chapter analyzes data on the price of selected similar foods across conducted prior to the name change,
different food retail outlet types. most program references are to the
Food Stamp Program.
61
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
These broad criteria enable FNS to authorize a wide variety of store types
and sizes in many locations so that participants have a range of food
shopping options. Table 5.1 compares the percentage of authorized retailers
and benefits redeemed by store type in Fiscal Years (FY) 1994 and 2008.
Store types are defined in terms of the dollar value of annual gross sales and
product lines offered. Supermarkets are defined as foodstores that provide a
full range of foods and have $2 million or more in annual gross sales. Large
groceries have annual sales between $500,000 and $2 million, while small
grocery stores have annual sales of less than $500,000.36 Convenience stores 36The 2008 data include a category
provide a more limited range of foods, usually excluding fresh produce. for medium sized stores. For this
Specialty stores primarily sell one or two product lines, such as produce, classification, stores with sales of
meats, or baked goods. Examples of other store types include nonprofit food $1 million to $2 million are large,
$250,000 to $1 million are medium,
buying co-op stores and combination grocery/other stores. and less than $250,000 are small.
Table 5.1
Percentage of Authorized Retailers and SNAP Redemptions by Category FY 1994 Versus FY 2008
Authorized retailers Benefits redeemed
Store type FY 1994 FY 2008 FY 1994 FY 2008
Percent
Supermarkets 15 12 77 47
Superstores na 8 37
Large grocery stores1 2 2
Medium grocery stores 25 6 11 2
Small grocery stores2 9 2
Convenience stores 27 35 4 4
Combination stores3 16 17 3 2
Farmers’ markets <1 <1 <1 <1
All other 17 9 5 2
Total 100 98 100 98
1In FY 1994, data for large grocery stores and supermarkets was combined.
2In FY 1994, data for small and medium grocery stores was combined.
3This category includes stores such as independent drug stores, dollar stores, and general stores.
na =not applicable.
Source: USDA, FNS calculations based on 1994 and 2008 administrative data on SNAP redemptions.
62
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
more noticeable in rural areas, where recipients shopped in relatively larger
population centers. Even in urban areas, however, households traveled
beyond their neighborhood supermarkets to more affluent areas and/or to
other low-income ZIP Codes to access stores offering items of particular
interest, such as fresh fruits and vegetables or ethnic products.
Ohls et al. (1999) reported that nearly 90 percent of each low-income group
used supermarkets as their main foodstore. Even among participants who
reported that they usually did not shop at supermarkets, all but 2 percent
reported that they sometimes used such stores.
More recently, FNS analyzed a national sample of EBT transaction data that
was linked to store and household characteristics (Cole, 2005). Participants
spent most of their food stamp benefits in supermarkets. Supermarkets
accounted for 64 percent of all EBT purchases and 83 percent of the dollar
value of food stamp benefits redeemed. Over 46 percent of food stamp
households shopped exclusively at supermarkets, while less than 6 percent
never shopped in supermarkets. The latter families were concentrated among
households receiving the minimum monthly benefit, $10 or less.
The aggregated redemption data for FY 2008 show that a majority of benefits
are spent in large stores: 87 percent of food stamp benefits were redeemed in
superstores, supermarkets, or large grocery stores. Only 4 percent of benefits
were redeemed in convenience stores, and another 4 percent were redeemed
in small to medium grocery stores.
This analysis uses the same data used by Rose and Richards (2004) to
examine the amount of food from different food groups purchased per week
by SNAP participants. The study by Rose and Richards (2004) is extended
in three ways. First, fruits and vegetables are separated into canned and
noncanned forms. Limited access to a supermarket is hypothesized to exert
greater effects on the purchase of noncanned produce than canned produce
since many smaller grocery stores and convenience stores sell mostly canned,
but not necessarily fresh fruits and vegetables. Additionally, potatoes
and dried beans are treated as a separate vegetable category. The analysis
also examines purchases of milk and other dairy products. Like fresh
produce, milk and dairy products are perishable, so that milk purchases are
hypothesized to be negatively affected by limited access to a supermarket.
64
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Second, survey respondents were also asked whether they did their major
food shopping at a supermarket as well as the frequency at which they
shopped for food. The analysis combines these two variables to develop
three mutually exclusive categories of access to a supermarket: major food
shopping was not at a supermarket, no matter how frequently shopping was
done; major food shopping was at a supermarket but shopping was infrequent
(less than once in 2 weeks); and major food shopping was at a supermarket
and was conducted at least once every 2 weeks. The measure of shopping
frequency is intended to capture difficulty in getting to a store—if stores
are relatively close and the costs (both time and travel costs) are low, then
it is expected that respondents will shop more frequently. But if costs to
getting to a supermarket are high, we would expect respondents to make less
frequent trips to the supermarket.
Third, the extension to the Rose and Richards study accounts for the
censored nature of food expenditures. In a given week, some households
may not make any purchase of the food groups in question. To accommodate
this data issue, the analysis employs the Tobit censored regression model, as
discussed later.
Data
Data for examining supermarket access on food purchases are drawn from
the NFSPS, conducted by Mathematica Policy Research, Inc. for USDA’s
FNS. The NFSPS employed computer-assisted personal interview (CAPI)
methods to collect data on household food purchases among food stamp
recipients between June 1996 and January 1997. Respondents reported their
7-day food use (some households were asked to provide four-day records),
which included data on both the quantities and prices of food used as well
as expenditures on food at home and away from home. This is the only
USDA survey in which household food use (quantity and expenditure) data
were collected since the 1987 88 Nationwide Food Consumption Survey.
Social, demographic, and economic characteristics of households were also
collected.
In total, 1,109 in-person interviews were completed from the SNAP list
frame, and 1,069 households provided complete information on quantity and
expenditure data. After excluding households with missing information, the
final sample totals 860 households.
There are more than 2,000 foods recorded in NFSPS, and they are aggregated
according to the research focus of the project. Insufficient intake of foods
rich in fiber and calcium (such as fruits, vegetables, and dairy products) is
a major dietary deficiency facing Americans, especially the low-income
subpopulation (Lin, 2005). The analysis hypothesizes that households with
65
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
limited access to supermarkets tend to spend proportionally less of their
food budget on perishable foods, such as fresh fruits, fresh vegetables,
and dairy products, than households that shop mainly at supermarkets.
This study focuses on household purchases of five food groups—dairy
products, noncanned fruits, noncanned vegetables, canned fruits, and canned
vegetables.
On any given week, some food stamp households did not purchase a
particular food group. Therefore, a cluster of zero consumption values for
a particular food group is observed in the data—making it necessary to
estimate a censored regression model. Any statistical procedure that does
not account for zero observations produces inconsistent parameter estimates.
Tobin (1958) was the first to propose a censored normal regression model
(Tobit model) to deal with censored data in regression, which can be
expressed as below
Results
66
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 5.2
Descriptive statistics of the NFSPS respondents
Supermarket shopping
Frequent Infrequent
Total shopper shopper Not shop at
Sample 860 578 218 64
Average purchase Pounds per week
Noncanned vegetables 3.73 3.93 3.32 3.30
Canned vegetables 2.06 2.05 2.14 1.79
Potatoes and beans 2.95 3.05 2.92 2.10
Noncanned fruits 5.92 6.23 5.38 4.94
Canned fruits 1.59 1.66 1.51 1.19
Milk and diary products 12.29 12.71 12.31 8.52
Proportion consuming Percent
Noncanned vegetables 85 86 83 78
Canned vegetables 69 67 73 70
Potatoes and beans 82 83 81 81
Noncanned fruits 83 85 80 78
Canned fruits 42 42 43 36
Milk and diary products 97 97 98 94
Mean values of explanatory variables:
Per capita income ($/month) 305
Meal number (# meals prepared from
food purchase) 48
Percent
Four day (sample report only 4-day purchase) 8
Asian (sample) 1
Black (sample) 39
Native American (sample) 1
Hispanic (sample) 13
White (sample) 46
Single-headed (sample with one head) 34
Child (sample with children under 18) 56
Elderly (sample with senior > 60) 27
Less than high school (head without HS diploma) 46
High school (head with HS diploma) 38
College (head attended college) 16
Rural (living in rural area) 15
West 38
South 20
Midwest 24
Northeast 18
Spring 24
Summer 24
Fall 29
Winter 23
Source: USDA, ERS calculations based on 1996-97 National Food Stamp Program Survey data.
67
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
matter how frequently shopping was done; major food shopping was at a
supermarket but shopping was infrequent (less than once in 2 weeks); and
major food shopping was at a supermarket and was conducted at least once
every 2 weeks (the reference group). The results are shown in table 5.3.
68
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 5.3
Tobit results
Vegetables Fruits
Non Potatoes Non
canned Canned Beans canned Canned Milk
Meal number 0.06 *** 0.02 *** 0.04 *** 0.07 *** 0.04 *** 0.13 ***
Literature review
Many studies have examined price disparities across income class, store
format, and accessibility (Andreyeva et al., 2008; Block and Kouba, 2006;
Broda et al., 2009; Chung and Myers, 1999; Hayes, 2000; Hendrickson et
al., 2006; Latham and Moffat, 2007; Talukdar, 2008). A limitation of these
studies is their use of observed prices in a regional setting rather than actual
prices paid on a national level. Kaufman et al. (1997) provides a review of
literature on food price disparity dating back to the 1960s and identifies the
complexities of undertaking such research.
69
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Andreyeva et al. (2008) replicated a 1971 study of food availability and
price in New Haven, Connecticut. Their findings show improvement in
availability and price since 1971. Findings indicate differences across
store types—grocery stores were approximately 4 percent cheaper than
convenience stores for a basket of goods. The study also found that high-
income areas faced higher prices than low-income areas.
Block and Kuoba (2006) compared prices for a market basket of goods in
different types of stores in the Austin and Oak Park sections of Chicago.
Austin is a lower-middle-class African-American community that borders
Oak Park, an upper-middle-income suburb. They find mixed results.
Discount supermarkets showed the lowest prices. Independent grocery stores
had higher prices for packaged goods than chain supermarkets, but lower
prices for fresh items.
Broda et. al (2009) analyzed actual consumer purchases and found that poor
households pay less for food items they purchase than households with
higher incomes—a 10-percent increase in income roughly induces a modest
0.1-percent increase in prices paid per food item. They also found that poor
households tend to shop more frequently at discount stores and supercenters.
Even after controlling for household characteristics and product fixed effects,
the study found that poorer households pay a lower price even in stores of the
same retail chain.
Chung and Myers (1999) conducted a survey in the Twin Cities metropolitan
area to determine how store type (nonchain/convenience store versus chain/
supermarket) and neighborhood quality (measured by percent of households
under the poverty level within a zip code) affects price of a food market
basket. They conclude that store type is more important in driving price
disparities than the geographic location of a household—the premium for
shopping at a convenience and/or nonchain store outweighs the premium for
shopping in a poor neighborhood. Limitations to their methodology include
the way missing price values were treated. When price for a selected item is
missing, the least expensive brand/size product (e.g., an in-store brand in its
largest package size) was used. When a selected item was not available in
the store, the sample mean price was used. Additionally, the use of regional
data based on “sticker prices” (those listed on the shelf) as opposed to
actual transaction prices at the national level do not control for promotional
purchases (e.g., on-sale and coupon use).
Hendrickson et al. (2006) studied prices of selected Thrifty Food Plan foods
in four Minnesota communities with higher than average poverty (two rural
and two urban). The study examined prices offered in grocery stores in
70
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
these communities for the TFP foods and compared their prices with those
of the TFP Market Basket Price (MBP). If a food was found in the grocery
store, the price of the lowest price version of the food (price per pound) was
recorded. The study found that in the two urban areas, 6 and 9 out of the
19 foods studied were more expensive than the TFP MBP. In the two rural
areas, 2 and 4 of the 19 foods studied were more expensive than the TFP
MBP. Over all of these communities, the prices per pound of fresh produce
were equal or less expensive than the TFP MBP price. Although this study
uses the lowest price per pound product in the store for a selected food, it
still only uses the available price instead of the actual paid price. Further,
the TFP MBP is a national price average so it is not clear if the prices in the
Minnesota communities studied are different from the TFP MBP because
prices in the neighborhoods are different or because the State or region has
different prices.
Talukdar (2008) investigated prices faced by the poor for both food
and nonfood items in Buffalo, New York, and surrounding suburban
neighborhoods. The study found that the inner-city neighborhoods
experience a weakened competitive market leading to cost-inefficient
“corner stores” which have a 6-7 percent premium over regional or national
chain grocery stores. Even after controlling for economies of scale and
competitive environments, prices were 2 to 5 percent higher in the poorest
neighborhoods.
71
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 5.4
Descriptive statistics of variables used in price analysis
Fluid milk RTE cereal Bread
Convenience All Convenience All Convenience All
Variables Definition store other store other store other
Unit value (expenditure net of any pro-
Price motions divided by the corresponding 2.47 2.59 16.17 16.71 8.67 9.03
quantity), cents per ounce
The ratio of household income over the
Income federal poverty level; where income is 3.57 3.83 3.59 3.73 3.18 3.73
the midpoint of the income class
72
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Data
The data source for investigating price differentials by store type is the 2006
Nielsen Homescan panel data. The panelists constitute a random sample that
is representative of the U.S. population and provides purchase information
of food items for at-home consumption. Each household is supplied with
a scanner device that the panelist uses at home to record grocery items
purchased at all retail outlets. The household either scans the Uniform
Product Code (UPC) or a designated code for random-weight purchases for
each food item. Each purchase records the date, the quantity purchased,
expenditures for that quantity, promotional information including whether or
not the item is on sale, and detailed product characteristics.
Total enrollment in the Homescan panel for 2006 was over 37,000
households, but to avoid would-be data problems resulting from incomplete
reporting, only those households that reported purchases for at least 10
months were included. Panelists report total expenditures and the quantity
of food purchased. Prices are derived as unit values – the ratio of reported
expenditures, net of any promotional and sale discounts, to the reported
quantities for each purchase record.
73
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Table 5.5
Hedonic results
Fluid milk RTE cereal Bread
Standard Standard Standard
Variable Coefficient error Coefficient error Coefficient error
Constant 2.419*** 0.025 22.879*** 0.220 6.338*** 0.137
74
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
where Pit is the price paid by the i-th household in time t; MKTit represents
a set of market factors such as income (a measure of neighborhood store
and product quality), type of store, promotional offering, season, region,
and urbanicity of purchase; PROit represents product attributes; and eit is
the error term. Interaction terms between convenience store purchases and
product attributes, as well as income, allow for the testing of additional price
differentiations observed in convenience stores.
Results
Milk, RTE cereals, and bread are three of the most frequently purchased
items at both convenience stores and grocery stores by Homescan panelists.
The hedonic model is specified in linear functional form so that estimated
coefficients represent price premiums or discounts. The hedonic results
are summarized in table 5.5. The R-squared is 51 percent for milk, 16
percent for RTE cereals, and 27 percent for breads. These goodness-of-fit
measures are quite high for cross-sectional studies, implying that the data
fit the model reasonably well. “Grocery” is treated as the reference store
in the model so that the estimated coefficient for “Convenience” measures
the price difference between the two types of stores. Specifically, a positive
(negative) coefficient for “Convenience” store indicates that consumers pay
a higher (lower) price at a convenience store than at a grocery store. Income
is included to capture store, product, and neighborhood quality attributes
unobserved in the data. In particular, a household’s income is expected to be
associated with the quality of shopping venue and product offering, which, in
turn, would be reflected in the price paid.
Fluid milk
TThe price of milk is expressed as cents per fluid ounce (128 ounces in a
gallon). The estimated constant term suggests an average price of 2.42 cents
per ounce (or $3.10 per gallon) for whole milk in a gallon container sold
in a grocery store in the Southern United States in the winter. The results
suggest that the same milk is sold at a price 0.13 cents per ounce higher at
a convenience store, or about 5 percent above the grocery store price. As
expected, consumers pay a lower price at discount stores, such as Wal-Mart,
at an average of 0.2 cents per ounce below the grocery store price. The
estimated coefficient for “Income” (a measure of store and product quality)
is significant but small, indicating that the price of milk in a grocery store
increases only slightly with income. Additional variation in price associated
with income in a convenience store (measured by the interaction between
“Income” and “Convenience”) is found to be insignificant.
75
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
oz (2.42 + 0.98) at grocery stores, but there was actually a price discount (0.3
cents/oz) for such milk sold at convenience stores.
RTE cereals
Like milk, RTE cereal in a grocery store increases in price with the affluence
of the neighborhood, as measured by household income. However, the
interaction term between income and convenience store is negative and
significant (-0.29) and must be interpreted in conjunction with the income
variable (0.33). This suggests less price variation in convenience stores
relative to store and product quality, as measured by income.
Bread
Bread in its largest 24-ounce size was priced at an average of 6.34 cents/
oz at grocery stores and about 0.62 cents (about 10 percent) more at
convenience stores. When bread was purchased on sale or with a coupon,
the average price dropped to 4.67 cents/oz (26 percent of the regular price).
Prices of bread also varied greatly by package size, and unlike whole-grain
RTE cereals, whole-grain bread commanded a large and significant price
premium, averaging 1.59 cents/oz (or 25 percent of the price of non-whole-
grain breads). Like the case of milk, price differentials between convenience
store bread and grocery store bread varied by package size. The estimated
76
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
coefficients indicate that medium- and large-sized bread (constituting 76
percent of convenience store purchases) was priced higher in a convenience
store than in a grocery store. Yet, bread sold in a small package size was
priced lower in a convenience store.
Discussion
The analysis of price variation for similar goods across different store types
shows that prices are higher, on average, at convenience stores than they
are at grocery stores, and this finding is confirmed in the literature (Broda
et al., 2009). Relatively easy access to convenience stores and smaller food
retailers in some neighborhoods may lead to higher prices for food for people
who live in those neighborhoods. But this argument assumes that people
who live in these neighborhoods do not shop at large stores or search for sale
items. Broda et al. (2009) use 2005 Nielsen Homescan data to address the
effects of access to different types of stores on overall shopping expenditures
and prices paid by lower income consumers. The analysis does not directly
consider access to stores or shopping patterns but instead focuses on
differences across household income levels.
Broda et al. (2009) show that across all income levels, 52-57 percent of all
food purchases are made at grocery stores. Spending on food at convenience
stores is a very small portion of shoppers’ food budgets, even for those at
the lowest income levels. Those with the lowest incomes (from $5,000 to
$11,999) spend 2 to 3 percent of their total food expenditures in convenience
stores, while the highest income consumers (annual income over $100,000)
spend only 0.7 percent of their total food expenditures at convenience stores.
Low-and middle-income households (incomes between $5,000 and $49,999)
spend 20-22 percent of their food dollars at supercenters, where prices are
lower. Households with incomes over $70,000 spend 13-17 percent of their
food dollars at these types of stores. Clearly, lower income consumers shop
at outlets offering lower prices.
77
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Examining prices paid for specific goods (through the Universal Product
Code of the good purchased) across household income level, Broda et al.
(2009) find that while households with incomes below $8,000 per year may
pay 0.5 to 1.3 percent more for their groceries than households earning
slightly more, those earning between $8,000 and $30,000 pay the lowest
prices for groceries. Households with the highest incomes, with earnings
over $100,000, pay the greatest amount for groceries, between 2-3 percent
higher than poorer households. These results suggest that the poor do not
pay higher prices for food.
One caveat to these results is that the Broda et al. (2009) study does not
directly address access issues as it is only approximated by household income
level. The study does not have information about the costs consumers face
to get to food retail outlets, which could be greater for those who live in areas
with limited access.
Summary
For SNAP to meet its mission, it is essential that participants have access to
foodstores offering good quality, variety, and reasonable prices without the
participants encountering undue burden. As a whole, participants live close
to an authorized retailer, which is often a supermarket or large grocery store.
More than 90 percent of participating households spend at least some of their
benefits in a supermarket, and close to 90 percent of all benefits are redeemed
in supermarkets or large grocery stores. Food stamp recipients reported
being largely satisfied with the stores in which they shop most frequently.
Studies of SNAP participant access and shopping patterns reviewed here
indicate that most SNAP participants have access to supermarkets and large
grocery stores. These findings, however, do not eliminate the possibility that
access may be challenging for some participants and nonparticipants in some
places.
The analysis of food purchases suggests that SNAP participants who did not
shop mainly at a supermarket purchased less noncanned fruit, noncanned
vegetables, and milk than SNAP participants who shopped frequently at a
supermarket. SNAP participants who did not shop at a supermarket also
purchased less canned fruits and vegetables than others, but the differences
are not statistically significant. Overall, the results suggest that lack of access
to a supermarket is associated with lower levels of expenditures on some
foods that are important for healthy diets.
It appears that only two studies have used longitudinal data to try to
determine differences in purchasing behavior over time as the degree of
access changed (Wrigley et al., 2003; Cummins et al., 2005). These studies
were conducted in the UK, and both considered changes in shopping
78
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
behavior and food intake when a new supermarket was opened in an
underserved area.41 Results from Wrigley et al. (2003) showed that a sizable 41Findings on food intake changes
number of residents in both studies switched their main shopping source to from these studies are summarized in
the new store, more walked to and from the store than before, and fewer took Chapter 4.
buses, taxis, or someone else’s car due to the change in access. Cummins et
al. (2005) also found that when access improved, many shoppers switched
to a new store, more walked to the new store, and fewer traveled to the store
using someone else’s car.
This study’s demand analysis does not consider the relative food prices
of these groups of foods that survey participants face. Those with limited
access could face higher prices of noncanned fruits and vegetables, which
could impact their purchase behavior. The analysis attempted to incorporate
prices and estimate a demand system, but the results were not reasonable,
possibly due to the rather limited size of the sample.
The results for estimating price differentials between grocery stores and
convenience stores are consistent with a priori expectations that consumers
pay more for food at a convenience store than at a grocery store. Likewise,
neighborhood quality, as measured by household income, affects prices
positively, but to a lesser magnitude for those purchases at a convenience
store. An important finding, however, is that on-sale purchases and
coupon use are frequently reported at convenience stores. The price
discount afforded by on-sale purchases or coupon use is quite large and
can compensate for the higher price registered at convenience stores.
Frugal shopping habits can effectively overcome the price disadvantage at
convenience stores. Of course, frugal shopping habits can also be effective
when shopping at grocery stores.
Analysis by Broda et al. (2009) is consistent with the idea that frugal
shopping habits can overcome high prices. This study finds small differences
in expenditures at different food retailers across income levels. It also
finds that, in general, the poor do not pay more for food. The study finds
that households earning between $8,000 and $30,000 per year pay the least
79
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
for groceries, while the poorest consumers, those with household incomes
below $8,000, pay between 0.5 to 1.3 percent more for their groceries than
households earning slightly more.
References
Andreyeva, T., D.M. Blumenthal, M.B. Schwartz, M.W. Long, and K.D.
Brownell (2008). “Availability and Prices of Foods Across Stores and
Neighborhoods: The Case of New Haven, Connecticut,” Health Affairs,
September/October, 27(5): 1381-1388.
Broda, C.E. Leibtag, and D.E. Weinstein (2009). “The Role of Prices in
Measuring the Poor’s Living Standard,” Journal of Economic Perspectives,
23 (2): Spring, 000-000.
Chung, C., and S.L. Myers (1999). “Do the Poor Pay Moor for Food? An
Analysis of Grocery Store Availability and Food Price Disparities,” Journal
of Political Economy, 91 (3): 480-93.
Cole, N., and Lee, E. (2005) Analysis of EBT Redemption Patterns: Methods
and Detailed Tables, Report submitted to U.S. Department of Agriculture,
Food and Nutrition Service, by Abt Associates, Inc., November.
80
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Cummins, S., A. Findlay, M. Petticrew, and L. Sparks (2005). “Healthy
Cities: The Impact of Food Retail-Led Regeneration on Food Access, Choice
and Retail Structure,” Built Environment, Vol. 31(4): 288-301.
Hayes, L.R. (2000). “Are Prices Higher for the Poor in New York City?”
Journal of Consumer Policy, 23 (2): 127-52.
Kaufman, P.R., J.M. MacDonald, S.M. Lutz, and D.M. Smallwood (1997).
Do the Poor Pay More for Food? Item Selection and Price Differences Affect
Low-Income Household Food Costs, Agricultural Economic Report No. 759,
U.S. Department of Agriculture, Economic Research Service.
81
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Pearson T, J. Russell, M.J. Campbell, and M.E. Barker (2005). “Do “Food
Deserts” Influence Fruit and Vegetable Consumption?--A Cross-Sectional
Study,” Appetite 45(2): 195-7.
Rose, Donald and Rickelle Richards (2004). “Food Store Access and
Household Fruit and Vegetable Use Among Participants in the US Food
Stamp Program,” Public Health Nutrition 7(8): 1081-1088.
Talukdar, D. (2008). “Cost of Being Poor: Retail Price and Consumer Price
Search Differences Across Inner-City and Suburban Neighborhoods,”
Journal of Consumer Research, 35 (3): 457-71.
82
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 6
There has been little consideration of the economics behind the variation
in food access across areas. It is important to understand the economic
conditions that may contribute to food deserts—that is, the costs that food
retail businesses face and the choices available to consumers who want to
buy foods. This chapter outlines an economic framework for considering
food access and why some areas may have limited access. This framework
considers the consumer and demand factors, business and supply factors,
and the market conditions that interact to create differences in the food retail
environment across areas and subpopulations. The chapter then provides
a broad overview of the history of supermarket development that captures
how food retailers have responded to different demand, supply, and market
conditions. ERS, through the National Poverty Center at the University of
Michigan, commissioned a paper to consider the economic framework of food
deserts (Bitler and Haider, 2009). This chapter draws heavily from that paper.
Travel costs and time costs of acquiring food as well as the time costs of
preparing foods are also likely to affect demand for particular foods. The
convenience of eating restaurant food or a prepared meal versus eating at
home may be an important part of demand for food. Even for foods prepared
at home, there may be relatively greater time costs than those for prepared
foods or takeout foods. Consumers may value the convenience of a fast food
or prepared meal more because it does not require spending much time to
prepare.
Demand for some foods could be affected if individuals do not know which
foods are healthy or unhealthy or if individuals do not know how to use or
prepare some foods. Tastes for different foods, or preferences as economists
42This is a bit of a chicken and egg
call them, may also drive demand and store location. For example,
problem—people of a specific ethnic
foodstores that sell many Asian or Hispanic food products are often located decent may choose to live in an area
in areas where there are high concentrations of people of Asian or Hispanic where they have relatively easy access
descent, presumably because demand for those foods is high in those areas.42 to the foods they wish to consume. We
discuss this further later in this chapter.
83
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Consumers choose where to live and thus, choose the food environment
available to them. Clearly, the choices of the poor, especially those living in
extreme poverty, are constrained by their budgets. But even among the poor,
there are choices in where to reside. This choice is an aspect of food access
that is not well considered in the literature. Not accounting for residential
choice limits the ability to assess whether limited access to affordable and
nutritious food affects food shopping, diet, and health outcomes. If two
people are equal in all other manners, the one who places a high value on
access to a large supermarket is probably more likely to live in an area
where those things are available relative to another who values food-related
amenities less and perhaps other amenities more.
Food retailers that face higher fixed costs will either need to charge a higher
price for goods sold or limit the range of products sold (Bitler and Haider,
2009). Fixed costs faced by food retailers probably vary greatly by the
type of area. In dense urban areas, land prices may be higher and zoning
requirements of local governments may be more cumbersome and costly
to meet relative to less dense suburban and rural areas. Consequently, it is
likely that the fixed costs in urban areas are greater. Further, the volume
of sales in a store could affect how much fixed costs matter to firms. For
example, stores that serve a higher volume of consumers (either in the
number of consumers or in terms of the amount they purchase) will be able
to spread fixed costs over more people and prices may be lower than in
stores that serve lower volumes of consumers. This could explain why some
rural, less populated areas do not have stores or why some foods may not be
stocked in low-volume stores, especially foods that perish.
If fixed costs drive store location decisions or the types of products sold,
then the fixed costs should be higher in low-income areas if these areas are
underserved. Within urban areas, for example, low-income areas are likely
to have lower land prices than high-income areas. In this case, it would be
surprising to find that land costs leave some poor areas without access. It is
possible that zoning requirements or security costs are higher in poor areas
and thus more important for stores to locate in those areas (regardless of size)
(Bitler and Haider, 2009).
One major variable cost for operating a foodstore is the cost of hiring
workers. Poor areas are often areas with relatively lower wages, which
should reduce the costs to operate a store, all else equal. King et al., (2004)
find that stores serving low-income shoppers (stores with greater shares
of revenue from SNAP participants) have significantly lower payroll costs
as a percentage of sales than stores that do not serve as many low-income
shoppers.
King et al., (2004) find that operating cost structures of stores with higher
portions of total revenues from SNAP redemptions are different than those
of other stores. For example, the stores had lower labor costs but also lower
sales margins, and they were more likely to be supplied by wholesalers
than to be part of a large chain with its own supply chain. However, overall
operating costs for these stores were not different than other stores.43 43Fixed costs were not considered in
this study.
Economies of scale, economies of scope, and economies of agglomeration
may also explain why product availability is differentially concentrated
across areas or stores (Bitler and Haider, 2009). Economies of scale, which
is when the costs of operating a store decrease as store size increases, and
economies of scope, which is when the costs decrease as more product
variety increases, suggests that larger stores that offer greater variety can do
so and offer lower prices. Both factors may account for the ability of larger
stores to survive more easily than smaller stores. For example, if wholesale
costs to stock store shelves are lower for larger stores because they can buy
in bulk and buy a greater variety of products more cheaply, larger stores may
be able to lower prices and smaller stores may not be able to compete. This
does not explain, however, why larger stores do not locate in low-income
areas.
The Market
The market is where consumers and suppliers meet, their interactions
determining the prices paid and the products bought and sold. The most
85
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
basic economic models assume that markets are perfectly competitive and
that food retailers and consumers do not have control over price. In such a
case, it is possible that if demand is sufficiently low, retailers would not be
willing to supply products because they could not sell them or could only sell
them at higher prices.
It is possible that food retailers actually have some market power, especially
in settings where there are few competitors. If so, then retailers would
have an incentive to increase price and restrict quantities to increase profits.
Supply-side conditions, such as high fixed costs or economies of scale, could
lead to food retailers having market power (Bitler and Haider, 2009).
It is possible that the supply-and-demand conditions are such that the market
does not operate efficiently—that some intervention could be implemented
that would improve circumstances for both retailers and consumers. A
market failure could arise, if for example, there is poor information on the
part of suppliers about the foods that consumers in an area will demand (for
a given price). For example, if the models that supermarkets use to predict
sales in an area do not adequately capture potential sales in low-income areas
as some have suggested, better information on the potential sales could lead
to more stores or products offered in those areas (Drake, 2009; Policy Link,
2007).
Hard discount stores, which target bargain and low-income shoppers, have
also developed. These stores are typically smaller than other supermarkets
(typically between 10,000 to 14,000 square feet, compared with an overall
average store size of 48,000 square feet), offer more private label brands and
fewer national brands, and offer fewer product size options. Many of these
stores operate in underserved or low-income areas. Examples include:
87
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
• Save-a-Lot (a subsidiary of SuperValu, Inc.) operates 1,200 stores,
including 4 in Prince Georges County, Maryland, and 1 store in Coahoma
County, Mississippi, a persistently poor county in the Mississippi Delta
region.
• ALDI stores (a German discount retailer which opened its first U.S.
stores in Southeastern Iowa) have 1,000 stores in 29 States.
Food retail has also expanded to retailers that have not focused on food
in the past. This “channel blurring” has occurred among pharmacies,
which typically carry snack and convenience foods and increasingly carry
refrigerated and frozen items, and among dollar stores (such as Family Dollar
and Dollar Store) (Martinez and Kaufman, 2008; Sharkey and Horel, 2009).
Warehouse clubs (e.g., Costco and Sam’s Club), which offer a limited
variety of food products and larger or bulk sizes, have also expanded into
food markets, serving primarily small businesses and middle-upper-income
consumers (Martinez and Kaufman, 2008).
Finally, grocery stores and supermarkets also compete with restaurants and
other sources of food away from home. Almost half of all food spending
in the United States is for food away from home (48.9 percent in 2006)
(Martinez and Kaufman, 2008).
Given these competitive forces, why is it that some areas lack access to a
supermarket or large grocery store? The range of reasons offered include
some that are identified in this chapter and others that have been summarized
elsewhere..44 Many apply specifically to urban areas, but some apply to rural 44See Drake, 2009; Policy Link,
areas as well. Examples include the following: 2008; Food Marketing Institute, 1998.
• Higher construction and operating (e.g., rent, real estate taxes, security
costs) costs in urban areas
• Local politics, where officials and groups may have competing goals for
development
88
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
underserved areas attractive for food retail development (Food Marketing
Institute, 1998).
Summary
Economic theory can help explain why some areas may have limited access.
Consumer behavior and demand-side issues, such as lack of knowledge of
the nutritional benefits of foods or food retailer behavior, and supply-side
issues, such as different fixed costs for developing a store in an underserved
area, may explain variation across areas in the types of foods offered and
the stores that offer them. Further research is needed to determine which
factors may be driving differences in access. This research could explore
how differences in fixed and operating costs of different areas may account
for variation in access to different types of stores or different products
within stores. Research could also consider how consumer knowledge and
preferences and the time cost tradeoffs affect consumer decisions of which
foods to eat and whether to make or to buy prepared foods or restaurant
meals. Research on price variation at the local level and spatial demand
models could also be used to help determine which factors contribute to
differences in access to food retailers.
References
Bitler, Marianne, and Steven J. Haider (2009). “An Economic View of Food
Deserts in the United States,” National Poverty Center Working Paper. http://
www.npc.umich.edu/news/events/food-access/index.php (Accessed 3/30/09).
King, Robert P., Ephraim S. Leibtag, and Ajay S. Behl (2004). “Supermarket
Characteristics and Operating Costs in Low-Income Areas,” U.S. Department
of Agriculture, Economic Research Service, Agricultural Economic Report
No. 839.
89
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Policy Link, and LISC Bay Area (2007). Grocery Store Attraction
Strategies, http://www.policylink.org/documents/groceryattraction_final.pdf
Rose, D., J.N. Bodor, C.M. Swalm, J.C. Rice, T.A. Farley, and P.L.
Hutchinson (2009). “Deserts in New Orleans? Illustrations of Urban Food
Access and Implications for Policy,” National Poverty Center Working
Paper, http://www.npc.umich.edu/news/events/food-access/index.php.
Tedlow, R.S. (1990). New and Improved: The Story of Mass Marketing in
America, Cambridge, MA: Harvard Business School Press.
Turner, M.A., F. Frieberg, E. Godfrey, C. Herbig, D.K. Levy, and R.R. Smith
(2002a). All Other Things Being Equal: A Paired Testing Study of Mortgage
Lending Institutions, U.S. Department of Housing and Urban Development.
Wyly, E.K., and S.R. Holloway (1999). “The New Color of Money:
Neighborhood Lending Patterns in Atlanta Revisited,” Housing Facts and
Findings 1(2): 1-11.
90
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 7
Legislative authority
Food is a very basic need. Seeing this need, community leaders all over the
country have begun to organize their communities around food access. This
is especially true in low-income communities where food sources often are
most scarce. The primary goals of the CFPCGP are to:
91
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
• Meet the food needs of low-income individuals;
CFPCGP was funded for $1 million in 1996, $2.5 million for 1997-2001, and
$5 million beginning in 2002 (current legislation). Since 1996, 290 projects
have been funded in 48 States. Currently, three types of projects are funded
each year, each type for 1 to 3 years:
Because the community food projects are designed to solve local problems,
they vary considerably based on local needs. Their specificity is both a major
strength and a challenge for evaluating them at a national level—and for
describing how they contribute to addressing issues within areas with limited
access to affordable and nutritious food.
The following are examples of the types of issues that are addressed within
CFP. These were extracted by Pothukuchi (2007) in a study of projects’
self-reported evaluation of performance on goals and objectives. The study
reflects summary data of grantee reports from 42 CFPs funded between 1999
and 2003.
92
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
neighborhoods, through food assistance programs, backyard and
community gardens, grocery stores, farmers’ markets, Community
Supported Agriculture shares, food buying clubs, and other resources.
• Local food marketing: Increase local markets for small- and family-
scale farms, including direct marketing and purchases by local
institutions and businesses.
Table 7.1
Community Food Projects Competitive Grant Program Activities, 2006
Percent
93
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
• Farm worker conditions: Promote safe and fair working conditions
for farmers, farm workers, and other food workers, such as those in
processing plants and wholesale and retail operations.
• Food heritages: Honor and celebrate diverse food cultures and traditions
in the community.
One way to see how these issues are addressed within CFP is to look at the
national level output reports. The CFPCGP works with grantees to develop
and implement a program-wide evaluation. Currently, grantees voluntarily
report common activities (Common Output Tracking Form) across their
projects. The latest analysis from 2006 enumerated 19 project-type activities
reported by 51 of a possible 65 projects (table 7.1). The projects are
comprehensive and have many component activities. The percentages in the
table refer to the share of projects that report having managed or participated
in the activity. It is hoped that the combination of these activities leads to
systems change within a community and that the multi-pronged approach to
getting more food into the community increases food security for the long
term. The net sum of these activities is to reduce the impact of limited access
to food. Here are some examples of how the combinations of different
activities have been implemented:
94
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Examples of Community Food Projects by Type of Project
Food Assessments
Fresno Fresh Access Project, Fresno Metropolitan Ministry, Fresno, CA,
www.fresnometroministry.org
Community Tradition, Foods, and Future, Legacy Cultural Learning Center,
Muskogee, OK, www.legacycultural.org/artwellness
Youth Programs
Healthy Harvests Initiative: Building Boston’s Food Security from the Ground Up,
The Food Project, Lincoln, MA, www.thefoodproject.org
Food for Life, Youth Farm and Market Project, Minneapolis, MN,
www.youthfarm.net
Urban Agriculture
Integrated Development Through Urban Agriculture, Nuestras Raices, Holyoke,
MA, www.nuestras-raices.org
Neighborhood Urban Agriculture Coalition, Greensgrow, Philadelphia, PA,
www.greensgrow.org
Rural Producers
Tribal Fish Market Connection Project, Ecotrust and Columbia River Inter-Tribal
Fish Commission, Portland, OR, www.ecotrust.org
Patchwork Family Farms: Value-added Processing for Community Food
Security, Missouri Rural Crisis Center, Columbia, MO,
www.inmotionmagazine.com/rural.html
Community Gardens
Urban Detroit Agriculture and Education Project, Capuchin Soup Kitchen,
Detroit, MI, www.cskdetroit.org.
95
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
at risk for food insecurity. The Hmong benefit from small business
and enterprise mentoring that allows them to develop farm- and
food-related micro-businesses. Self-sufficiency and self-help
are putting people on the road to food security.” (Healthy Food
Healthy Communities, 2006).
There are many types of CFPs, including food assessments, food policy
councils, youth projects, entrepreneurial food and agriculture activities, urban
agriculture, rural producers, and community garden projects. This section
describes the general goals and structures of these types of CFPs (see box,
“Examples of Community Food Projects by Type of Project”). Detailed
descriptions of these projects are given in appendix D.
Food assessments
Food policy councils (FPCs) may take many forms but typically are
commissioned by State or local government, with participation from
diverse stakeholders engaged in food and agriculture. They foster a more
comprehensive and coordinated approach to addressing food system issues
by bringing together stakeholders to develop and implement solutions. FPCs
play various roles, including educating officials and the public, making
policy recommendations, improving coordination between existing programs,
and starting new programs. FPCs help identify ways that governments can
mobilize existing resources or shape policy to improve the food system
(Winne, 2008).
FPCs have sometimes been the focus of applicants who see the need to
address policy issues early on in their community food work. Other grantees
have found that they need an FPC to address barriers confronted during
the operation of their project. Projects have often run into local ordinances
or legislative barriers and have formed formal and informal policy groups
to deal with these issues. For example, a project in Washington State
established a market garden on public land only to find that legislation
disallowed marketing from the land. The amendments to the CFPCGP
legislation in 2002 put an added emphasis on funding planning and
policy activities within the program. Examples of Food Policy Councils
projects funded by CFPCGP include a food policy council in New Orleans,
96
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Louisiana, and a nationwide policy project funded to educate State legislators
about food systems issues and CFP grantees about policy issues.
Youth projects
Some CFPs are specifically targeted toward youth. The goals of these
projects are often educational—providing agricultural, nutrition, meal
preparation, and culinary training to young people. A basic idea of these
projects is to strengthen the connections between youth and the foods
they eat. Two examples of youth-focused CFPs are the Healthy Harvests
Initiative and the Food for Life program.
The goals of these types of projects are to build the capacity of communities
to develop and promote their own agricultural activities. The projects often
involve entrepreneurial training, technical assistance in expanding local
markets, developing and producing value-added products or infrastructure
development. Examples of projects that have served rural and low-income
communities include the Appalachian Center for Economic Networks
in Athens, OH, and Building an Integrated Sustainable Food System in
Abington, Virginia.
These types of projects offer support for developing commercial gardens and
food production for sale or for home consumption in underserved urban or
rural areas. Funds are used for developing capacity to grow, sell, and market
food for commercial enterprises or to grow and prepare food for at-home
consumption. The projects cover a wide range of areas and communities.
Lessons Learned
Annually, national experts are selected as peer reviewers for the CFPCGP.
Reviewers select the most meritorious proposals through a competitive
process for funding. The reviewers are people who run community food
projects, work in academia, or do similar work and know what works in
communities. Through the years, lessons have been learned about what tends
to work in communities and what does not. These lessons include project
development issues, challenges faced by CFPs, and general lessons about
community food security work. Some of the lessons learned include the
following:
Project Development
97
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
• Three years is a very short time to develop a project that is going to be
sustainable and economically viable. Often, more than three years is
needed.
Challenges
Programmatic lessons
98
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
• Projects employing youth in community food projects are finding that
they are successfully engaging the future leaders in the community food
security movement.
• Corner stores are seen as a potential tool for improving the diets of
low-income residents of communities with limited food access. Corner
stores tend to not have the resources to make the transition to selling
more fresh foods. One project found that corner stores could not afford
the refrigeration necessary, their regular suppliers did not handle fresh
produce, and they lacked the expertise to handle fresh produce.
99
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
References
Abi-nader, J. (2007). “Community Food Security Coalition and National
Research Council Common Output Tracking Form, Fiscal Year 2006
Results” (Internal, unpublished document).
FNS-315 (2000). The National Nutrition Safety Net. Tools for Community
Food Security, U.S. Department of Agriculture, Food and Nutrition Service,
January.
Hamm, M.W. and A.C. Bellows (2003). “Community Food Security and
Nutrition Educators”. Journal of Nutrition Education and Behavior, 35(10):
37-43.
Jessup, E., J. Hofer, P. Moua (2005). The Fresno Metro Ministry Fresno
Fresh Access Community Food Assessment was staffed by the Hunger and
Nutrition Project: Fresno Metro Ministry, www.fresnometroministry.org
100
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
World Hunger Year (2009). “Community Food Projects Database,” Food
Security Learning Center, http://www.yhunger.org/programs/fslc/features/
community-food-projects-database.html
101
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 8
Policy Options
Some traditional grocers have opened stores in more densely populated areas
(although not necessarily low-income areas), adapting their store models to
fit into smaller spaces. Further, many other nontraditional retailers, such as
pharmacies and dollar stores, have expanded into food retail as well (Sharkey
and Horel, 2009). Smaller corner stores and convenience stores that already
serve areas without access to a supermarket have also expressed interest in
offering more nutritious foods in their stores (Gittelsohn, 2009). These stores
are sometimes unwilling to take the risk of offering these foods when there is
uncertainty about whether they can sell enough of them (Gittelsohn, 2009).
103
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Map 8.1
Locations of new or planned foodstores in Philadelphia
through Pennsylvania Fresh Food Financing Initiative
Blue areas represent the areas of greatest need, which have low
supermarket sales, low income, and high rates of deaths due to
diet-related diseases. The orange dots represent stores that have
been opened under the Pennsylvania Fresh Food Financing Initiative.
The yellow dots represent stores funded by the program that are
under construction.
There are many other examples of grocery stores that have opened stores or
expanded service in underserved areas (see, for example, Food Marketing
Institute, 1998).
As explained in Chapter 6, both cost factors for food retailers and demand
factors for consumers may leave some areas underserved by food retail
stores or types of food retail stores. If cost factors keep food retailers from
developing new stores or expanding services in existing stores (e.g., if fixed
costs of obtaining land and permits for building a new store are higher in
these areas), then efforts to reduce these costs or to subsidize development of
new or expanded stores may be effective policy solutions.
Many localities and some States have utilized a number of methods to entice
new stores to open in these areas or to improve existing stores. These efforts
range from financing for new large-scale supermarkets, to small incentives
offered to existing stores to stock healthier foods, such as gift cards at fruit
and vegetable wholesalers. This section highlights a few of these efforts.
Previously published studies provide very good detail about the potential
benefits and the hurdles of different approaches and the strategies that can
be used to ensure successful store development (Policy Link, 2007; Food
104
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Marketing Institute, 1998). Less has been published about the potential
drawbacks of these approaches. Two examples of potential drawbacks
that have not been explored are the use of tax revenues to encourage
grocery stores when other uses of the revenues could be more beneficial to
low-income areas or the degree to which changes in competition faced by
existing stores could affect the local market.
During the 1990s, USDA’s Food and Nutrition Service pursued a research
agenda to address questions about food access among SNAP participants
and other low-income households (see Chapter 5). The agency concurrently
implemented three projects to identify strategies for improving access in
underserved areas. Activities included a conference with experts who
shared their ideas for increasing food access among low-income Americans
(Koralek, 1996), along with two studies. The studies examined successful
supermarkets in low-income, inner-city communities (O’Connor and Abell,
1992) and identified widely used means to improve consumer food access in
underserved neighborhoods (CRP, 1998).
Two key themes emerged from the conference and were reinforced by the
two descriptive studies:
106
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
jobs and helps to keep money in the community. These points can be used
to obtain support from a mayor or city council when it comes to handling
zoning, tax, and other issues. Buy-in from consumers will affect the ultimate
success of the business, so matters of location, preferred store type, and other
consumer interests should be addressed up front.
Community-Level Interventions
Another type of intervention to increase the availability of healthy foods is
through community-level programs, such as farmers’ markets, community
gardens, or mobile carts or trucks that sell fruits and vegetables. These
options for improving the food environment are often less expensive, require
less space, and can be quicker to implement than programs that encourage
new store development. These programs typically operate on a more limited
scale (e.g., seasonally or only 1 day a week) or in the case of community
gardens can require large time commitments on the part of consumers. The
goal of these interventions is to give consumers more options for purchasing
nutritious food by increasing the supply of these foods. Often these
interventions include nutrition and food education components or incentives
to consumers to increase demand for nutritious food.
In addition to the Healthy Bodegas initiative, the City of New York has
implemented two other community-level initiatives to increase access to
fruits and vegetables. One initiative is to increase the number of farmers’
markets in underserved neighborhoods and increase their use by residents
through the Health Bucks program. The Health Bucks program offers $2
coupons for the purchase of fresh fruits and vegetables at participating
farmers’ markets. Health Bucks were used as a tool to introduce consumers
to farmers’ markets, generating business for producers and helping reduce
food access barriers for consumers. The program was expanded specifically
for SNAP participants, with an additional $2 Health Buck offered for every
$5 spent using EBT at the farmers’ market. Through this program, EBT
sales at farmers’ markets more than doubled from $40,000 in 2007 to over
$89,000 in 2008 (Nonas, 2009). The program is being expanded into upstate
New York as the Fresh Bucks program.
In addition to the farmers’ market and Health Bucks program, the city is also
in the process of implementing a “Green Cart” program, which is intended
to increase the number of mobile carts that sell fresh produce, especially
107
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
in underserved areas. Green Cart vendors are subject to the same permit
process as other vendors, except they are restricted to operate in underserved
areas and sell only fresh produce under the NYC Green Cart umbrella. The
program is still recruiting vendors, a process that has been more difficult
than expected (Nonas, 2009). To aid in the recruitment process, private
foundation grants are being used to offer low-cost microloans to finance carts
for vendors.
Transportation-Related Improvements
In contrast to bringing the stores to people, an alternative approach is to
bring the people to the stores. Programs that improve public transportation
routes from areas with limited access or provide transportation subsidies to
low-income individuals or subsidized supermarket shuttle services could be
effective ways to reduce access problems. Transportation improvements
such as these have some advantages, depending on characteristics of the
areas. First, as the analysis in Chapter 2 showed, supermarket access
may also be a problem for low-income people who live outside of low
income areas. In cases where those with access problems are dispersed,
transportation interventions may be more efficient than new store
development since targeting an area for a store to locate to reduce access
problems for this group would be difficult (relative to areas where there are
high concentrations of low-income people with access problems). If public
108
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
transportation in the area already exists, it may not be too expensive or take
as much time to change or add routes that could be used to improve access
to stores. A transportation subsidy given to those with limited access may
enable consumers to access foodstores that better fit their budgets. Such
improvements could also help consumers access other services that may be
lacking in their area (such as banks or health care providers).
Using the NFSPS survey data and data from the Louisiana Neighborhood
Environment and Consumption Survey (LANECS), Rose et al. (2009)
estimated travel costs for different transportation modes and across areas in
New Orleans that had poor or good access to supermarkets. This exploratory
exercise considered both out-of-pocket travel costs and time costs for the
46The
different travel options for grocery shopping.46 time cost estimates use the
hourly minimum wage to value time.
See table 5 in Rose et al., 2009, for
Not surprisingly, a taxi was the most expensive option ($66.60 per month), details.
while a bus was $38.70, a ride from a friend or relative was $21.90, and
47Costs are to the nearest
walking was $21.00.47 Driving one’s own car was the least expensive
supermarket by mode of transport
method for getting to the store ($5.90 per month). The study also considered based on approach by Feather, 2003.
differences in time costs for New Orleans residents living in areas with poor Costs for each mode were weighted
access to supermarkets (defined as census tracts more than 2 kilometers from using the distributional data on the
a supermarket), compared with the time costs for those living in areas with mode of transportation.
good access (defined as census tracts within 2 kilometers of a supermarket)
(Rose et al., 2009). The average difference in travel cost between areas with
poor access and areas with good access was $10.58 per month—meaning
that SNAP participants in poor-access areas of New Orleans had total travel
costs (both time and out-of-pocket costs) that were on average (across mode
of transportation) almost $11 higher than those in areas in New Orleans with
good access.
109
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
SNAP Retailer Policy
Retailers authorized to participate in SNAP must sell staple foods for home
preparation and consumption. In addition, the store must offer for sale, on a
continuous basis, three different varieties of foods in each of four staple food
categories, with perishable foods in at least two categories; or have more than
50 percent of total gross sales in staple foods. Most stores are authorized
under criterion a. A store technically can meet this criterion with a small
number of items that meet the variety and perishable requirements.
Anti-Poverty Policy
Chapter 3 provided analysis of the factors correlated with whether an area
has limited access to affordable and nutritious food. Findings showed
that some household characteristics are associated with greater risk of low
access. Household characteristics such as low educational attainment, low
employment levels, and reliance on public assistance were found to be
the most influential determinants (in addition to low-income, upon which
the analysis was conditioned). Federal anti-poverty programs, such as the
Earned Income Tax Credit, Temporary Assistance for Needy Families;
and food and nutrition assistance programs, such as SNAP, WIC, and the
school meal programs are targeted to households with these characteristics.
These programs often have goals to encourage and support employment
and to offer income support, food, and better nutrition. Thus, some of the
populations for which a food access policy intervention may be targeted
already receive public assistance from these general anti-poverty programs.
If these programs are successful in raising incomes, employment prospects,
and educational opportunities, then low-income populations may be able to
access better foods as well.
Summary
This chapter outlined the types of policy options that could be considered
to reduce the effects of limited access to affordable and nutritious foods. A
formal evaluation of the benefits and costs of these policy interventions was
beyond the scope of this study. There are, however, some broad overarching
guidelines to consider if future policy interventions are needed.
The analysis in chapter 2 shows that the number of people that are likely
affected by limited access is small. Further, data on shopping patterns
and prices actually paid show that many, but perhaps not all, low-income
consumers have strategies for finding stores that offer the products they want
at lower prices. Both of these findings suggest that a large-scale, national-
level program may have difficulty addressing what are likely to be quite
localized pockets of limited access.
The analysis in Chapter 2 and from other studies (Sparks et al., 2009;
Apparicio et al., 2007) also shows food access may be a problem for some
low-income families who live outside of areas with high concentrations of
poverty. This observation is also important for policy considerations because
111
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
the type of intervention that may be effective in areas with concentrated
poverty are probably different than the type of interventions that may
be effective if the population with limited access is more geographically
dispersed. For example, improving existing stores or enticing new stores
or farmers’ markets to areas where the population with limited access is
concentrated may be an effective solution. On the other hand, if those
who lack access are more dispersed, it may be more effective to improve
transportation options or provide transportation subsidies for this population
to travel to stores rather than encouraging new stores.
On the demand side, the key question is if better access to affordable and
nutritious food is provided, will people buy these foods? Evidence is mixed
(see Cummins, 2005; Wrigley, 2003; Gittelsohn, 2009; Ayala et al., 2009).
These studies show small increases or no changes in consumption of healthy
foods in areas where access was improved. Related evidence suggests
that price subsidies for fruits and vegetables can increase consumption
of these foods for low-income consumers in general, but the effects are
small and consumption is still below recommended levels (Dong and Lin,
2009). It is possible that demand for more nutritious foods like fruits and
vegetables may also be driven by the time costs associated with preparing
and cooking them, especially considering the relative convenience and price
of alternative sources of foods. It is also possible that some consumers
may not know the nutritional benefits of such foods. Some may also lack
familiarity with preparation and the taste of these foods. If either of these is
true, then increasing access through supply-side factors alone (e.g., enticing
new stores) will not change food purchasing and dietary habits. A public
health campaign to promote healthier eating and cooking habits may be
more effective. Or, a program that gives low-income consumers incentives
to purchase more fruits and vegetables, such as the Healthy Incentive Pilot
that is being planned by FNS, could be more effective in increasing demand
for these foods. If the time-cost tradeoffs are the major determinants behind
lower demand, then such products as already-cut-and-cleaned fruits and
vegetables or healthier prepared food options may be useful. There are
usually price premiums for these prepackaged alternatives, however.
If supply-side factors, such as high fixed costs for new store development
in low-income areas or restrictive zoning policies, are an obstacle for larger
food retailers, then local policies toward zoning or government programs
(at the Federal, State, or local level) that subsidize these costs for stores
may be effective. There is some evidence that overall operating costs of
stores serving low-income consumers are similar to those of other stores
(King et al., 2004). But it is still possible that high fixed costs (as opposed
to operating costs) could be barriers to entry for stores trying to locate in
underserved areas.
Chapter 4 and Rose et al, (2009) raised the possibility that food swamps and
not food deserts may be more of a factor in BMI and obesity concerns. If
it is the case that energy-dense foods are too easily accessed, then policy
interventions may need to reduce the incentive to eat less healthful foods
or reduce access to these foods. This could be much more difficult because
even stores stocking the healthiest of options also stock the least healthy
options. Taxing less healthy foods (e.g., soda or sweetened beverages) is
one example of such a policy, but these taxes inherently have distributional
tradeoffs that must also be considered.
References
Apparicio, P., M. Cloutier, R. Shearmur (2007). The Case of Montreal’s
Missing Food Deserts: Evidence of Accessibility to Food Supermarkets,”
International Journal of Health Geographics, 6(1), 4.
Dong, D., and B.H. Lin (2009). Fruit and Vegetable Consumption by
Low-Income Americans: Would a Price Reduction Make a Difference?
Economic Research Report No. 70, U.S. Department of Agriculture,
Economic Research Service, January.
Feather, P.M. (2003). “Valuing Food Store Access: Policy Implications for
the Food Stamp Program,” American Journal of Agricultural Economics,
Vol. 85(1): 162-172.
114
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Fisher, A. (2009). “Community Level Food Environment” Presentation
at the IOM/NRC Workshop on the Public Health Effects of Food Deserts,
January 27, Washington, DC.
FNS-315 (2000). The National Nutrition Safety Net: Tools for Community
Food Security, U.S. Department of Agriculture, Food and Nutrition Service,
January.
King, Robert P., Ephraim S. Leibtag, and Ajay S. Behl (2004). Supermarket
Characteristics and Operating Costs in Low-Income Areas, Agricultural
Economic Report No. 839, U.S. Department of Agriculture, Economic
Research Service.
New York State Department of Health (2007). New York State WIC
Program 2006 Vegetable and Fruit Demonstration Project, Bureau of
Supplemental Food Programs Food Delivery Systems Unit, Division of
Nutrition, Albany, NY.
Policy Link, and LISC Bay Area (2007). Grocery Store Attraction
Strategies, http://www.policylink.org/documents/groceryattraction_final.pdf
Pothukuchi, K., and J. L. Kaufman (2000). “Longer View: The Food System:
A Stranger to the Planning Field,” Journal of the American Planning
Association, Vol. 66, Part 2.
115
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Rose, D. (2009). “Discount Bus Passes Proposed,” Wisconsin State Journal,
Friday, February 20, http://www.madison.com/wsj/home/local//index.
php?ntid=431881
Rose, D., J.N. Bodor, C.M. Swalm, J.C. Rice, T.A. Fraley, and P.L.
Hutchinson (2009). “Deserts in New Orleans? Illustrations of Urban Food
Access and Implications for Policy,” National Poverty Center Working
Paper, http://www.npc.umich.edu/news/events/food-access/index.php
Sparks, A., Bania, N., and L. Leete (2009). “Finding Food Deserts:
Methodology and Measurement of Food Access in Portland, OR,” National
Poverty Center Working Paper, http://www.npc.umich.edu/news/events/
food-access/index.php
116
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
CHAPTER 9
Progress has been made in characterizing the problem of food deserts and
in describing their consequences for affected populations. Many aspects
of food deserts, however, are not well defined or understood. This chapter
outlines future research needs for each of these aspects of food deserts. It
first focuses on conceptual and measurement issues of food access. These
were discussed in greater detail in Chapters 1 and 2. Next, it considers the
research issues for determining the causes of limited access, which were
discussed in Chapter 6. Finally, it highlights future research considerations
for understanding the economic and health consequences of limited access.
It is clear from the research that some areas and individuals have relatively
less access to some food retailers than other areas and individuals. This is
useful information which highlights issues of equity and areas where policy
117
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
interventions or private store development may be targeted. But it is not
clear whether these relative deprivations in access to food constitute an
absolute inadequacy of access to food. That is, just knowing that an area has
less access to food than another area may not be enough to say that access is
inadequate and thus, whether policy-interventions may be necessary. This
is not to say that there are no areas where access is inadequate. Rather,
such a designation has not yet been systematically considered to make such
judgments. Further conceptual clarification of the adequacy of food access is
needed in order to do so.
Actually measuring food access, even when the key conceptual constructs are
well-defined, will require a great deal more data. National-level assessments,
such as those conducted in Chapter 2, are important for monitoring broad
characteristics of the food environment in the U.S. The use of supermarkets
and large grocery stores as proxy measures of the availability of affordable
and nutritious food is a feasible way to operationalize a measure of access
on a broad national scale. Such stores typically offer competitive prices
and have greater variety of foods. But examining only large supermarkets
and grocery stores inevitably understates the sources of food available
to consumers, since food is available from many other retailers and from
restaurants and because consumers do not limit their food shopping to stores
near their homes. Thus, national-level assessments can only go so far.
It would also help to know the prices of goods because while some goods
may be available, their prices may be so high that they are not reasonably
affordable. Data collected by the Bureau of Labor Statistics to compute the
Consumer Price Index, or data from Nielsen’s Homescan panel could be
linked to food availability data (Bitler and Haider, 2009).
119
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Methodological and data improvements could make it possible to go beyond
cross-sectional relationships to isolate the effects of access, or changes in
access, on shopping and consumption behavior separately from the effects of
individual preferences for foods or other goods and services, which also may
affect shopping and food intake decisions.
Cross-sectional data sets like NHANES and the American Time Use
Survey could be further exploited to uncover the causal links. Doing so
may require linking geographical data to the individuals or households in
these surveys so that aspects of food access and the food environment can
be linked to food intake or time-use data.50 Localized data on food prices 50The estimates of time spent
from Nielsen is being linked with the NHANES survey by National Center traveling to grocery stores across
for Health Statistics staff and used by ERS staff to better understand how distances from supermarkets in tables
2.8 and 2.9 used such linked data.
price variation affects food consumption decisions. The 1996 National
Food Stamp Participant Survey collected a wealth of information on food
access, food shopping behavior, travel costs and time costs of food shopping,
and food use (see Chapters 2 and 5). This survey is over 10 years old.
An updated version of the survey could be considered to help improve
understanding of food access issues. In addition to providing more recent
data, additional information on food access, food shopping, and food use
could be collected in an updated round, including complete information
for both SNAP participants and eligible nonparticipants. It may also be
feasible to oversample individuals who live in areas that have low access to
supermarkets to provide greater sample size to study those who live in areas
with low access.
Longitudinal data on both consumer shopping and food intake and on the
food environment could help tease out the causal pathways between access
and consumption by linking changes in the food environment to changes in
shopping or food intake. Currently, none of the national-level surveys collect
detailed longitudinal data on consumer shopping or food intake. Given the
lack of national longitudinal data on food shopping and food intake, it may
be more feasible to add questions on intake of a few important foods (e.g.,
the frequency of eating fruits and vegetables or whole grains) to existing
longitudinal surveys such as the Panel Study of Income Dynamics, the
National Longitudinal Survey of Adolescent Health, or the Early Childhood
Longitudinal Surveys. These longitudinal surveys or others, such as the
Survey of Income and Program Participation, could also ask basic questions
120
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
of whether households have enough food and if not, why, such as those
questions that were at one time part of the CPS-FSS.
References
Bayer, P.J., R. McMillan, K.S. Rueben (2004). “An Equilibrium Model
of Sorting in an Urban Housing Market,” National Bureau of Economic
Research Working Paper No. W10865.
Bitler, Marianne, and Steven J. Haider (2009). “An Economic View of Food
Deserts in the United States,” National Poverty Center Working Paper, http://
www.npc.umich.edu/news/events/food-access/index.php.
Chen, S.E., R.J.G.M. Florax, and S.D. Snyder (2009). “Obesity, Fast Food,
and Grocery Stores: Evidence from Geo-referenced Micro Data,” National
Poverty Center Working Paper, http://www.npc.umich.edu/news/events/
food-access/index.php
King, Robert P., Ephraim S. Leibtag, and Ajay S. Behl (2004). “Supermarket
Characteristics and Operating Costs in Low-Income areas Areas,”
Agricultural Economic Report No. 839, U.S. Department of Agriculture,
Economic Research Service.
Rose, D., J.N. Bodor, C.M. Swalm, J.C. Rice, T.A. Farley, P.L. Hutchinson
(2009). “Deserts in New Orleans? Illustrations of Urban Food Access and
Implications for Policy,” National Poverty Center Working Paper. http://
www.npc.umich.edu/news/events/food-access/index.php.
121
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Appendix A
Waugh Auditorium
Economic Research Service
1800 M Street, NW
October 9, 2008
123
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
LaDonna Redmond, Institute for Community Resource
Development, Chicago, IL
Pamela Roy, Farm to Table & the New Mexico Food and
Agriculture Policy Council
Ken Meter, Crossroads Resource Center, Minneapolis,
MN
3:00 – 3:15 Break
3:15 – 4:45 Session 4: Roundtable Discussion of Program and Policy
Options
Chair: Neil Conklin, Farm Foundation
Tim Hammonds, Food Marketing Institute
John Weidman, The Food Trust
Andy Fisher, Community Food Security Coalition
Kami Pothukuchi, Wayne State University
4:45 Closing and Adjourn
Neil Conklin, Farm Foundation
124
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Appendix B
Box B.1
Measures of Access Used in Food Desert and Related Studies
1) Algert, Susan J., Aditya Agrawal, and Douglas S. Lewis (2006). "Disparities in
Access to Fresh Produce in Low-Income Neighborhoods in Los Angeles." American
Journal of Preventive Medicine 30(5): 365-370.
Area Studied Measure of Access
Pomona, CA Divided food stores in Pomona as having a “variety” of produce
or “limited produce” based on onsite inspections, telephone
interviews regarding produce content, or inference from store
description. Stores selling four or fewer produce items were
counted as “limited” while those serving more than four were
counted as having a “variety” of produce. Access was mea-
sured by distance from a store offering a variety of produce.
Addresses of food pantry clients were geocoded. Those living
outside of a “walkable” distance to a store selling a variety of
produce (0.8 km or about a 15 minute walk) were highlighted
as not having access to a variety of produce.
2) Alwitt, Linda F., and Thomas D Donley (1997). "Retail Stores in Poor Urban
Neighborhoods." The Journal of Consumer Affairs 31(1): 139-164.
Area Studied Measure of Access
Chicago, IL Examined number of retail stores of different types and sizes
in “poor” ZIP Codes. Poor ZIP Codes are defined as: 1) poverty
rate in highest quartile of ZIP Codes in Chicago, 2) lowest
quartile of high school graduation rate, 3) lowest quartile of la-
bor force participation, and/or 4) highest quartile of unemploy-
ment rate. Store size was defined by the number of employees.
3) Apparicio, Philippe, Marie-Soleil Cloutier, and Richard Shearmur (2007). "The
case of Montréal's missing food deserts: Evaluation of accessibility to food super-
markets." International Journal of Health Geographics 6(4).
Area Studied Measure of Access
Montreal, Canada Measured access by proximity (nearest supermarket), diversity
(number of supermarkets within 1,000 meters), and variety in
terms of food and prices (mean distance to the three closest,
different chain-name supermarkets). Supermarkets were de-
fined as grocery stores associated with one of the seven major
chains in Quebec.
4) Baker, Elizabeth A., et al. (2006). "The Role of Race and poverty in Access to
Foods That Enable Individuals to Adhere to Dietary Guidelines." Preventing Chronic
Disease: Public Health Research, Practice, and Policy 3(3).
Area Studied Measure of Access
St. Louis, MO Supermarket audits were used to assess “healthy food” avail-
ability in stores, where healthy food was defined by USDA
dietary intake guidelines. A checklist of 78 canned, fresh, or
frozen fruits and vegetables were used by auditors. Availability
of lean, low-fat and fat-free meat, poultry, and dairy products
was also considered. Composite score for each supermar-
ket was developed based on these audits. The composite
scores were then divided into tertiles of high, medium, and low
availability. Census tracts were divided by racial composition
(greater than 75% African-American, greater than 75% White,
and mixed) and poverty level (less than 10% poor, 10-20%
poor, and greater than 20% poor).
5) Berg, Nathan, and James Murdoch (2008). "Access to grocery stores in Dallas."
International Journal of Behavioural and Healthcare Research 1(1): 22-37.
125
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area Studied Measure of Access
Dallas County, TX Mapped all of the mainline chain grocery stores in Dallas
County and then counted how many stores were within a
1-mile radius of neighborhood Census block groups. The study
linked this geospatial data with neighborhood income levels,
poverty levels, and racial composition.
6) Blanchard, Troy C., and Thomas Lyson (2006). "Access to Low Cost Groceries in
Nonmetropolitan Counties: Large Retailers and the Creation of Food Deserts." Jour-
nal of Regional Studies.
Area Studied Measure of Access
Mississippi A food desert is classified as having no supermarkets or
supercenters within a 10-mile radius of a ZIP Code centroid
(with exceptions for highways). A supermarket classification
was based on having 50 or more employees and a supercenter
classification was based on NAICS category 452910.
7) Horowitz, Carol R., et. al. (2004). "Barriers to Buying Healthy Foods for People
With Diabetes: Evidence of Environmental Disparities." American Journal of Public
Health 94(9): 1549-1554.
Area Studied Measure of Access
East Harlem and Classified stores as desirable if they had at least one item of
Upper East Side, the following five and undesirable if it had none of the items:
New York City (1) Diet soda (1-L or 2-L size)
(2) 1% fat or fat-free milk (1-quart, half-gallon, or 1-gallon size)
(3) High-fiber bread, low-carbohydrate bread, or both high-fiber
and low-carbohydrate bread (defined as 2 g or more fiber, 10 g
or less carbohydrate per slice, or both)
(4) Fresh fruits
(5) Fresh green vegetables or tomatoes.
8) Block, Daniel and Joanne Kouba (2005). “A comparison of the availability and af-
fordability of a market basket in two communities in the Chicago area.” Public Health
Nutrition 9(7): 837-845.
Area Studied Measure of Access
Austin and Oak The study compared item availability, price, and quality of food
Park in Chicago, IL in 10 different types of foodstores in Austin (poor and African-
American) and Oak Wood (upper-middle class and diverse
races). Item availability was calculated by how many items
a store stocked from a food list based on the USDA’s Thrifty
Food Plan. Price was calculated from the items in the food
basket and the average price from all stores was assigned if
an item was not sold at that store. Quality rating was subjec-
tive and only given two options, “satisfactory” or “poor.” The 10
store types were national and regional chain markets, indepen-
dent supermarkets, independent groceries, chain drug stores,
gas stations, liquor stores with food, chain convenience stores,
dollar stores, and specialty stores.
9) Block, JP, RA Scribner, KB DeSalvo (2004). "Fast food, race/ethnicity, and income:
a geographic analysis." American Journal on Preventive Medicine 27(3): 211-7.
126
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area Studied Measure of Access
New Orleans, LA The study geocoded all fast food restaurants and used a 1
mile and 0.5-mile radius as buffers around Census tracts to
determine “shopping areas” in each tract. Fast food restaurant
density was calculated by the number of fast food restaurants
per square mile. Fast food restaurants have two or more of the
following characteristics: expedited food service, takeout busi-
ness, limited or no wait staff, and payment tendered prior to
receiving food. These data were then compared with neighbor-
hood characteristics, such as percentage Black population and
low income.
10) Bodor, J Nicholas, Donald Rose, Thomas A Farley, Christopher Swalm, and
Susanne K Scott. "Neighborhood fruit and vegetable availability and consumption:
the role of small food stores in an urban environment." Public Health Nutrition 11(4):
413-420.
Area Studied Measure of Access
Four Census tracts Adequate access was calculated by two components; distance
in central-city New from the household to a foodstore and the instore availability of
Orleans healthy items. The study included all stores selling food within
the four Census tracts and all supermarkets within 5 km of the
tract borders. Trained observers determined instore availabil-
ity by calculating the linear shelf space devoted to fruits and
vegetables and the number of fresh produce varieties available.
Neighborhood availability was then determined by summing all
of the shelf space devoted to fresh fruits and vegetables in all
small food stores within 100 m of the household residence.
11) California Center for Public Health Advocacy, PolicyLink, and UCLA Center for
Health Policy Research (2008). Designed for Disease: The Link Between Local Food
Environments and Obesity and Diabetes.
Area Studied Measure of Access
California Retail Food Environment Index (per adult): Total number of fast
food restaurants and convenience stores divided by the total
number of grocery stores and produce vendors within a given
radius of the person’s home. The radius used was 0.5 miles for
urban areas and 5 miles for rural areas.
12) Clarke, Graham (2002). "Deriving Indicators of Access to Food Retail Provision
in British Cities: Studies of Cardiff, Leeds, and Bradford." Urban Studies 39(11):
2041-2060.
Area Studied Measure of Access
Cardiff, Leeds/ The study used GIS to map store locations. A circle with a
Bradford (UK) 500-meter radius was drawn around each multiple/co-op store,
which represents the maximum walking distance. Potential
food deserts were only the areas that met the Carstairs indices
of multiple deprivation and only had a few, small independent
stores. The Hansen accessibility and spatial interaction model
take into account the location and size (30,000, 60,000, and
90,000 square feet) of store in relation to its customers and
social class structures.
13) Examining the Impact of Food Deserts on Public Health in Chicago. Mari Gal-
lagher Research & Consulting Group (2006).
127
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area Studied Measure of Access
Chicago, IL Food Balance Score: average distance (blocks) to any main-
stream food venue divided by the average distance to a fringe
food venue (includes fast-food). A mainstream food venue is a
grocery store and a fringe food venue is everything else, such
as convenience stores and fast food restaurants. The study
compared food ratio scores and their impact on Years of Po-
tential Life Lost (YPLL) and BMI outcomes, holding education,
income, and race constant.
14) Examining the Impact of Food Deserts on Public Health in Detroit. Mari Gal-
lagher Research & Consulting Group (2007).
Area Studied Measure of Access
Detroit, MI Food Balance Score: average distance (blocks) to any main-
stream food venue divided by the average distance to a fringe
food venue. A mainstream food venue is a grocery store and
a fringe food venue is everything else, such as convenience
stores and fast food restaurants. Researchers manually clas-
sified store types based on appearance and items they sold.
Diet-related death data are used to calculate Years of Potential
Life Lost, a statistic that measures the total number of life years
lost due to premature death per 100 people in a population
from a certain cause. The Food Balance Score was then com-
pared with the YPLL statistic in each neighborhood.
15) Guy, Cliff, Graham Clarke, and Heather Eyre (2004). “Food retail change and the
growth of food deserts: as case study of Cardiff.” International Journal of Retail and
Distribution Management 32(2): 72-88.
Area Studied Measure of Access
Cardiff, UK The study measured the spatial distribution of multiple and co-
op grocery stores (healthy food stores) using electoral divisions
as the geographical unit. Areas with a high deprivation are
considered food deserts, and this was calculated by the Welsh
Index of Multiple Deprivation (IMD). IMD includes income,
employment, health, education, housing, and geographic
components.
16) Hendrickson, Deja, Chery Smith, and Nicole Eikenberry (2006). "Fruit and veg-
etable access in four low-income food deserts communities in Minnesota." Agricul-
ture and Human Values 23: 371-383.
Area Studied Measure of Access
Four urban and The study used a survey based on the Thrifty Food Plan to
rural communities measure the average price per pound of a food item, how
in Minnesota much of the food item was offered, and the quality of the food
item. Quality was determined by expiration dates on pack-
aged food and by the opinion of surveyors for produce items.
Consumer focus groups also revealed what residents thought
of the availability, quality, and price of food.
17) Hosler, Akiko, et. al. (2008). "Assessing Retail Fruit and Vegetable Availability
in Urban and Rural Underserved Communities." Preventing Chronic Disease 5(4).
http://www.cdc.gov/pcd/issues/2008/oct/07_0169.htm. Accessed December 4, 2008.
128
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area Studied Measure of Access
Urban and rural Instead of categorizing stores based on business type (such
areas in New York as a gas station or supermarket), businesses were labeled as
State a fruit and vegetable stores if they stocked at least two types
of fresh fruit and at least three types of fresh vegetables. There
were three types of fruit and vegetable stores: super produce
stores, year-round produce stores, and seasonal produce
stores. Fruit-for-snack stores were stores that carried at least
one type of ready-to-eat fresh fruit but didn’t meet the fruit and
vegetable measure.
18) Sharkey, Joseph R. and Scott Horel (2008). "Neighborhood Socioeconomic
Deprivation and Minority Composition Are Associated with Better Potential Spatial
Access to the Ground-Truthed Food Environment in a Large Rural Area." The Jour-
nal of Nutrition 138(3): 620-627.
Area Studied Measure of Access
6-county rural The study calculated the distance to the nearest foodstore from
region in Texas the population-weighted center of each CBG and then exam-
ined associations among deprivation, minority composition,
population density, and distance to foodstore.
19) Kaufman, Phil R. (1999). "Rural Poor Have Less Access to Supermarkets, Large
Grocery Stores." Rural Development Perspectives 13(3): 19-25.
Area Studied Measure of Access
Lower Mississippi Net accessibility is the ratio of available large grocery store
Delta (36 rural, sales to potential food spending by households in a ZIP Code-
high-poverty based area (region broken up into quartiles). The stores in the
counties) study all have annual sales of $500,000 or more.
20) Morton, Lois Wright and Troy C. Blanchard (2007). "Starved for Access: Life in
Rural America's Food Deserts." Rural Realities 1(4): 1-10.
Area Studied Measure of Access
Rural Iowa Low access: 50% of population lives more than 10 miles from
large foodstore (supermarkets or supercenters).
Food Desert: All residents live more than 10 miles from large
foodstore
Large foodstores are large supermarkets and supercenter.
21) Short, Anne, Julie Guthman, and Samuel Raskin (2007). "Food Deserts, Oases,
or Mirages? Small Markets and Community Food Security in the San Francisco Bay
Area." Journal of Planning Education and Research 26: 352-364.
Area Studied Measure of Access
San Francisco, CA The study measured accessibility by distance (mapping),
(Bayview, Mis- affordability and nutritional adequacy (market basket analy-
sion, Central East sis), cultural acceptability (types, variety and quality of goods
Oakland) carried and characteristics that may affect the shopping
experience), and quality (produce observation) for small (less
than 3,000 sq ft) full-service foodstores. Instead of measuring
overall community access, the study measured whether small
full-service stores can enhance access to food compared with
chain and large supermarket availability.
22) Smoyer-Tomic, Karen E., John C. Spence, and Carl Amrhein (2006). "Food Des-
erts in the Prairies? Supermarket Accessibility and Neighborhood need in Edmon-
ton, Canada." The Professional Geographer 58(3): 307-326.
129
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Area Studied Measure of AccessLocations of Food Access Studies
Local/ State: Regional/ Nation:
Edmonton,
U.S. Kindergarten
Canada TheEdmonton
study compared
was broken the
upchange
into neighborhoods
in the children’s
based
BMIs on(fiZIP
rst
children Codeand
year andthird
accessibility
year) withwasthemeasured
prices for by
meat,
distance
fruits, (shortest
dairy, and
1-Pomona, CA 13-Chicago, IL 26-South USA
route)
fast food
and andGIS the
mapping
per capita
to thenumber
closest
of foodstores
supermarket. andA restau-
super-
2-Chicago, IL 14-Detroit, MI 27-USA
market
rants in was
the residence
defined asZIP offering
Code.a full range of grocery items
3-Montreal, Canada 15-Cardiff, UK 28-USA
(dairy,
29) Powell, Lisa M., 4-St.
et. fresh produce,
al.Louis,
(2007). meat,Between
"Associations and baked goods) and having Storesa
MO 16-MinnesotaAccess to Food29-USA
and Adolescent Body minimum
Mass Index." of 10 employees. Two areas within Edmonton were
5-Dallas County,American
TX JournalYork
17-New of Preventive Medicine 30-USA
then compared.
33(4S): S301-S307.6-Mississippi 18-Texas 31-UK
23)
Area Wrigley,
StudiedNeil, Daniel
7-New Warm,
Measure York,ofand
NY Barrie Margetts
Access 19-Lower (2003). "Deprivation, diet, and
MS Delta
8-Chicago,
food-retail access: findings fromILthe Leeds 'food 20-Iowa
deserts' study." Environment * 26-31andnot shown on
U.S. adolescents
Planning The study compared the BMI of adolescents to the distance
maps
A 35: 151-188.
and type
9-New of foodstores
Orleans, LA located
21-Sannear residence
Francisco, CA (ZIP Code).
Area Studied Measure of four
Access
There were
10-New Orleans, types of22-Edmonton,
LA stores: chain supermarkets,
Canada nonchain
Leeds, England The study measured
supermarkets,
11-California groceryboth
stores,
the physical
23-Leeds, andUKconvenience
(distance stores.
and mode
30) Rose, Donald and ofRickelle
transportation)
12-Cardiff, UK
Richards and economic"Food(income)
24-Detroit,
(2004). MI access
store accessandbarriers to
household
supermarkets before and after
25-Detroit,an intervention.
MI
fruit and vegetable use among participants in the US Food Stamp Program." Public
Health
24) Zenk,
Nutrition
Shannon,
7(8):et.1081-1088.
al. (2005). "Neighborhood Racial Composition, Neighbor-
hood Poverty,
Area Studied and the MeasureAccessibility
Spatial of Access of Supermarkets in Metropolitan Detroit."
American
UK Journal of Public Health 95(4): 660-667. Canada
US Supermarket Access Variable: Distance of store, travel time,
Area Studied Measure of Access
individuals’ car ownership, and type of store was used to
Detroit, MI measure
The studyhousehold
measuredfruitthe use and household
Manhattan vegetable
block distance use
to near-
(expressed in grams
est supermarket (onlyper male
chain adult equivalent
supercenters per day).
and chain full-line
grocery
31) Shaw, Hillary J (2006). stores)
"Food from the
Deserts: middle
Towards theofDevelopment
the neighborhood for 869ca-
of a Classifi
tion." Series B: Human neighborhoods.
Geography 88(2): 231-247.
25)
AreaZenk, Shannon N.,
Studied et al. (2005).
Measure "Fruit and Vegetable Intake in African Ameri-
of Access
cans: Income and Store Characteristics." American Journal of Preventive Medicine
UK
29(1): 1-9. Urban areas were divided into 250 sq m blocks and rural
23
areas were divided into 500 sq m blocks. Each neighborhood
Area Studied Measure of based
Access
was ranked on whether the area was residential, had a
Eastside Detroit, grocery
The study store, and whether
measured 10 orofmore
the effects kinds of fresh
sociodemographics fruit
and and
22
MI (97% African vegetables were sold (scale
food-store characteristics on1-4).
fruit and vegetable intake. So-
American) 12, 15
ciodemographics include age, education, and family income.
3
The foodstore characteristics were location, type (classified by
definitions from the Food Marketing Institute), and perceived
quality and availability of fruit and vegetables.
26) Blanchard, Troy C., and Thomas Lyson (2006). Food Availability and Food
Deserts in the Nonmetropolitan South. Assistance Needs of the South’s Vulnerable
USA
Populations, Number 12.
Area Studied Measure of Access
16
South US A food desert is when a certain percentage of a county’s
population is not located within 10 miles of a large food retailer
(supermarkets with 50 or more employees, supercenters, and 17
wholesale clubs).
11 M, et. al. (2006). “Food store availability and20neighborhood 14, 24, 25 7
27) Powell, Lisa charac-
teristics in the United States.” Preventive Medicine 44: 189-195. 2, 8, 13
Area Studied 21Measure of Access
4
28,050 ZIP Codes The study compared the availability of chain and nonchain
in United States supermarkets, grocery stores, and convenience stores to
1
neighborhood characteristics (race, ethnicity, and SES) at the
ZIP Code level. “Availability” was defined as the6number of
each type of store in the ZIP Code.
28) Sturm, R., and A. Datar (2004). "Body mass index in elementary school chil-
dren, metropolitan area food prices and food outlet density." The Royal19
Institute of
9, 10
Public Health 119: 1059-1068.
130
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Appendix C
I. Methods
Store Directory Development. The directory of authorized SNAP
foodstores was merged with the TDLinx listing of supermarkets to take
advantage of the strengths found in each source. SNAP-authorized stores
that did not match with the TDLinx directory were examined in detail.
After reviewing SNAP stores classified as either SM (supermarkets) or SS
(superstores) in the 2006 listing, it was concluded that many of the SNAP
superstores did not meet the same criteria as the industry supermarkets.51 51 SNAP-authorized stores were, for
To further sort through which of the SNAP stores met the definition of a the most part, self-classified by the
supermarket, researchers examined the annual total sales and food sales of applicant. A large number of SNAP
superstores are known to carry some
all SNAP stores classified as SM and SS that did not match with TDLinx packaged foods but are less likely to
supermarkets. Stores that did not meet the annual $2 million in sales include fresh fruit and vegetables or
requirement or that did not have significant food sales were eliminated. fresh meat products.
Researchers used the Trade Dimensions Marketing Guidebook (2008)
to verify where food sales data were unavailable or not current. Online
sources, such as company Web sites and other online sources for additional
information about a store’s annual sales and the kinds of foods sold, were
checked. By combining the two store listings and using outside sources
for verification, researchers obtained a more comprehensive national list of
supermarkets and supercenters.
131
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The walkability range is categorized as high, if a supermarket is within a
half mile; medium, if a supermarket is between one-half and 1 mile; and
low, if the nearest supermarket is more than a mile away. Obviously,
whether walking to a supermarket is feasible or not depends on more than
just distance—it could also depend on whether the individual is capable of
walking that distance, whether there are safe sidewalks on which to walk and
controlled intersections, and whether there are other barriers, such as crime
that may make walking to a store dangerous. Furthermore, the measures of
distance are all Euclidian, or straight-line distance, which may not represent
the actually distance that must be walked to access a supermarket. Despite
these limitations, this definition of walkability is grounded in the literature.
Algert, Agrawal, and Lewis (2006) defined access to stores selling a variety
of produce as a walkable distance of 0.8 km., or about a 15-minute walk.
Apparicio et al. (2007) measured supermarket access proximity by the
number of stores within 1,000 meters, or about 0.6 miles. In a study by
the California Center for Public Health Advocacy and the UCLA Center
for Health Policy Research (2008), a store radius of 0.5 miles was used to
determine adequate access in urban areas.
132
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
While the choice of search area to use is not empirically derived, researchers
experimented with different search areas and observed the resulting plots of
low-income areas. An effort was made to avoid applying search criterion
less than 3 km., in which results gave highly fragmented (pixilated) areas,
interspersed with many higher income areas. Conversely, the use of a large
search criterion, such as 5 km., resulted in very large low-income areas that
included many higher income populations within it. In determining the
kernel density search area, these factors and outcomes were considered.
II. Tables.
Table C.1
National levels of access households living in higher income areas: walking and driving distances1
Number of Share of total
low-income low-income
Share of Total Share of total persons in persons in
Number total square higher-income higher-income higher-income higher-income
of square- kilometer area persons, area persons, areas, by areas, by
Access level kilometer grids grids by access level2 by access level access level access level
Percent Millions Percent Percent
Walking:
Driving:
High 3,776,567 67.0 205.2 98.5 42.4 97.9
Medium 1,292,894 22.9 2.8 1.3 0.8 1.9
133
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
III. Figures.
Measuring Access Using Cumulative Density Functions (CDFs). In the
area-based analysis, researchers first separated the U.S. population into one
of three urbanicity types (urban, urban clusters, and rural) to compare access
to supermarkets among areas having similar levels of built environment. For
each urbanicity, low-income areas were compared with non-low income
areas, as well as for three additional vulnerable subpopulations. Within
each urbanicity-subpopulation class, distance was measured to the nearest
supermarket for each grid. All grids were ranked according to distance,
from closest to farthest. The corresponding populations of the ranked grids
were tallied and accumulated. The resulting distances and corresponding
cumulative share of total population were plotted in a cumulative density
function (CDF) chart, in which for any point on the curve(s), distance
(measured on the horizontal axis) and its corresponding cumulative share of
population (measured on the vertical axis) can be determined.
Figure C.1. plots separate CDFs for each of the three urbanicities: urban
areas, urban clusters, and rural areas. The vertical axis shows the cumulative
share of the total U.S. population, with the horizontal axis indicating distance
to the nearest supermarket. Using CDFs, one can examine the CDF curves
of two or more subpopulations for potential access inequalities. Large
differences in access appear as gaps between the individual CDF curves.
The overall shape of individual curves also reflects the range of access
experienced for a given share of the population or subpopulation. Access
curves initially having very steep slopes and gradually flattening at the top
indicate superior access relative to access curves that have a more gradual
slope extending to the upper right corner of the chart. These differences
can be seen in figure C.1., where dotted vertical lines demarcate distances
of 0.5 and 1.0 miles, indicating high, medium, and low access for each of
the three urban categories. The intersections of the horizontal lines on each
curve indicate the corresponding share of the total population for each urban
category. Accordingly, 74.4 percent of the urban population was within 1
mile of the nearest supermarket. Access in urban clusters was only slightly
lower, where 64.9 percent of the population had a supermarket within 1 mile.
In contrast, rural populations had considerably lower levels of access. Only
12.4 percent of the rural population was within 1 mile of a supermarket.
This result is not unexpected, due to the much greater spatial dispersion
of the rural population and its greater dependence on owned-vehicle 52In the analysis of rural area access to
transportation.52 supermarkets, we employ both walking
and driving distance criteria. We do not
Similar CDF charts are presented below for selected urbanicity- show the drivable distance on figures
subpopulation combinations. that include urban and urban cluster
areas because very few people in these
areas are outside of 10 miles of a super-
market (in fact, almost all are within 4
miles, as shown in figure 3.1.
134
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Figure C.1
Supermarket access by urbanicity for the total population
0.8
0.6
Rural areas
0.4
0.2
0.0
0 0.5 1 2 3 4
Distance to nearest supermarket (mi)
Figure C.2
Access to supermarkets by urbanization level for non low income areas
1.0
0.8
Low income area
Non−low income areas
0.6
0.4
0.2
0.0
0 1 2 3 4
Distance to nearest supermarket (mi)
135
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Figure C.3
Supermarket access for low-income individuals in low-income and
non-low-income areas within urban areas
Cumulative share of low income population (%)
1.0
0.6
0.4
0.2
0.0
0 1 2 3 4
Distance to nearest supermarket (mi)
Figure C.4
Supermarket access for households without vehicle by low-income and
non-low-income areas within rural areas
Cumulative share of households without a vehicle
1.0
0.4
0.2
0.0
0 5 10 15 20 25
Distance to nearest supermarket (mi)
136
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Figure C.5
Supermarket access for households without vehicle by low-income and
non-low-income areas within rural areas
Cumulative share of non−white population
1.0
0.8
0.6
0.4
0.2
0.0
0 5 10 15 20 25
Distance to nearest supermarket (mi)
Figure C.6
Supermarket access for elderly households by low-income and non-low-
income areas within rural areas
Cumulative share of people 65 and older
1.0
0.8
0.6
0.4
0.2
0.0
0 5 10 15 20 25
Distance to nearest supermarket (mi)
137
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Methodology for Average Time Spent
in Travel to Grocery Shopping
The estimates of average time spent in travel to grocery shopping were
made using the pooled 2003-2007 American Time Use Survey (ATUS)
data. The U.S. Bureau of Labor Statistics’ ATUS is a continuous survey
that began in 2003. One individual age 15 or older from each sampled
household is interviewed about his or her use of time for the 24-hour period
from 4 a.m. the day before the interview to 4 a.m. on the interview day. The
time diary information includes where the respondent was and whom the
respondent was with for each activity. The ATUS also collects information
on the respondent’s household, labor force participation, and demographic
characteristics. (For more information on the ATUS, see www.bls.gov/tus).
The ATUS time diary data contain extensive information about Americans’
travel. The data specify travel as an activity, and record mode of
transportation and whom the respondent was with when traveling. If the
travel was by vehicle, the data include whether the respondent was the driver
or passenger. Travel is defined as moving from one location (or address) to
another.
To deal with the complexities of trip chaining, ATUS codes the travel
activities as to their purpose, looking ahead to the next activity and location.
For example, the time a respondent travels from home to work is coded
as 180501, “Travel related to working.” If the respondent went grocery
shopping after work, the time spent traveling to the store is coded as 180701,
“Travel related to grocery shopping.” The exception to the “looking ahead”
rule is when the respondent is traveling from one location to home, in which
case the purpose of the travel is coded as a previous activity. As a result,
calculating travel time to the grocery store is complicated by the fact that
some diaries will have only one “side” of travel related to grocery shopping
coded as travel related to grocery shopping and others will have both
sides—the going and coming home—of the trip coded as travel related to
grocery shopping.
To best deal with trip chaining and the ATUS coding, researchers decided
to estimate average time to the grocery store as follows. For each time
diary with grocery shopping as an activity in the respondent’s time diary,
researchers added all legs of travel from home to the grocery store and then
138
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
added all legs of travel from the grocery store to home. The total travel time
home-to-shopping was compared with the total travel time shopping-to-
home, and the shorter total time was chosen as the “time distance” to grocery
shopping. In doing this, researchers did not have to consider the coded
purpose of the travel, and they also did not have to consider the “dwell time,”
that is, the time spent on an activity between two travel occurrences. All the
characteristics of travel to grocery shopping, such as mode of transportation,
were characteristics of the shorter travel side. In cases where the respondent
did not start the day at home or did not end the day at home, researchers only
had information for one side (home to grocery shopping or grocery shopping
to home). In these cases, the total travel time for that side was used as the
time distance to the grocery shopping.
For 6.4 percent of the grocery shoppers in the ATUS data, the shortest time
distance is actually from work and not from home. It was decided that the
work location is a relevant means of access to grocery shopping, so for these
respondents the travel time is work-to-store or store-to-work. Consequently,
the average times presented use two “anchors,” home and work.
Grocery shopping was defined as the ATUS activity 070101 with the
location of grocery store, restaurant or bar, other store/mall, outdoors away
from home, or other place. Grocery shopping with the locations home,
work, church, and library were not included so as to exclude online grocery
shopping. If the respondent was not at home for any activity in the 24-hour
time diary, that observation was not included. If the mode of travel was by
airplane, the observation was excluded as the grocery shopping was likely in
an airport during out-of-town travel. Thirty-seven observations with grocery
shopping were excluded for the following reasons: it was determined that
the respondent was out of town when the grocery shopping took place; the
time diary was not clear as to what the respondent did on the diary day,
usually due to “can’t remember” or “none of your business” responses;
or the respondent had a large number of errands or other activities on the
diary day such that the total travel time to grocery shopping would likely
be an overestimate of the time distance to the grocery store. The resulting
dataset used to calculate the estimates contains 11,569 observations. A small
number of extreme cases were included—respondents who had zero minutes
travel to grocery shopping, which is possible if the store is in the same
building as the respondent’s previous activity, and respondents who had
120 minutes or longer travel to grocery shopping. Additional details about
variable coding and definitions are available on request.
The Bureau of Labor Statistics (BLS) and the Census Bureau calculated
the estimates to ERS specifications to have estimates by whether or not the
respondent was in a low-access area. Without BLS and Census cooperation,
this analysis could not have been done as detailed geographical information
is not publicly available in the ATUS data files.
139
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Appendix D
Examples of Community Food Projects
by Type of Project
Food Assessments
The Fresno Fresh Access Project was funded at $200,000 for 2 years.
The grantee organization was the Fresno Metropolitan Ministry, Fresno,
California. The purpose of this community food assessment was to identify
key factors in food availability, accessibility, affordability, and quality in
Fresno County. The objectives were to assess a number of districts, involve
local residents and community volunteers in the survey process, empower
community members to make food policy recommendations to local officials,
and create an action plan from data collected during the assessment.
141
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The FSAT has been widely used by Native American groups in assessing
their food systems.
The Legacy Cultural Learning Center’s project goal was to create the
first organized collection of data focused on the food, diet, and traditional
agriculture issues of the Muskoke People and their neighbors. The project
assessed the food assets, strengths, needs, and deficiencies of rural,
low-income communities in the Tribal Nation. They conducted meetings
and surveys in 17 of the 23 Muskoke communities and in one neighboring
Seminole Nation community to assess food system assets, strengths, needs,
and deficiencies. There was good participation in discussions focused on
diet-related health problems such as diabetes, heart trouble, and high blood
pressure.
Data was collected from written surveys and also from more than 17 person-
to-person interviews with tribal and community leaders, academics, and
project collaborators. Through the surveys, the project learned that the Tribal
Elderly Nutrition Service operates 10 fully equipped kitchens and dining
facilities serving more than 18,000 meals monthly, including an extensive
home delivery network. The surveys uncovered a heavy reliance on fast
food, a preference for local food but limited access to it, and significant
concerns about rising food costs. Approximately one-third of respondents
have gardens. Surveys found a significant positive consumer response to the
availability of locally produced foods available directly from farmers and
farmers’ markets. However, there are few local growers and small farmers do
not have the ability to supply the quantities needed.
Following the surveys and interviews, project staff developed a draft plan
of action to address the most urgent needs and concerns, including Elderly
Nutrition Service purchases of local products, initiation of community
gardens on donated land, purchase of a large RV to serve as a mobile
resource center, and establishment of a Native Foods restaurant (The Legacy
Learning Center, Muskogee, Oklahoma, 2006).
The project, Food Access and Food Justice in New Orleans: Rising
Above the Waterline, was funded in 2006 for $265,000 for 3 years.
The organization funded was the New Orleans Food and Farm Network
(NOFFN). It was designed to assist in the Hurricane Katrina recovery.
The project supported the development of sustainable food projects
and neighborhood capacity by facilitating innovative linkages at the
neighborhood level. In brought together diverse groups including public
health, social work, urban agriculture, and food security sectors via an active
food system coalition.
142
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
and community members, and created a formalized food policy network to
inform public policy and private development.
The project helped to rebuild the food system by developing a food access
and food justice action plan and by creating neighborhood food access
maps. These inform both the general public and those involved in rebuilding
efforts. To further include and mobilize the public, NOFFN involved
communities and leaders to conduct neighborhood-based participatory food
assessments. They plan to create sustainable community food projects based
on neighborhood appraisals and regional assets, which contribute to local
food security. NOFFN plans to facilitate three to five targeted community
food projects, which will lead to a sustainable and just food system (World
Hunger Year, 2009; Food Security Learning Center).
The project consisted of two components: site visits and training. Three
2-day site visits provided a forum for participants to discuss problems
and progress in implementing community food systems with agriculture,
transportation, land-use, nutrition, food service, food security, and education
experts. Participants, including legislators, legislative staff, CFP grantees
and local stakeholders, visited sites ranging from small farms, distribution
centers, schools, or other State institutions that purchase local produce, or
farmers’ markets. Preference for selection of programs for the site visits were
given to past or current CFP grantees. The second component was a 1-day
pre-conference training at the 2006 Community Food Security Coalition
(CFSC) annual conference to educate CFP grantees about the legislative
process and about how to incorporate policy into current or future programs.
NCSL partnered with the CFSC throughout this entire project (World Hunger
Year, 2009).
143
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
Youth Programs
Healthy Harvests Initiative: Building Boston’s Food Security from the
Ground Up was funded in 2000 for 2 years at $150,000. The organization
funded was The Food Project in Lincoln, MA. The Food Project made
substantial progress in demonstrating the economic viability of a sustainable
metropolitan food system by linking food production with youth and
community-based enterprise. It opened and equipped a 1,500-square foot
urban center that included a commercial kitchen and involved youth in
business startup and management by working with them to research, design,
and test a value-added food enterprise: farm-fresh salsa. It also tripled the
membership in its Community Supported Agriculture program and increased
both its urban and rural farmland.
The kitchen has become an integral part of The Food Project’s work. Youth
use the kitchen to process and prepare produce for value-added products,
a catering business, and weekly food deliveries. All of the programs
incorporate culinary and nutrition training, expanding the scope of the
involvement of youth in the creation of a local food system. Children from
local elementary and middle schools have classes in the kitchen, which is
also available to local growers for canning food grown in their gardens.
Healthy food production and consumption skills were further modeled in the
community through land remediation, education, and outreach. Under the
grant, an additional urban lot was cleared, cleaned, and put into production
growing herbs and vegetables. Youth further began a series of regular
community lunches and now prepare 11 lunches annually serving over 800
people. The Food Project has also established an education center with a
library of cookbooks and resources on food enterprise development. The
project also runs two farmers markets a week for the entire local community
where youth are employed in marketing and selling the produce they grow
(World Hunger Year, 2009).
“Food for Life” was funded in 2001 at $180,000 for 3 years. The
organization funded is the Youth Farm and Market Project (YFMP) in
Minneapolis, MN. A central objective of the grant to the YFMP was to
help youth both understand and strengthen the local food system (i.e., to
learn to grow food, to plan and prepare nutritious meals, and to increase
community awareness, health, and diversity). To that end, YFMP collected
recipes that emphasized nutritional content, as well as other cooking class
materials. They created and implemented a comprehensive, cultural nutrition
curriculum, which called for the participation of parents and local chefs.
YFMP has grown a variety of cultural foods, specifically Asian and Latin
American, for summer lunches. Meal plans emphasizing the importance of
balance were also designed.
YFMP worked with ethnic farmers of the Twin Cities area to plan purchases
and distribution in low-income housing developments. YFMP staff also met
with school officials to discuss securing more local, organic produce for
school meals. They worked to sell produce to several local businesses and
144
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
restaurants and to secure culturally appropriate foods that would supplement
project-grown foods for summer lunches. Program participants had at least
three opportunities to prepare a meal alongside local chefs. Youth-led harvest
festivals provided recognition, community meals, and awards for project
participants.
Another objective of the grant was to make a wider variety of foods available
to low-income consumers via public housing markets. They assisted
immigrant growers who were able to sell a variety of vegetables and prepared
foods to the project. They also worked with a number of schools on gardens
and on getting local foods into school meals. They ran a youth farmstand at a
senior public housing high rise, which was very popular with senior residents
(World Hunger Year, 2009).
145
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
This project promoted a “field to table” strategy in southern Virginia and
eastern Tennessee to stimulate the development of food and farm-based
economic opportunities for limited resource farmers and entrepreneurs, to
increase access to local foods for low-income households, to enhance the
value-added infrastructure, and to educate the public about a sustainable food
system.
The main needs this project addressed were poor or inadequate nutrition; the
decline in farm incomes and family farming; lack of access to larger, better
paying markets for local farm products; the lack of infrastructure needed
to add value to local farm products; the lack of affordable, locally raised
organic and sustainable farm products; and inadequate research, education,
and technical assistance available to farmers attempting new enterprises and/
or production practices. The overall purpose of this project is to help create a
more just and sustainable food system in southwestern Virginia and northeast
Tennessee, one that can serve as a model for other rural regions around
the Nation (World Hunger Year, 2009; Food Security Learning Center,
Community Food Projects Database).
Urban Agriculture
Integrated Development Through Urban Agriculture was funded in 2002
and 2005 for a total of $184,000. The organization funded was Nuestras
Raices in Holyoke, Massachusetts. The grantee built an urban agriculture
center with classrooms and a community gathering space, as well as a bakery
and restaurant that were later spun off to private entrepreneurs. The grantee
conducted market research and promotion of Puerto Rican specialty produce,
assisted experienced community gardeners in establishing commercial
gardens, and aided youth in establishing a market garden.
A second CFP grant in 2002 allowed Nuestras Raices to build on and expand
its agriculture-based community development work. The organization took
a leadership role in the development of the Holyoke Food Policy Council,
a partnership of 30 members from various sectors of the food system.
The council conducted a visioning/mapping exercise and interviewed
stakeholders to complete a study of the local food system. Nuestras Raices
established two new community gardens at public housing projects with
plots for 6 families and 40 youth. An existing 17-plot community garden
was transformed into a market garden where adults and youth have begun
farming commercially. The organization also conducts a Spanish-language
promotion program for two farmers’ markets, supports a youth-managed
stand at one of the markets, provides technical assistance and shared tools to
residents growing Puerto Rican specialty crops not available elsewhere, and
is developing seed sources for those crops to sell at farmers markets.
146
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
now includes six new market farms, nature trails, an outdoor stage, tropical
flowers and crops, and a farmstand. There are plans for a youth center and
an office on the land as well, in part to reflect the substantial involvement of
community youth in the efforts to purchase the land and to draft a business
plan for the farm operation (World Hunger Year, 2009; Food Security
Learning Center, Community Food Projects Database).
Rural Producers
The Tribal Fish Market Connection Project was funded in 2003 for
$125,000 for 3 years. The organization funded was Ecotrust, Portland, OR.
The project works with the Columbia River Inter-Tribal Fish Commission
(CRITFC) on wild salmon marketing and helps provide more diverse sources
147
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
of revenue and employment to the 400 tribal fisher families on the Columbia
River. The project helped build a business plan and product development
capacity for the four Columbia River treaty fishing tribes represented by
CRITFC—the Confederated Tribes of the Umatilla Indian Reservation, the
Confederated Tribes of the Warm Springs Reservation, the Yakama Nation,
and the Nez Perce Tribe. The tribes are brought together by the struggle to
save the salmon and by shared spiritual traditions. Fishers involved in the
project are selling their “treaty right tribal caught fish” at Portland Farmers’
Market, with a devoted fan base and the capacity to increase their income
fourfold. With Ecotrust spreading the word, consumers are quickly learning
that there are plenty of high-quality regional foods that not only taste great
but can help maintain a way of life closely tied to the identity and spirit of
tribal nations (World Hunger Year, 209).
During 1999, when hog prices for Missouri farmers averaged 28 cents per
pound, Patchwork producers received 43 cents per pound. With the grant,
Patchwork Family Farms was able to double their product line, test value-
added marketing, and provide a “best sellers” product list. To achieve the
goal of expanding the food cooperative system, this project grew the capacity
for marketing and distribution of locally raised produce, meat, and other
local products to low- and moderate-income families in Mid-Missouri. This
expanded marketing capacity allows for local growers and producers to
supply the enlarging cooperative system. Also, the project worked to increase
the number of farmers and grantee chapter members who are growing for
themselves and for sale through the coop program.
To grow the coop system, through increased demand, the Missouri Rural
Crisis Center will develop an outreach campaign targeting rural Missourians
that explains the economic, environmental, social, and public health benefits
of community-based food systems. To further the public’s appreciation of
148
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
local foods and increase or continue interest in coops and local products, they
will also provide education and resources pertaining to the preparation and
processing of locally raised food and related products (World Hunger Year,
2009; Food Security Learning Center, Community Food Projects Database).
Community Gardens
Urban Detroit Agriculture and Education Project received funding of
$150,000 in 2003 for 2 years. The organization funded was Capuchin Soup
Kitchen, Detroit, Michigan. The project provides a haven for low-income
people to socialize with their neighbors, gives access to affordable produce,
and educates people about nutrition and the origins of their food. The project
has evolved to focus on four objectives: launching a Garden Resource
Program, organizing educational opportunities for community and backyard
gardeners, developing the Romanowski Farm Park (a public park centered
on a 5-acre farm in Detroit), and increasing access to nutritious food for all
Detroiters. The project works with 80 community gardens connected to many
grassroots organizations that are critical to the success of the program. This
Detroit project is an example of the extraordinary types of collaborations that
make community food projects work. The project works with the following
collaborators: The Greening of Detroit, Detroit Agriculture Network,
Michigan State University Extension, O.W. Holmes Elementary School,
American Indian Health and Family Services, City of Detroit Recreation
Department, Michigan Association of Community Gardeners, Michigan
Department of Agriculture, Gleaners Food Bank, Michigan Department
of Community Health, Catherine Ferguson Academy, Detroit 4H Center,
Foundation for Agriculture Resources in Michigan and Michigan State
University.
149
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture
The Capuchin Soup Kitchen and its partners have increased food security
throughout Detroit by increasing the number of urban gardeners citywide
and by seeking new partners. The Urban Detroit Agriculture and Education
project grant allowed project partners to form the Detroit Agriculture
Network, which will become its own nonprofit organization focused on food
security (World Hunger Year, 2009).
150
Access to Affordable and Nutritious Food: Measuring and Understanding Food Deserts and Their Consequences
United States Department of Agriculture