007V-A 01 fi7?^.A 3//3
USDA
United States
Department of
Agriculture
Food and
Consumer
Service
Food Retailers in the Food Stamp
Program: Characteristics and
Service to Program Participants
Office of
Analysis and
Evaluation February 1997
A€?f'#l&fS%'
USDA United States
Department of
Agriculture
Food and
Consumer
Service
Office of
Analysis and
Evaluation
Food Retailers in the Food Stamp
Program: Characteristics and
Service to Program Participants
February 1997
Authors:
Richard E. Mantovani
Lynn Daft
Theodore F. Macaluso
Katherine Hoffman
Submitted by:
Macro International Inc.
11785 Beltsvilie Drive
Calverton, MD 20705-3119
Project Director: Richard Mantovani
Submitted to:
Office of Analysis and Evaluation
USDA Food and Consumer Service
3101 Park Center Drive, Room 214
Alexandria, VA 22302
Project Officer: Ken Offerman
This study was conducted under Contract No. 53-3198-3-007 with the Food and Consumer Service,
United States Department of Agriculture, under the authority of the Food Stamp Act of 1977, as
amended.
Food Retailers in the Food Stamp Program: Characteristics & Sen ice to Participants
ACKNOWLEDGMENTS
The authors would like to acknowledge all those who contributed to this study. We would like to
acknowledge Ken Offerman, the Food and Consumer Service (FCS) project officer. In addition to
providing critical input into the various components of the study, his ability to bring together the various
stakeholders in a productive fashion was instrumental in developing this study. Additional important
contributions to this effort from the Office of Analysis and Fvaluation were made b> Steven Carlson and
Margaret Andrews, both of whom provided valuable insights and direction throughout the study and
during the development of the final reports. Others from FCS who made significant contributions
include Suzanne Fecteau, Lynn Jordan, Jill Herndon and Judy Love from the Benefit Redemption
Division, Food Stamp Program, and Debbie Mclntosh. Laurie Hickerson and Chris Casey with the
Supplemental Food Program Division (WIC). A special thanks to Mark Denbaly and Phil Kaufman with
the Economic Research Service (USDA/ERS) who formally reviewed a key early draft final report
providing valuable comments and for their insights on the retailer food industry at critical points during
the study. We also wish to acknowledge the valuable input provided by Zy Weinberg (Director Inner-
City Food Access Program, Public Voice), Celia Slater (Manager, Community Relations. Food
Marketing Institute, FM1) & her staff, and Jean Kinsey (Professor and Director, The Retail Food
Industry Center, University of Minnesota), all of whom participated in the formal review of the first draft
final report.
In addition to the authors, other staff critical to this effort include: Lisa Hammer, whose management of
the retailer survey was critical to the success of this study. Ms. Hammer also conducted site visits that
provided valuable insight on food availability and access in several communities. Sara Sullivan is also
to be commended for her role in both the survey effort and the site visits. Finally, we would like to thank
both Pedro Saavedra and Joseph Steinberg for their valuable contributions to the sample design.
3r-
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants __^_
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY i
INTRODUCTION 1
METHODOLOGY
Data Sources 3
Store-based Measures 4
Analytic Strategy 7
FINDINGS
The National Situation 10
Store Presence at The Local Level 15
Availability and Cost of Food at the Local-Level 21
High Poverty Urban Areas 21
High-Poverty Rural and Mixed Areas 26
Conclusions 30
APPENDIX (Reference Tables)
IV)
Food Retailers in the Food Stump Program: Characteristics & Service to Participants
TABLE OF EXHIBITS AND FIGURES
Table I Profile of Retailers Authorized for The Food Stamp Progra .i,
by Major Store T\ pe 5
Figure 1 The Core Market Basket 6
Table 2 Distribution of Geographic Area. Total U.S. Population, and Population in Poverty,
by Urbanization Level 8
Fable 3 Summary of Availability, Cost, Quality of Foods by Major Store Type Categories,
and Total Redemptions of the Stores in These Categories 10
Table 4 Number and Percent of Stores.
by Store Type and Urbanization Level of Store's Location II
fable 5 Distribution of ZIP Code Areas. Total Population, and Population Living in Poverty,
by Urbanization Level and Availability of Retailers 13
Table 6 Number and Percent of Stores in Urban Areas.
by Store Type and Poverty Level of Store's Location 15
Table 7 Number and Percent of Stores in Mixed Areas,
by Store Type and Poverty Level of Store's Location 16
Table 8 Number and Percent of Stores in Rural Areas,
by Store Type and Poverty Level of Store's Location 17
Figure 2 Market Factors Relating to Predicting Store Location 19
Figure 3 The Average Number of Stores in the Area Controlling for Demographics,
by Store Type, Urbanization and Poverty Level 20
Table 9 Cost Index, by Store Type and Poverty Level, in Urban Areas 22
Table 10 Redemption-Based Costs, by Store Type and Poverty Level, in Urban Areas 23
Table 11 Average Percent of Market Basket Available,
by Store Type, and Poverty Level in Urban Areas 24
Table 12 Extent to Which Supermarkets Have Full Service Departments and Non-Food
Product Lines, by Poverty Levels, and Urbanization Level of Area 26
Table 13 Store-Based Costs, by Store Type, and Poverty Level, in Rural and Mixed Areas ... 27
Table 14 Redemption-Based Costs, by Store Type and Poverty Level, in Rural
and Mixed Areas 27
\M
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
TABLE OF EXHIBITS AND FIGURES
(continued)
Table 15 Average Percent of Market Basket Available,
by Store Type, and Poverty Level and Urbanization Level of Area 28
Table 16 Extent to Which Supermarkets Have Full Service Departments and Non-Food
Product Lines, by Poverty Level, and Urbanization Level of Area 29
Figure A. 1 Full Market Basket Used for Data Collection 32
Table A. 1 Geographic, Population and Retailer Characteristics of ZIP Code Areas,
by Urbanization Level and Poverty Qunitile 33
Table A.2 Regression Coefficients Demonstrating the Effects of Selected Demographic and
Housing Measures on Store Location 34
\l
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Executive Summary
Food Retailers in the Food Stamp Program:
Characteristics and Service to Participants
PURPOSE
This study presents a national assessment of the variety, quality and cost of food available at food
retailers authorized by the Food Stamp Program (FSP). For over 20 years the FSP has been the
cornerstone of the national commitment to protect the nutrition, health, and well being of Americas low-income
families. By design and law, the program seeks to achieve its nutritional goals by working
through "normal channels of trade" - food retailers. It is therefore critical to know whether food stamp
families are in fact able to purchase a variety of quality food at a reasonable price from food retailers
authorized to accept food stamps.
In recent years, researchers and advocates for the poor have argued that access to food of reasonable
quality and price through normal channels of trade may be problematic in low-income urban areas and
sparsely populated rural areas. The concerns have been that the poor pay more for less, that chain
supermarkets have left the inner city, and that food stamp families living in high-poverty urban and rural
areas must buy their food from small stores with limited selection and high prices. The primary question
addressed in this report is: do food stamp families have the same degree of access—not only in terms of
proximity to food stores but also in terms of the quality, variety and price of food available in nearby
stores—as families with higher incomes?
METHODS
We collected information on a market basket of foods from a nationally representative sample of almost
2,400 retailers authorized by the FSP. A market basket was analyzed to calculate three measures
pertaining to the foods offered within each store: the percent of the market basket available for purchase
(a measure of variety), an index of the quality of the items available for sale, and the annualized cost of
purchasing the market basket at the store. In addition, we obtained the complete national listing of all
200,000 food retailers authorized by the FSP and analyzed it to show how the major types of food stores
are distributed geographically1. We linked census demographic data by ZIP Code to the street address for
each store in both the market basket survey and the national listing in order to explore how store
characteristics and service to food stamp participants vary by location in urban or rural ZIP Codes as
well as by location in ZIP Codes where the percent of the population in poverty is high or low.
FINDINGS
The type of store at which food stamp customers buy food is critical, for store type determines the
selection of food available and exerts a large influence on the cost of food. Quality of food does not
vary by type of store. Our results confirm the common belief that supermarkets supply, on average.
1 We use six categories: supermarkets, large groceries, small groceries, convenience stores and/or grocery/gas
combinations, specialty stores (such as meat or produce markets) and "other" (such as general stores, co-ops, route
vendors). .
■v\
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
nearly all food items in a market basket and have the lowest cost of an) store type. Large grocer) stores
have an important role in food access. While a step down from supermarkets, the) are closer to
supermarkets than to other types o\ food retailers on both selection and cost I his is especiall) true in
rural areas where large groceries provide the same level of selection and cost as supermarkets.
People and food stores appear to be located together. As a result lew people lack access to
supermarkets or large groceries. The population in poverty has about the same access to
supermarkets as the general population. We sorted ever) ZIP Code into mutual I) exclusive
categories: one or more supermarkets present, no supermarket but one or more large groceries present,
small stores but no supermarket or large grocery, and no authorized store of any type present
Nationally, only 2 percent of the total population and 2 percent of the population under the povert) line-live
in ZIP Codes with no authorized food stores; °-() percent of the total population and *■>{) percent of the
population under the poverty line live in ZIP Codes u ith at least one supermarket or large grocer)
present.
Store presence in high-poverty areas
The preceding analysis (which sorts ZIP Codes by the type of stores present within them) suggests that
on the whole the food retailing system serves poor and non-poor alike. Mut this finding leaves open the
possibility that specific communities may still have inadequate access \\ e therefore looked specifically
at areas with high concentrations of people in povert).
The average number of supermarkets in high-poverty urban areas is slightly less than in other
urban areas, even when controlling for many of the market factors that influence store placement.
We examined whether population, geographic size of the area, and supply and demand factors that
influence food retailing can account for the number of supermarkets observed in an area (our anal) sis
could not include the effect of zoning regulations or difficulties in assembling parcels of land). The
estimated average number of supermarkets in high-poverty urban areas (0.9 supermarkets) is lower than
the average number in lower-poverty urban areas (1.14 supermarkets). In rural high-povert) areas.
market factors are sufficient to explain the number of supermarkets and large groceries that exist
Availability and cost Qf food in high-poverty urban areas
Food stamp families shopping in high-poverty urban areas do not need to spend significantly more
for food than those shopping in other areas. The cost of our market basket in supermarkets in urban
high-poverty areas is nearly equivalent to stores in lower-poverty areas. When we examine where food
stamp households actually shop, we find that they save money by selecting the stores that they v isit. For
those frequenting supermarkets in high-poverty areas, shoppers save approximately l cents on the dollar.
Food shoppers are able to find nearly the same percentage of our market basket available among
supermarkets in high-poverty urban areas as in other urban areas. Some differences on specific
fresh items were found. Only 33 percent of supermarkets in high-poverty areas cany fresh seafood
compared to 83 percent in other urban areas. Among large groceries, fresh meat is more available in
high-poverty areas than elsewhere; fresh produce slightly less available in high-poverty areas Fresh
produce and meat, however, are available in almost all supermarkets in urban areas regardless of
location. Shoppers can find a high level of acceptable quality food in urban authorized stores, regardless
of store type or location.
Although the cost, availability, and quality of food do not vary between urban superm... kets in
high-poverty and other areas, the total shopping experience does. Supermarkets in high-pov erty
urban areas offer substantially fewer full-service department and non-food product lines than
1/1/
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
supermarkets in other urban areas. In addition, supermarkets in high-poverty urban areas offer 5 to 10
percent less variety in brands and package types than those in other areas.
Availability and cost of food in high-poverty rural areas
Among stores in rural areas, prices were always close to the same in high-poverty areas as in other
areas. This was true both when calculated on a store basis or when adjusted for where participants
actually shop.
Food stamp participants are able to find a slightly higher percentage of our market basket
available among supermarkets in high-poverty rural areas as in other rural areas. Large groceries
in rural areas provide the same level of selection as supermarkets. There is very little difference between
high-poverty and other areas in level of selection. Moreover, in rural areas, shoppers can find
acceptable quality food at virtually every authorized store: quality levels were identical across different
store types and poverty levels.
CONCLUSION
Overall, these findings confirm that the design of the Food Stamp Program—to work through normal
channels of retail trade—effectively reaches low-income populations and provides them with high
quality food at reasonable prices. In most parts of the country, the low-income population can find
supermarkets and large groceries that stock a wide selection of food that meets quality standards at
reasonable prices. Other kinds of stores fill market niches when needed.
About forty percent of the rural population reside in localities without supermarkets or large groceries.
However, this appears to reflect the economics of food retailing. Moreover, the absence of such stores
does not fall disproportionately on the poor: proximity to stores is identical for both the population in
poverty and the total population.1 Finally, in rural areas, the price of the market basket was about the
same among stores in high-poverty and lower-poverty areas.
In urban areas, the number of supermarkets and large grocery stores is lower in high-poverty areas than
in other areas and the shopping experience in supermarkets (as measured by the number of full-service
departments, non-food lines, availability of fresh seafood and variety of package types) is more
restricted. However, there appears to be little effect on the cost of food. The price of our market basket
was either about the same or lower among supermarkets and large groceries in high-poverty areas as
among those in lower-poverty areas. The mix of stores in high-poverty urban areas is characterized by
an extraordinarily high abundance of small groceries with less variety and higher prices than
supermarkets. However, supermarkets exist in those same high-poverty urban areas and. based on actual
food stamp redemption data, food stamp participants shop heavily at those supermarkets and appear to
save about four cents on the dollar compared to supermarkets in other urban areas.
1 Many observers point out that access to transportation may be problematic for low-income families and it is
possible that the transportation burden of living in rural areas falls more on the poor than the non-poor.
in
Jlft
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Introduction
For more than 20 years the Food Stamp Program (FSP) has been the cornerstone of the national
commitment to protect the nutrition, health, and well being of America's low-income families. Serving
low-income families who need assistance to purchase nutritious food, the FSP is the only assistance
program open to virtually everyone who meets income and asset levels. The program helped to provide
food for an average of 25 million Americans each month of fiscal year 1996.
To meet its nutrition goal, it is critical that food stamp families be able to purchase a variety of quality
food at a reasonable price. By design and law, the FSP works through "normal channels of trade," of
which food retailers are the major component. Approximately 200,000 retail food stores are now
permitted to accept and redeem food stamps. Food stores range from large chain supermarkets to small
"mom and pop" groceries. In addition, the program has authorized other stores that offer a single line of
food (e.g., produce markets) as well as those that sell food as a sideline to their primary business (e.g.,
general stores).
However, in the last ten years, low-income, urban areas and sparsely populated rural areas have been
identified as places where access to food through normal channels of trade may be problematic' In low-income
urban areas, the major concern has been that residents do not have ready access to chain
supermarkets within a reasonable distance and thus have to travel to do their shopping or purchase food
from higher-priced, smaller stores within the area that offer less variety and quality. Rural areas are said
to pose similar problems, but on a larger scale since distances required to reach population centers that
have supermarkets are greater, and nearby smaller stores may not be as plentiful as in urban areas.
1. Recent studies include: (1) M. Green, The Poor Pay More for Less. New York City Department of Consumer
Affairs, New York. 1991; (2) L. Ashman, et. al.,. Seeds of Change: Strategies for Food Security for the Inner City.
Southern California Interfaith Hunger Coalition, Los Angeles, 1993; (3) D. M. Ambrose "Retail Grocery Pricing: Inner
City, Suburban and Rural Comparisons," Journal of Business, Vol. 52, No.l, 1979 p. 993; (4) P. M. Morris, et. al.
"Food Security in America: A Study of the Availability and Costs of Food," Journal of Nutrition Education, Vol 24.
No. 1. Jan/Feb Supplement 1992. p.525; (5) E. G.Crockett, et. al." Comparing the Cost of a Thrifty Food Plan Market
Basket in Three Areas ofNew York State," Journal of Nutrition Education, Vol. 24 Jan/Feb Supplement, 1992. p. 765;
(6) P.E. Nelson, Analysis of the Impacts of Food Stamp Redemptions on Food Stores and Regions, Fiscal Year 1978.
Economic and Statistics Research Service, USDA. Technical Bulletin No. 1946, Washington DC. April 1981; and (7)
R. Cotterill. and A. W. Franklin, The Urban Grocery Store Gap. Food Marketing Policy Issue Paper No. 8. Food
Marketing Policy Center, University of Connecticut, April 1995. Another important issue paper that provides an
overview of the issues is "No Place to Shop: An Issue Paper", which is published by Public Voice for Food and Health
Policy, Washington, D.C. February, 1996.
I
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
The purpose of this report is to present a national, authoritative assessment of access of food stamp
participants and low-income Americans to food of good quality and reasonable prices. In presenting the
findings, we first focus on the Nation at-large and examine differences in the ability of different types of
stores authorized by the Food Stamp Program to supply foods of acceptable quality at a reasonable price.
We then examine whether the mix of stores accessible to the low-income population is different than the
mix available to the general population. Next we turn to the local level, highlighting high-poverty urban
and rural areas. Finally, we explore whether local differences in the mix of stores has an impact on the
selection, quality, and cost of foods. To a large extent, the question addressed in this report is: do food
stamp families have the same degree of access—not only in terms of proximity but in terms of assured
quality of services and food—as families with higher incomes'.'
This research is part of a larger Food and Consumer Serv ice initiative to examine access to food from a
variety of perspectives, including case studies of the proximity of food stamp households to authorized
food stores, a conference on access to food, and an analysis of food purchasing patterns of low-income
households.1 I ...s report provides an overview of the access issue, drawing from detailed analyses
presented in a larger companion report and a detailed geographic analysis of access to food stores.
I. Two reports constitute the effort under the Authorized Food Retailer Characteristics Study. The companion report,
which provides a more detailed analysis of the issues discussed here, is the Authorized Food Retailer Characteristics
Study, Technical Report IV, February 1997, Food and Consumer Service., USDA. A second published report on 9
separate case studies of geographic access is: R. Mantovani and J. Welsh; The Authorized Food Retailer Characteristics
Study: Technical Report III, February 1996, Food and Consumer Service. Anothc.- initiative was a conference on access
to food (See R. Koralek Conference on Access to Food: September 18 and 19, 1995. Report of the Proceedings.
November 1996. Food and Consumer Service, USDA.
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Methodology
This section pros ides an overview of the Study's sources of data and key measures.1 We also present a
road map to guide readers through the analytical strategy that underlies our findings.
Data Sources
The study is based on three sources of data. The first is a new. nationally representative survey of 2,378
FSP authorized retailers conducted specifically for this study. Sampled food retailers included every
type of store open to the public and authorized to accept food stamps and ranged from chain
supermarkets to informal food distribution centers.: As described below, this survey collected
information on the prices, variety, quantity, and quality of foods from each store in the sample. The
survey was conducted between April and August 1994. The response rate for the survey was
approximately 95 percent.
The second data source is the Store Tracking and Redemption Subsystem (STARS), a national database
of retailers maintained by the food and Consumer Service (FCS). From STARS we drew a data set
containing the universe of approximately 200,000 food retailers that were open to the public and had
authorization from FCS to accept food stamps at any time during calendar year I993.J Available data
includes store type, street address, and level of sales and food stamp redemptions.
1. A detailed description of methods is available in Appendix C to the Authorized Food Retailer Characteristics
Study: Technical Report IV. op. cit.
2. The survey - as well as the entire study - intentionally exclud -s Alaska, Hawaii and the territories and is
representative of the 48 contiguous States. Because they are not "open to the public", military commissaries and
food wholesalers were also excluded and arc the only categories of authorized "store" excluded from study
databases.
3. A representative sample was selected from a frame of retailers located in the contiguous 48 States and
authorized to redeem food stamps in 1993. A three-stage sampling approach with 40 Primary Sampling Units
(PSUs) was used. Efforts were made to contact all sampled food stores; however only 2,378 stores were in business
and willing to cooperate. A full description of the survey, the instruments, sample design, and data collection
procedures are provided in Appendix B in Authorized Food Retailer Characteristics Study: Technical Report IV, op.
cit.
4. Specifications on the data base supplied for this study can be found in the Retailer Characteristics Study Data
Documentation Code book.
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
The final data source is census demographics for ZIP Code areas. In order to describe retailer
characteristics by community we linked the stores in the first two data sets to a database of census
demographics that allowed us to measure the urbanization and poverty characteristics of the ZIP Code in
which the store is physically located.1
Store-based Measures
Type of Store. The food retailing industry is characterized by a variety of stores that serve distinct
market niches. Because the cost and selection of food may vary greatly by whether one shops at a
supermarket or a gas 'n go convenience mart, our first analytic measure is type of store. The ISP
recognizes 20 different types of stores. Almost 90 percent of the stores, however, are identified as one of
five types: supermarkets, groceries, convenience stores, grocery/gas combination outlets, and specialty
stores. Table 1 provides the definitions used in this study, the approximate number of each type of store
as estimated by national trade organizations and the number of stores authorized by the FSP.
The Food Stamp Program "small to medium" sized grocery store category includes self-declared grocery
stores with less than $2 million in gross sales. Stores in this category range from very small "mom and
pop" stores to larger stores indistinguishable from small supermarkets. Because of this large variation,
we differentiated large grocery stores with gross sales of between $500,000 and $2 million from small
grocery stores with gross sales of less than $500,000.: This distinction is used throughout the analysis
and applies to both the survey and STARS-based retailer data.
1. ZIP Code areas were the smallest geography that would allow us to map retailers to specific communities and
their characteristics. Because these areas are smaller than counties, they allow us to more closely approximate
access patterns within specific communities.
2. In classifying large groceries from small groceries, we selected a gross sales value that could reasonably define
stores that could stand on their own in terms of meeting shopper's food needs. Preliminary work related to the
retailer survey indicated that $500,000 was a reasonable value.
Food Retailers in (he Food Stamp Program: Characteristics & Service to Participants
Table 1
Profile of Retailers Authorized for The Food Stamp Program,
by Major Store Type
Store Type Definition Number Listed With
Major Trade
Organizations'
Number Authorized by
FSP*
Supermarkets Food stores with $2 million or more in annual gross
sales and able to provide a full range of foods
30.450 30 400
Groceries Food stores that can provide a full range of foods
with less than $2 million in annual gross sales In
this study, large groceries (stores with annual gross
sales between $500 000 and $2 million) are
differentiated from small groceries (stores with
annual gross sales of less than $500,000)
42,550 Large Groceries 13,541
Small Groceries 38,042
Convenience
Stores and
Grocery/Gas
Combinations
Stores providing limited range of foods usually
excluding fresh foods These stores are generally
aimed at supplementing larger stores and providing
convenience in terms of their proximity to shoppers
and hours of operation
84,000 76 185
Specialty Stores Stores specializing in one or two product lines such
as produce, meats, or baked goods
18,500 17 352
Other Retailers Includes health food stores, co-op food stores
routes, multi-stall and produce stands, general
stores, and combination stores that sell food in
addition to other goods
23 881
1 Sources for industry estimates are Supermarkets and Groceries Progressive Grocer Annual Report April 1995.
Convenience Stores and Grocery/Gas Retailers Food Institute, Food Retailing Review, 1995, Specialty Stores Food
Marketing Review 1993-94 (AER-678), Economic Research Service, USDA, April 1994
2 Source for authorized retailer figures is Store Tracking and Redemption Subsystem (STARS) December 1993 Total
number of stores equals 199.401 This number excludes authorized stores in Alaska, Hawaii, and the territories as
well as retailers identifying themselves as wholesalers and military commissaries
Source: Food Retailers in the Food Stamp Program Characteristics & Service to Participants, February 1997
Market Basket Measures. To obtain data on the price, quality, and variety of food available in
authorized stores we sent data collectors to each store in the survey with instructions to identify both a
core (42 item) and expanded (I42 item) market basket of foods. Both market baskets included foods in
all major food groups. Selection of items in the baskets was guided by the USDA Thrifty and Low Cost
Food Plans. The items in the core market basket are presented in Figure I while the items in the larger
market basket are presented in Appendix Figure A-l,
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Figure 1
..._
The Core Market Basket
Our measures of food cost and quality are based on the following items:
Fresh Ground Beef Fresh Lettuce Ice Cream Canned Macaroni
Fresh Pork Chops Fresh Tomatoes Eggs Catsup
Fresh Chicken Frozen Orange Juice Whole Wheat Flour Peanut Butter
Fresh Fish Fillets Frozen Potatoes Dry Spaghetti Canned Chicken Soup
Packaged Bacon Canned Applesauce White Rice Canned Spaghetti Sauce
Frankfurters Canned Com Com Flakes Soft Drinks. Cola
Canned Tuna Canned Apple Juice Bread M&Mst>pe candy
Fresh Apples Potato Chips Crackers Sugar
Fresh Bananas Cheddar Cheese Frozen Pot Pie Coffee
Fresh Oranges Stick Margarine Frozen Pizza
Fresh Potatoes Whole White Milk Dry Macaroni & Cheese
Source: Food Retailers in the Food Stamp Program Characteristics & Service to Participants, February 1997.
In each store, data collectors attempted to fill the 142 item market basket, which was used to calculate
the percentage of these items actually available for purchase. The 42 items in the core market basket
were priced and used to calculate the annual cost of purchasing the core items at that store. The items in
the core market basket were also used to measure quality, which we calculated as the percentage of items
that met standards set out by USDA.' Three key measures are used in this report:
• Percent of Market Basket Available - For each store we calculated the weighted percentage of
the expanded market basket available to customers on the shelf when the data collector was in
the store. Because some items are more frequently purchased than others calculating the simple
percentage of all items available would have been misleading. To reflect the differential
importance of each item in the expanded market basket, availability was calculated by using
consumption weights derived for each item available in the market basket.2
1. The use of a core and expanded market basket reflected the greater burden associated with collecting cost quality
and quantity information. For more information on the rationale for specific items selected, please see data collection
forms in Appendix C of the Authorized Food Retailer Characteristics Study: Technical Report IV, op. cit.
2. The consumption weights reflect the portion of the market basket accounted for by each item. It therefore
reflects the importance of the items in the diets of low-income shoppers. The weights were derived from the
1987/1988 Nationwide Food Consumption Survey. The methodology relating to these weights is provided in
Authorized Food Retailer GVnacteristics Study: Technical Report IV, op. cit.
Food Retailers in the Food Stamp Program: Characteristics «& Service to Participants
• Annual Cost of Market Basket—Annual cost was calculated by weighting the price per pound
of each product by its importance to the diet as determined by USDA's 1987/88 Nationwide
Food Consumption Survey. Importance reflected amount purchased by a family of four during a
year. The market basket for computing cost, quality and quantity was the core 42-item market
basket, not the expanded basket used to measure availability. In developing this cost, we
employed a low-cost strategy that used the item representing the lowest per-unit or pound cost
within the product categories specified in Figure I.
• Percent of Acceptable Quality Items—This measure was assessed by the proportion of
acceptable items found in the core market basket. In assessing quality within a particular
product, data collectors were told to attempt to find acceptable items and avoid poor quality
items, if possible. For each food item, we established a quota representing the number of items
that have to be purchased to meet shopping requirements of approximately 10 households.
Within the food item, the quality measure represented the number of acceptable items over the
total number present, up to the number established by the quota. Quality was determined by
guidelines established in a 1975 USDA publication on "How to Buy Food For Enjoyment and
Quality: Recommendations of the United States Department of Agriculture". Quality
information on each item was weighted by its importance to the market basket
Analytic Strategy
Table 1 confirms that the United States Department of Agriculture, through the Food Stamp Program,
has authorized a broad range of stores, covering virtually every category of food retailing and virtually
every store that is a member of one of the trade associations that represent the different aspects of the
food retailing industry. We therefore can eliminate prior Food Stamp Program authorization policy as a
potential influence on access. In the study, we focus strictly on whether the normal channels of food
retail trade adequately serve food stamp clients.
Previous examinations of the access issue imply that inner-city urban areas and poor rural areas have less
access than other areas. Our primary analytic strategy therefore is to sort stores by their geographic
location in areas categorized by varying levels of urbanization and poverty. Then, for the nation and for
each combination of urbanization and poverty we present data on the mix of store types and the market
basket characteristics of those stores.
Food Retailers in thfe-Food Stamp Program: Characteristics & Service to Participants
In the study we utilize ZIP Code areas as our local areas. Every store in both our survey and STARS
databases was linked by the ZIP Code of the store's location to demographic data compiled by the
Census Bureau for that ZIP Code.1 These ZIP Code areas were categorized by urbanization and poverty
level in the following manner.
Urbanization. We utilize three categories based on urbanization level:2
• Urban—stores located in ZIP Codes with an urbanized population of 90 percent or
more.
• Rural—stores located in ZIP Codes with an urbanized population of 10 percent or less.
• Mixed—stores located in the remaining ZIP Code areas.
Table 2 presents demographic information on each of these areas.
Table 2
Distribution of Geographic Area, Total U.S. Population, and Population in Poverty,
by Urbanization Level*
Urbanization
Level
Percent of Total U.S. Land
Area
Percent of Total U.S.
Population
Pnrctnt of Total U.S.
Population In Poverty
Percent of Total Authorized
Food Stamp Retallara
Urban 3 2% 56 1% 52 8% 47 8%
Rural 64 2% 12 4% 14 8% 19 1%
Mixed 326% 31 5% 32 5% 33 1%
Total 100 0% 1000% 100 0% 100 0%
N=29,073
* Excludes Hawaii. Alaska and U S territories
Source: Macro International Inc .Food Retailers in the Food Stamp Program Characteristics & Service to Participants February 1997
1. This procedure has several known limitations. For example, rural ZIP Code areas can be very large and access
to a store located within the ZIP Code may not be easy. In urban areas, ZIP Code areas can be very small: there
may be no supermarket within a particular urban ZIP Code but there may be a few blocks away in the neighboring
ZIP Code. Despite problems such as these, ZIP Codes are the best of all feasible units for geographic analysis. For
the purpose of this study, the potential inaccuracies in specific ZIP Codes will counterbalance each other and do not
diminish the validity of the national picture as the focus of this study. For a fuller discussion of this issue see
Authorized Food Retailer Characteristics Study: Technical Report IV, op. cit.
2. Urbanized population was estimated based on data provided by the 1990 Census. A ZIP Code area could contain
an urbanization mix and therefore reflect both urban and rural experiences.
8
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Poverty. Depending upon the analyses we categorize poverty in one of two ways:
• Quintiles—Within urban, mixed and rural areas, we divided ZIP Code areas into five
quintiles based on their poverty rate. The quintiles are used for that portion of the
analysis that describes the distribution of retailers nationally. The breakpoints (poverty
rates) for each of the quintiles vary slightly from one table to the next. This signified
different poverty rate distributions across the various urbanization levels. A footnote for
each table provides the breakpoints.
• High-Poverty/Other Areas—With regard to the retailer characteristics survey data, we
subdivided the retailers into two groups. The first included stores located in ZIP Codes
where the number of households under the poverty level was equal to or more than 20
percent. These high-poverty areas are equivalent to the poorest quintile used in the
national level analysis.
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Findings
THE NA TIQNAL SITUA TIQN
The type of store at which food stamp customers buy food is critical, for store type determines the
selection of food available and exerts a large influence on the cost of food. In general, the quality of
food is high and does not vary notably by type of store. However, there is some variation in the
quality of fresh produce by store type. Table 3 presents our three key market basket measures by store
type.
T»W»3
Summary of Availability, Cost. Quality of Foods by Major Store Type Categoric*,
and Total Rttemptfontof tha Store* tn Th—Categort—
Store Type
'• Average Percent of
Market Basket '■ . Annual Costof
*^*aaa*»e%^^ ^e>pp*peTW*
Average Percent of
Acceptable Quality
Kama
Pood Stamp Redemptions
Amount
On millions of $)
Pet Of
Total
Supermarket* 95% $871 99% 16,074 2 767%
Large Grocerle* 81% $1,000 98% 1,275 1 6 1%
Small Grocerle* 51% $1,169 97% 1.148.2 5 5%
Convenience Store* and
Grocery/Gas
Combination*
51% $1,303 99% 1,075 9 5 1%
Specialty Store* 20% $1,155 99% 8186 39%
Other Retailer* 29% $1,183 98% 5648 27%
Total 20,9568 100 0%
Source: Food Retailers in the Food Stamp Program: Characteristic* S Service to Participants, February 1997.
The results confirm the common belief that supermarkets supply, on average, nearly all food
items in a market basket and have the lowest cost of any store type.
Large grocery stores, although not providing the food selection and cost advantages of
supermarkets, are closer to supermarkets than to other stores on both selection and cost.
Other types of stores (smaller groceries, convenience stores, grocery/gasoline outlets, specialty
stores, and other stores) have a significantly smaller selection of items available and charge
notably more than supermarkets and large groceries. These stores are commonly perceived to
occupy important niches in the food retailing industry, catering to ethnic food tastes, convenience
or other aspects of the shopping experience.
10
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Quality does not vary substantially across store types for most foods. The variation in quality is
greatest for fresh produce with supermarkets being able to supply quality items for 97 percent of
the fresh produce in the market basket and small groceries being able to supply quality items for
86 percent of fresh produce in the market basket.'
Food stamps are used largely in store types with the largest selection of foods and lowest costs.
Redemption information presented in Table I indicates food stamp households use most of their food
stamps in supermarkets and large groceries, rather than in store categories with lower availability of food
items. Seventy-seven percent of the food stamps are redeemed by supermarkets and 6 percent are
redeemed by large groceries. It should be noted that supermarkets account for 15 percent and large
groceries account for about 7 percent of all authorized stores. In other words, 80 percent of all food
stamps are redeemed in about 20 percent of all stores.
The mix of store types varies considerably across urban and rural areas. Because the availability
and price of food varies so much by store type, it is critical to know whether stores with higher
availability and lower prices are uniformly present throughout the nation. Not surprisingly, there are
substantial urban and rural differences (Table 4).
Table 4
Number and Percent of Storm,
by Store Type and Urbanization Level of Store's Location
Store Type Urban Areas Mixed Areas Rural Areas Total Areas
No. Pa.:' NO. Pet NO. Pit NO. Pet
Supermarkets 14.876 15 6% 12,228 18 6% 3296 8 6% 30,400 15 2%
Large Groceries 5,349 56% 3,675 5 6% 4,517 11 8% 13,541 6 8%
Subtotal 20,225 21 2% 15,903 242% 7,813 204% 43,941 22 0%
Small Groceries 22,978 241% 7,572 11 5% 7,492 19 6% 38.042 19 1%
Convenience Stores and
Grocery/Gas
Combinations
31.809 333% 28,557 43 4% 15,819 41.5% 76185 38 2%
Specialty Stores 9,670 10 1% 5,650 8 6V 2.032 53% 17.352 8 7%
Other Retailers 10,725 11.3% 8,169 12 3% 4,987 13 2% 23,881 12 0%
AH Retailers 95,407 100 0% 85,851 100 0% 38,143 100 0% 199.401 100 0%
Source Food Retailers in the Food Stamp Program Characteristics 8 Service to Participants. February 1997
I We also examined the percentage of stores within each store type category that can supply 90 percent of the market
basket of 10 households with foods meeting quality standards. Virtually ail supermarkets (99.75 percent) meet this 90
percent test, but 7 percent of small groceries and 10 percent of "other" stores do not.
II
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
• Supermarkets constitute a larger share of authorized retailers in urban than in rural areas: they
account for 16 percent of the authorized stores in urban areas but only 9 percent in rural areas. On
the other hand, large groceries account for 6 percent of the authorized stores in urban areas but
almost 12 percent in rural areas.
• Other stores contribute differently to the overall mix of retailers across the three urbanization
categories. For instance, small groceries have a larger relative presence in urban and rural areas,
while convenience stores have a larger relative presence in mixed and rural areas.
• Convenience stores and grocery/gas outlets are the most prevalent regardless of area. In urban areas,
they account for one-third of the stores and in mixed and rural areas they account for more than 40
percent of the stores.
While the mix of food stores is not constant across urban and rural geographic areas, people and
food stores appear to be located together. As a result few people lack access to supermarkets or
large groceries. Somewhat surprisingly, the population in poverty has about the same access as the
general population. We sorted every ZIP Code in the continental United States into the following five
mutually exclusive categories that reflect the presence or absence of authorized supermarkets and large
groceries:
• Two or more supermarkets present in the ZIP Code
• One supermarket present
• No supermarkets, but I or more large groceries present in the ZIP Code
• Only small stores in the ZIP Code, and
• No authorized food stores of any type present in the ZIP Code.
We then looked within each of these categories to see whether the population under the poverty line fares
poorly compared to the general population. The results are provided in Table 5.
• Only 4 percent of the total population in urban areas and just over 3 percent of those below the
poverty line live in urban areas served only by small stores (i.e., with no authorized supermarket
or large grocery). About 1 percent of both the total population and the population in poverty live in
areas with no stores. In urban areas, about 92 percent of the general population and 92 percent of the
population under the poverty line are located in ZIP Code areas served by at least one supermarket.
12
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Table S
Distribution of ZIP Code Areas, Total Population, and Population Living in Poverty,
by Urbanization Level and Availability of Retailers
Urbanization Level and
Availability of Retailers
ZIP Code Areas Total Population Population Living in
Poverty
No. Pet. No. Pet. No. Pet.
Urban Areas
Two or Mort Supermarkets 3 559 59 8% 108 614 577 78 5% 13.190754 78 5%
One Supermarket 1.215 20 4% 19 0/4 365 13 8% 2218 138 13 2%
Large Grxery But No Supermarkets 270 4 5% 3.364.483 2 4% 711 600 4 2%
Smaller Sto'es but No Large Stores 581 9 8% 5.398489 3 9% 556 925 3 3%
No Stores 322 5 5% 1 962051 1 4% 129 720 0 8%
Total 5947 100 0% 138413.965 100 0% 16 837 13" 100 0%
Mixed Areas
Two or More Supermarkets 3 122 46 9% 57,107.355 73 4% 8 085.750 78 2%
One Supermarket 1,473 22 1% 13.065 598 16 8% 1.448 395 14 0%
Large Grocery But No Supermarkets 460 6 9% 2484 642 3 2% 293 669 2 8%
Smaller Stoies But No Large Stores 1.030 15 5% 4,028049 5 2% 439 940 4 3%
No Stores 567 8 6% 1 158 136 1 4% 73.515 0 7%
Total 6652 100 0% 77,843.780 100 0% 10.341,269 100 0%
Rural Areas
Two or More Supermarkets 546 3.3% 3 149 332 10 3% 525 588 11 2%
One Supermarket 2.053 12 5% 7 972,658 26 1% 1,169 241 24 9%
Large Grocery But No Supermarkets 3178 19 3% 7 180,605 23 5% 1,142213 24 3%
Smaller Stores but No Large Stores 6.808 41 3% 9520,590 31 2% 1 542 304 32 8%
No Stores 3.889 23 6% 2,706,677 8 9% 319 853 6 3%
Total 16 474 100 0% 30 529 862 100 0% 4.699.202 100 0%
All An at
Two or More Supermarkets 7.227 24 9% 168.871,264 68 4% 21 802 092 68 5%
One Supermarket 4.741 16 3% 40112,621 16 3% 4 835 774 15 2%
Large Grocery But No Supermarkets 3.908 13 4% 13,029,730 5 3% 2.147482 6 7%
Smaller Stores but No Large Stores 8419 29 0% 18947 128 7 7% 2539169 8 0%
No Stores 4.778 16 4% 5.826,864 2 3% 523088 1 6%
Total 29.073 100 0% 246 787 607 100 0% 31.847.608 100 0%
N=29.073
• Further information on characteristics of the vanous urbanization categories by poverty level is provided in Table A-1 in the Appendix Information
provided gives the population, population density, the area, poverty rate and retailer density of each of the a-eas
Source Food Retailers in the Food Stamp Program Characteristics & Service to Participants. February 1997
13
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
• Mixed areas with only small stores contain 5 percent of the general population and 4
percent of the poor. In mixed areas, the proportion of individuals living in ZIP Code areas with
at least one supermarket exceeds 91 percent for both the general population and the population in
poverty. It must be noted, however, that mixed ZIP Codes tend to be substantially larger than
urban ZIP Codes (79 square miles versus 8) so the similarity of the general and poverty
populations is less certain.1
• Rural areas with only smaller stores present account for 31 percent of the overall
population and a marginally higher 33 percent of the population in poverty. Rural areas
without any stores account for 9 percent of the overall population and 7 percent of the
poor. In rural ZIP Code areas (which average 68 square miles), about 36 percent of the
population live in areas that have a supermarket available within their boundaries. An additional
24 percent of the rural population live in areas that have a large grocery but no supermarket.
Both these latter figures are the same for the rural poverty population and general rural
population.
• Nationally only about 8 percent of the general population and 8 percent of the population
in poverty live in areas with only smaller stores. Only 2 percent of either the general or
poverty population live in areas without stores of any type.
Thus, when viewed from a population perspective, a relatively small proportion of the population seems
to live in areas that have no food stores, and those living in poverty seem to have the same access to
supermarkets or large groceries as the general population (at least at the level of ZIP Code areas).
The preceding analysis (which sorts ZIP Codes by the type of stores present within them) suggests that
on the whole the food retailing system adequately serves poor and non-poor alike. But this finding
leaves open the possibility that specific communities may still have inadequate access. In the next
section therefore we identify urban and rural ZIP Codes where there are very high concentrations of
people in poverty and compare them to areas where there are very few people in poverty. In doing so,
our research interest changes slightly: we are no longer asking whether the normal channels of food retail
trade reach the poverty population as a whole, rather we seek to determine whether there are some who
are left behind by virtue of living in specific geographic areas.
I See Table A-l in the Appendix. In mixed areas, square miles of the ZIP Code area increases as poverty increases but
the number of stores per 5,000 residents also increases.
14
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
STORE PRESENCEAT THE LOCAL LEVEL
In areas with high concentrations of persons in poverty there are relatively more small stores than
in other areas, regardless of urbanization. Tables 6 to 8 provide the share of stores within each
urbanization level by poverty quintile:
• In high-poverty urban areas, supermarkets account for 8 percent of authorized stores
compared to 40 percent in low-poverty areas (Table 6). There are actually more supermarkets
in high-poverty urban areas than low-poverty areas (2,882 vs. 2,384). However, there are 42
times more small groceries in high-poverty urban areas than low-poverty areas (12.991 vs. 307).
As poverty level rises, the shares of all stores accounted for by small groceries increases from 5
percent to 35 percent, respectively. Small groceries are most prevalent in high-poverty urban
areas.
TaW««
Number and Percent Of Stores in ' "ban Areas,
by store Type and Poverty Level of Store's Location
Mara Type
Poverty Rate QuIntUe of ZIP Code in Which Store ii Located"
Lows* Poverty
tes
Second Third Fourth
Highest Poverty
Rates
Total
NO. Pet No. Pet Mo. Pet No. Pet No. Pet No. Pet
Supermarkets 2,384 39 8% 2,760 27 6% 3.354 20 3% 3,496 13 4% 2,882 7 8% 14,877 15 6%
Large Groceries 196 3 3% 443 4 4% 750 4 5% 1,477 5.7% 2.483 6.7% 5.349 5 6%
Subtotal 2.580 43 1% 3.203 32 0% 4.104 24 9% 4973 19 1% 5,365 145% 20 226 21 2%
Small Groceries 307 5 1% 950 9 5% 2,426 14 7% 6,304 242% 12.991 35 2% 22.979 24 1%
Convenience
Stores and
Grocery/Gas
Combinations
2.085 34.8% 3,856 386% 6.586 39 9% 9,096 35.0% 10.186 27 6% 31,810 33 3%
Specialty Stores 440 7 4% 921 9 2% 1,585 9 6% 2,716 10 4% 4.008 10 9% 9,670 10 1%
Other Retailers 571 9.6% 1.068 10 7% 1,800 11 0% 2.934 11 3% 4,352 11 8% 10,725 11 3%
All Retailers 5,983 100 0% 9.998 100 0% 16,501 100 0% 26,023 100 0% 36,902 100 0% 95.411 100.0%
* The median poverty rates by quintile are 2 9 percent. 5 3 percent. 8 3 percent. 13 3 percent, and 26 5 percent.
Source: Food Retailers in the Food Stamp Program Characteristics & Service to Participants. February 1997
15
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
• In high-poverty mixed areas, the representation of supermarkets is about half that in other
mixed areas. There are almost twice the number of supermarkets and three times the number of
large groceries as there are in areas with the lowest levels of poverty. The difference in the
number of small stores between high-poverty and other areas, while still dramatic, is less than in
urban areas. Overall and by store type, there are more authorized retailers located in higher
poverty mixed areas (Table 7). There are about ten times as many small groceries and five times
as many convenience stores and grocery/gasoline outlets in high-poverty areas as in low-poverty
areas. As a result, the share of all stores accounted for by supermarkets decreases from 29
percent to 13 percent as the poverty level increases.
• Tabte7
Number ami Percent of Stores in Mixed Area*,
by Store) Typo and Poverty Level of Store's Location
Store Type
Poverty Rat* Qutntiie of ZIP Code In Which Store la Located*
Lowest Poverty Second - Third Fourth Highest Poverty Total
No Pet No. Pet No Pet No. Pet No Pet No Pet
Supermarket* 1,594 29 2% 2,168 231% 2.524 20 1% 3,047 180% 2.895 13 4% 12,229 18 6%
Large
Groceries
309 5.7%
34 9%
540 5 8% 717 5 7% 976 5.8% 1,133 5 3% 3.675 5 6%
Subtotal 1,903 2,708 28 9% 3,241 25 8% 4 023 23 8% 4 028 18 7% 15 904 24 2%
Small
Groceries
309 57% 724 7.7% 1,194 95% 1,835 10 9% 3,510 163% 7.572 11 5%
Convenience
storti and
GroceryrOas
Combination!
2.130 39 1% 4,013 428% 5,422 43 1% 7,388 437% 9.604 44 6% 28.559 43 4%
Specialty
Storet
437 8 0% 789 84% 1,154 9 2% 1.445 85% 1.825 8 5% 5650 8 6%
Other Retailer* 675 123% 1.137 122% 1,568 12 4% 2.222 131% 2.567 11 9% 8.170 123%
All Retailers 5.454 100 0% 9.371 100 0% 12,579 100 0% 16.913 100 0% 21.534 100 0% 65.855 100 0%
* The median poverty rates by quintile
Source Food Retailers in the Food S
are: 4.2 percent. 8.0 percent, 11 8 percent. 16 3 percent, and 24 2 percent
amp Program: Characteristics & Service to Participants. February 1997
16
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
• In high-poverty rural areas, the representation of supermarkets is again less than in other
areas. However, the difference between high-poverty and other areas in supermarket
representation is not as large as the difference in urban areas. There again are more small stores
in high-poverty areas. As a result, supermarkets account for 15 percent of all the stores in low-poverty
areas and 6 percent in high-poverty areas (Table 8). Large groceries account for almost
twice as many of the stores in low-poverty areas as they do in high-poverty areas. As in urban
and mixed areas, small groceries, and convenience stores and grocery/gas outlets have a larger
share in higher poverty rural areas. Across all store types, there are greater similarities in store
mix between low and high-poverty areas in rural areas than in urban areas.
Tables
Number and Percent of Stores in Rural Areas,
by Store Type and Poverty Level of Store's Location
Store Type
Poverty R.te Quirtlle of ZIP Code in Wilch Store It Located"
Lowest Poverty Second TWftl Fourtti HigheetPoverty Total
No. Pot No. Pet No. Pet No. Pet No. Pet NO. Pet
Supermarket* 751 14 6% 608 10 3% 633 9 6% 619 7 4% 685 5 6% 1,304 6 4%
Large
Groceries
733 14 2% 895 15 2% 929 14 1% 996 12 0% 964 7 9% 1.960 9 6%
Subtotal 1.484 28 8% 1,503 25 5% 1.562 23 7% 1,615 19 4% 1,649 135% 3.265 16 0%
Small
Groceries
692 134% 954 162% 1,174 17 8% 1,829 196% 3,043 24 9% 4.673 22 8%
Convenience
Stores and
Grocery/Gas
Combinations
1.979 38 4% 2.270 38 6% 2.630 39 9% 3.567 42 9% 5,373 44 1% 8941 43 6%
Specialty
Stores
321 6 2% 381 6 5% 402 6 1% 419 5 0% 509 4 2% 928 4.5%
Other Retailers 675 132% 768 13 2% 827 125% 1,094 131% 1,623 13 3% 2,717 13 1%
All Retailers 5.151 100 0% 5,876 100 0% 6.595 100 0% 8.324 100 0% 12.197 100 0% 20.524 100 0%
• The median pov
Source Food Re
wty rates by quinble are 6 9 percent. 11 1 percent. 14 6 percent. 19 0 percent, and 27 5 percent
tailers in the Food Stamp Program Charactenstics & Service to Participants, February 1997
17
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Because of the economics of food retailing, some areas are less able than others to sustain
supermarkets and large groceries. However, the average number of supermarkets in high-poverty
urban areas is slightly less than in other urban areas, even when controlling for many of the
market factors that influence store placement. Counts of the number of authorized supermarkets and
large groceries by urbanization, and by poverty level do not adequately explain the greater presence of
supermarkets in some areas over others. We investigated one possible explanation for this phenomenon:
that the differences reflect the economics of the food retailing industry. This section details our findings.
To investigate this possibility, we examined whether population, geographic size of the area, and other
market factors that influence food retailing can account for all or part of the differences we found.
Because many factors come into play at once, we employed a Poisson regression framework.1 The
regression is set up to draw conclusions about the number of stores present in each of six geographic
areas (our three urbanization levels crossed against two poverty levels). In these analyses, we have
modified our approach with regard to defining poverty level. Instead of quintiles. we use two levels
with high-poverty areas defined as those with a 20 percent or greater poverty rate. This split contrasts
the very poorest communities with other communities. The regression provides the mean number of
stores that should be expected to exist in each area based on the factors specified in Figure 2.
To keep our findings comparable with those in the previous section (which utilized poverty data
available at the ZIP Code level), we limited the regression to variables available for ZIP Code areas from
the 1990 Census. Therefore the equations do not reflect several important influences—such as zoning
restrictions, the availability of food wholesalers and distributors, and the ability to assemble parcels of
land within specific geographic locations. The regression nonetheless enables us to approximate the
degree to which areas differ in supermarket and retailer availability after controlling for many supply and
demand factors. These factors include population as well as other factors, and thus in effect provide an
expected retailer density measure.
1. Poisson regression was used to estimate the number of supermarkets and large stores separately. A Poisson
restriction was needed due to address the highly skewed nature of the store count data in which most cases had a value
of zero.
18
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Figure 2
Market Factors Relating to Predicting Store Location
Population Based Factor*
I. Population
2 Population < 20 Years
3. Population > 65 Years
4. Household Size
5. Households with children
6. Access to vehicle
7. High School Graduates
8. Female Headed Households
Male Headed Households
Single Persons
9 Hispanics, Blacks, Asian.
Foreign
general measure of demand
included because young generally consume more food than the elderly
included because elderly food intake and shopping differs from the rest ol the population
included because large households can achieve economies of scale in shopping, influencing
demand for food separately from population
included because the food purchases of households with children differ substantially from
households without children
included because vehicles expand individuals shopping choices beyond the local area and
is therefore a potential influence on local demand for food
used as a proxy for education which can affect shopping preferences
used to reflect the effect in how household composition affects management
of food shopping decisions
used to indicate preference variables that can affect the type of stores frequented
Household or Community Context Variables
The following variables are used to describe the population of the community identified by the ZIP Codes. These factors may relate to
supply and to demand, but mostly provide a context for describing the community.
1. Average No. of Rooms
2 Average time to commute
3. Use of public transportation
4. Area
5. Population Squared
6. Absence of Plumbing
7. Median Rent
S. Employed in Services
9. Employed as an Operator
Interaction Terms
density of residential housing development
used as proxy for efficiency of transportation infrastructure
used as proxy for efficiency of transportation infrastructure
general measure likely to be associated with the supply of food
included to allow for economies of scale. As population increases relative to geographic
areas, at the higher density levels store size and efficiency levels car. increase,
used to approximate the state of the infrastructure in the area and therefore the case/difficulty
of building or maintaining stores
used to approximate the cost of land in the area and therefore the cost of doing business
used to approximate the local labor supply and, indirectly the cost of doing business in the
area
also used to approximate the local labor supply and indirectly, the cost of doing business in
the area
To capture the outcome we are most interested in—the joint effect of urbanization and poverty on the number of supermarkets and large
groceries—we included five interaction terms. The five included terms are: Urban High-poverty. Mixed High-poverty. Rural High-poverty,
Urban Other, Mixed Other The effect including the preceding five interaction terms in the regression is that the regression
intercept—i.e., the "constant" in Table 5—is "normed on Rural Other (i.e., non-high-poverty) areas
Source: Food Retailers in the Food Stamp Program Charactenstics & Service to Participants. February. 1997
19
Food Retailers in tl.t Food Stamp Program: Characteristics & Service to Participants
Figure 3 presents a graphical depiction of the coefficients estimated for each of the urbanization and
poverty level variables. The actual regression results are presented in Table A.2 in the Appendix.
• jstimated average number of supermarkets in high-poverty urban areas (0.9 stores) is lower than
the average number in lower-poverty urban areas (1.14 stores). The regression coefficient are
ally significant.
• t number of estimated supermarkets is larger in high-poverty mixed areas than in lower-poverty
mixed areas; and there is virtually no difference in the number of supermarkets in high-poverty and
lower-poverty rural areas.
Figure 3
The Average Number of S'ores in the Area Controlling for Demographics,
by Store Type, Urbanization and Poverty Level
2 5
2
o5
*1.5
i§Z
1 ■
a5
< 0.5 •
Suptrmiriciti Laroe Stores
Urban High-Poverty Areas
Urban Other Areas
Mixed High-Poverty Areas
^ Mixed Other Areas
Rural High-Poverty Areas
[iji; Rural Other Areas
Source Food Retailers in the Food Stamp Program Characteristic* & Service to Participants. February 1997
20
tm
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Looking at the regression for all large stores (i.e., supermarkets plus large groceries), the coefficient in
urban high-poverty areas (1.36 stores) is slightly lower than the average number of large stores in lower-poverty
urban areas (1.5 stores). Again, both of the regression coefficients are statistically significant.
These results mean that the market factors measured in our model do not sufficiently explain the fewer
number of supermarkets and large groceries we find in urban high-poverty areas.1
In rural high-poverty areas, the coefficient was not statistically significant, suggesting that the market
factors in our regression are sufficient to explain the number of supermarkets and large groceries that
exist.
4WUWJXUmCOSTQF FQQP 4 T THE WCAl LEVEL
The previous sections document that supermarkets and large groceries account for a relatively smaller
share of all stores in high-poverty areas than other areas and that in high-poverty urban areas there are
fewer supermarkets and large groceries than expected based on a model of many factors that influence
the economics of food retailing. This raises the possibility that food stamp participants living in high-poverty
areas may be at a disadvantage either because prices may be higher and/or the variety of food
available for purchase narrower. The following analyses compare authorized food retailers in high-poverty
areas to retailers in other areas in terms of costs, selection, quality and services provided.
Because the data for the coming analyses derive from our nationally representative survey of 2,378
retailers, we again collapse our categorization of poverty areas from five to two to improve the validity
of comparisons. Therefore, high-poverty areas will now be defined as those areas with a poverty rate of
20 percent or more. This is very close to the areas defined as the fifth quintile in the previous section.2
High-Poverty Urban Areas
Food stamp families shopping in high-poverty urban areas do not need to spend significantly more
for food than those shopping in other areas. The cost of our market basket in supermarkets in
1. In our model we were unable to control for some factors related to the cost of doing business (such as rent) and
other difficulties in establishing viable stores (e.g., land use restrictions, land accumulation), which may account for
some of the differences between high-poverty and other urban areas.
2. The high-poverty areas defined in these next tables are roughly equivalent to the fifth (high-poverty) quintile used
previously. It also corresponds to one of the definitions used by HUD to qualify areas for the Enterprise
Zone/Empowerment Community Program.
21
Food Retailers in the Food Stamp Program: Characteristics A Service to Participants
urban high-poverty areas is nearly equivalent to stores in lower-poverty areas, and the cost in the
"other store" category in high-poverty urban areas is lower than similar stores in lower-poverty
areas. Table 9 presents the costs (in dollars) of purchasing the market basket over or under the cost of
purchasing the market basket in a typical supermarket.1
• Among supermarkets in urban areas, the cost of our market basket is 2 percent higher in high-poverty
areas than in other areas. Among supermarkets nationally, the cost of our market basket
is 4 percent lower in high-poverty areas than in other areas
• Among large groceries (both urban and national), prices are lower at stores in high-poverty areas
than in stores in other areas.
TabJas
Cost Index, by Store Type and Poverty Laval, in Urban Areas
l" .parmaricet Coat -1 00)
I MoJ^a>ovaityAiaia Other Area* ■ Ai Arm
Urban Areas
Supermarket* 1.02 1.00 100
Large Groceries 1.13 1.27 1.24
Other Stores 1.41 1.41 1.46
AH Areas preludes Urban, Mxed and Rural Areas)
Supermarkets .96 1.00 1.00
Large Groceries 1.09 1.16 1.15
Other Store* 1.37 1.43 1.42
N-2.378
Source: Food Retailers in trie Food Stamp Program: Characteristics & Service to Participants, February 1997
The costs computed for each of the major store types above assume that shoppers are equally likely to
purchase an item from any store. This "store-based" cost therefore does not reflect the actual purchase
patterns of FSP recipients as represented by the level of food stamp redemptions within individual stores.
Redemption-based costs consider where food stamp households shop.
1. The cost measure was calculated as an index with the cost of the market basket at supermarkets used as a base.
Costs at other types of stores were calculated as a percentage of this base. The cost index provides several
advantages over reporting actual costs. First, because the market basket on which price information is collected is
only a partial and selective version of a list of foods that would be actually purchased, the cost should not be the
focus. The index provides information on relative costs. Second, the index provides a method to avoid price
imputation when stores in a certain category do not carry a line of food. In this case, the index is formed only from
those items available in all stores within that category. The supermarket base then consists of those items that are
present in the comparison category.
22
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Therefore, to examine how choice of store might alter the comparison, we calculated cost weighted by
the redemptions of food stamps in the sampled stores and indexed the result to the store-based measure.
This redemption-based measure reflects the savings (or additional costs) experienced by participants
using some authorized stores within a given type over others. The redemption-based costs for urban
areas are presented in Table 10. The redemption-based costs are indexed on overall supermarket cost
presented in Table 9. Thus, all comparisons reflect the percentage saved by participants by shopping in
particular types of stores.
When we examine where food stamp households actually shop, we find that they save money by
selecting the stores that they visit. For those frequenting supermarkets in high-poverty areas,
shoppers save approximately 4 cents on the dollar. Food Stamp Program households are
apparently using food stamps in less expensive supermarkets and large groceries. Even for small
stores, redemption-based costs are less than store-based costs.
Tabs* 10
Rederr.ption-Basad Costa, by Stars Type and Poverty Lav*},
<S»<M*-Ba*ed Supermarket Cost-1.00)
in Urban Areas
High-Poverty Area* Other Areas All Areas
Urban Area*
Supermarkets 0.94 0.M 0.96
Large GroctrlM 1.0* 1.17 1.12
Other Store* 1.27 1.39 1.J1
All Areas (Includes Urban. Mixed and Rural Areas)
Supermarkets .94 .96 .96
Large Groceries 1.0S 1.0* 1.06
Other Stores 1.27 1.32 1.30
N-2.37J
Source: Food Retailers ir
1997
the Food Stamp Program Characteristics & Service to Participants February
In urban areas, redemption-based costs in supermarkets are slightly lower in high-poverty areas. This is
reversed from the store-based cost figures, in which slightly higher costs are detected in high-poverty
areas. The redemption-based cost in large groceries was still higher than in supermarkets in urban areas,
but not as high as in large groceries in general.
23
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Food shoppers are able to find nearly the same percentage of our market basket available among
supermarkets in high-poverty urban areas as in other urban areas. Among large groceries
availability of items in the market basket was marginally lower in high-poverty areas than other
areas.
Table 11 gives the average percent of the market basket filled in stores located in urban areas:
• Whereas supermarkets, on average, were able to supply more than 95 percent of the market basket,
large groceries could supply just over two-thirds, and other stores were able to supply two-fifths.
• When all food stores are considered regardless of type (and regardless of redemptions), food stores
in high-poverty areas could supply 46 percent of the market basket compared to 57 percent for food
stores in other areas. This reflects the greater number of smaller stores in high-poverty areas.
Table 11
Average Percent of Market Basket Available,
by Store Type, and Poverty Level in Urban Areas
Store Type High-Poverty Areas Other Areas All Areas
Urban Areas
Supermarkets 94% 96% 96%
Large Groceries 67% 70% 69%
Other Stores 40% 44% 43%
All Stores 44% 55% 52%
AHAreas
Supermarkets 96% 95% 95%
Large Groceries 78% 82% 81%
Other Stores 41% 45% 44%
All Stores 46% 57% 54%
N 2.378
Source Food Retailers in the Food Stamp Program Characteristics & Service to Participants February
1997
Shoppers can find a high level of acceptable quality food in urban authorized stores, regardless of
store type or location. The lowest average proportion of foods in the market basket of acceptable
quality (97 percent) is found in the "other store" category located in high-poverty urban areas.
24
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
While the overall percentage of the market basket available did not substantially vary according to
a store's location in high-poverty or lower-poverty urban areas, some differences can be noted in
the availability of fresh meat, fresh seafood and fresh produce.
• Fresh meat was available in all urban supermarkets. It is found in 92 percent of the large groceries in
high-poverty areas and 75 percent of the large groceries in other urban areas. Fresh meat was found
in 45 percent of the other stores in high-poverty areas and 38 percent of the other stores in other
areas.
• Except for supermarkets, fresh seafood is generally unavailable in most stores in high-poverty areas.
Only 14 percent of the large groceries and 6 percent of the other stores carry this item. Even among
supermarkets, only 33 percent in high-poverty areas carry fresh seafood. This compares to 83
percent in other urban areas.
• Fresh produce is available in almost all supermarkets in urban areas in both high-poverty and other
areas; it is carried in about 85 percent of the large groceries in high-poverty areas and 95 percent of
the large groceries in other areas.
Although the cost, availability, and quality of food do not vary between urban supermarkets in
high-poverty and other areas, the total shopping experience does. Supermarkets in high-poverty
urban areas offer substantially fewer full-service departments and non-food product lines than
supermarkets in other areas. Table 12 displays these data for supermarkets in urban and ail areas:
• Supermarkets in high-poverty urban areas have half the number of full-service departments as
supermarkets in other urban areas.
• Moreover, supermarkets in high-poverty urban areas offer about a fourth fewer non-food product
lines than supermarkets in other areas.
• Finally, supermarkets in high-poverty urban areas offer 5 to 10 percent less variety in brands and
package types than those in other areas.'
1 With regard to variety in brands, stores in high-poverty areas offer an average of 1.90 different brands while those
in other areas offer an average of 2.05 different brands. The corresponding figures for package types are 2.11 and 2.31.
25
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Table 12
Extant to Which Supermarkets Have Full Service Departments and Non-Food Product Lines,
by Poverty Levels, and Urbanization Level of Area
Htgh-Poverty Araes Other Areas AS Are**
Urban Area*
Full Service Department* 1.7 3.5 3.4
Non Food Product Una* 7.9 10.1 9.9
All Area* (Include* Urban, Mixed and Rural Area)
Full Service Department* 2.2 3.1 3.0
Non Food Product Line* 9.3 10.0 «0.0
N=2.478
Source Food Retailers in Ihe Food Stamp Program Characteristics & Service to Participants. February 1997
While food stamps can only be used for food, food stamp participants use their cash income for
non-food necessities. Because they are less able to do so in high-poverty urban area
supermarkets than those in other areas, urban food stamp participants may perceive high-poverty
area supermarkets as less satisfactory than those in low-poverty areas.
High-Poverty Rural and Mixed Areas
Among stores in mixed and rural areas, prices were always lower or close to the same in high-poverty
areas as in other areas. This was true whether calculated on a store basis or adjusted for
where food stamp participants actually shop (redemption basis).
• Market basket costs in rural supermarkets are less in high-poverty areas than other rural areas. This
is true when calculated on either a store basis (Table 13) or a redemption basis (Table 14).
• Market basket costs among supermarkets in mixed areas are about the same for high-poverty and
other areas.
• In rural areas, the market basket in large groceries in high-poverty areas had nearly the same cost as
supermarkets.
26
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
TaWeU
Store-Baaed Costs, by Store Type, and Poverty Level, in Rural and Mixed Area*
(Suparmarkat Coat «1.00)
High-Poverty Areas Other Are as AH Area*
Mixed Area*
Supermarkets 0.93 0.9S 0.98
Large Groceries 1.14 1.13 1.13
Other Store* 1.35 1.43 1.41
Rural Areat
Supermarkets 0.93 1.05 1.04
Large Grocerlei 0.91 1.11 1.09
OtherStores 1.31 1.34 1.34
All Area*
Supermarkets 0.96 1.00 1.00
Large Groceriee 1.09 1.16 1.15
Other Stores 1.37 1.43 1.42
N-2,378
Source: Food Retailers
February 1997
in the Food Stamp Program: Characteristics & Service to Participants,
':'.">::.' toy Store
TaWa14
Redemption-Based Coeta.
Ty pa and. Poverty Laval, in Rural and Mixed Areas
(Store-Based Supermarket Coat-1.00)
High-Poverty Arm j Other Area* All Area*
Mixed Areas
Supermarkets 0.94 0.96 0.96
Large Groceries 1.05 i.oa 1.06
Other Stores 1.27 1.32 1.30
Rural Areas
Supermarkets 0.93 0.96 0.95
Large Groceries 0.90 1.04 1.02
Other Stores 1.21 1.29 1.28
iMl Areas (Includes Urban, Mixed and Rural Areas
Supermarkets 0.94 0.96 .96
Large Groceries 1.05 1.06 1.06
Other Stores 1.27 1.32 1.30
N-2.378
Source: Food Retaito
February 1997
s in the Food Stamp Program: Characteristics & Sen ice to Participants,
27
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Food shoppers are able to find a greater percentage of the market basket in high-poverty rural
supermarkets than in other rural areas. Supermarkets in high-poverty mixed areas also can
supply a slightly higher proportion of the market basket than supermarkets in other mixed areas.
Large groceries in high-poverty rural areas also provide more of the market basket than large
groceries in other rural areas. Large groceries in rural areas provide about the same level of
selection as supermarkets.
Table 15 gives the average percent of the market basket filled in stores located in mixed and rural areas:
• In mixed areas, supermarkets were able to supply 95 percent of the market basket, while large
groceries supplied 84 percent and "other" stores supplied 42 percent. When all food stores are
considered, those in high-poverty areas could supply 47 percent of the market basket compared to
56 percent in other areas.
• In rural areas, supermarkets and large groceries were able to supply about 90 percent of the market
basket and "other" stores were able to supply 50 percent. Again, when availability of the market
basket is considered apart from store type, those stores in high-poverty areas could supply 49
percent of the market basket compared to 63 percent in other areas.
Tablets
Average Percent of Market Basket Available,
by Store Type, end Poverty Level and Urbanization Level of Area
High-Poverty Areaa Other Areas AH Areas
Mixed Areas
Supermarkets 97% 94% 95%
Largt Grocartaa n% 64% 84%
Other Store* 39% 43% 42%
All Store. 47% 56% 54%
Rural Araas
Suparmarkata 97% 91% 91%
Larga Grocariaa 95% 90% 90%
Other Storts 44% S2% 50%
All Stores 49% 63% 60%
All Areas (Includes Urban, Mixed and Rural Areaa)
Supermarkets 96% 95% 95%
Larga Groceries 78% 82% 81%
Other Stores 41% 45% 44%
All Stores 46% 57% 54%
N=2.378
Source: Food Retailers in ttle Food Stamp Program Characteristics & Service to Participants. February 1997
28
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
In rural and mixed areas, shoppers can find acceptable quality food at virtually every authorized
store: quality levels were identical across different store types and poverty levels.
With the exception of seafood, fresh food is available among large groceries and supermarkets in
rural and mixed areas.
• Fresh meat is available in more than 96 of the supermarkets and large groceries in rural areas. In
mixed areas, fresh meat is available in more than 97 percent of the supermarkets and 90 percent of
the large groceries. There are only slight differences between high-poverty and other areas.
• Fresh seafood is generally unavailable in most rural area stores. This item is available in just half of
the supermarkets in high-poverty rural areas and about one-third of the supermarkets in other areas.
Outside of supermarkets, it is not generally available.
• Fresh produce is available in over 95 percent of the supermarkets and large groceries in mixed and
rural areas. It is also available to a larger degree in smaller stores in rural areas than mixed areas. In
rural areas, approximately 80 percent of the stores carries fresh produce.
Unlike urban areas, the shopping experience in supermarkets is similar in high-poverty and other
rural areas (Table 16). The number of full-service departments and non-food product lines offered
by supermarkets in rural and mixed areas does not vary by location.
Table 16
Extent to Which Supermarkets Have Full Service Departments and Non-Food Product Lines,
by Poverty Level, and Urbanization Level of Area
H^h-Poverty
Areas
Other Are** All Area*
Mixed Areas
Full Service Departments 2.7 2.9 2.9
Non Food Product Lines 10.2 10.2 10.2
Rural Araaa
Full Service Departments 2.2 2.1 21
Non Food Product Lines 9.3 9.4 9.4
All Area* (Include* Urban, Mixed and Rural Area*)
Full Service Department* 2.2 3.1 30
Non Food Product Line* 9.3 10.0 10.0
N-2,478
Source: Food Retailers in the Food Stamp Program Characteristics S Service to Participants, February 1997
29
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
CONCLUSIONS
Overall, these findings confirm that the design of the Food Stamp Program—to work through normal
channels of retail trade—effectively reaches low-income populations and provides them with high
quality food at reasonable prices. In most parts of the country, the low-income population can find
supermarkets and large groceries that stock a wide selection of food that meets quality standards at
reasonable prices. Other kinds of stores fill market niches when needed.
About forty percent of the rural population resides in localities without supermarkets or large groceries.
However, this appears to reflect the economics of food retailing. Moreover, the absence of such stores
does not fall disproportionately on the poor. Proximity to stores is identical for both the population in
poverty and the total population.1 Finally, in rural areas, the price of the market basket was uniformly
lower among stores in high-poverty areas.
In high-poverty uroan areas, the number of supermarkets and large grocery stores is lower than predicted
by a model of retailing supply and demand variables. In addition, the shopping experience in
supermarkets (as measured by the number of full-service departments, non-food lines, and availability of
fresh seafood, and in the variety of brands and packaging available) is more restricted. However, there
appears to be little effect on the cost of food. The price of our market basket was either about the same
or lower among supermarkets and large groceries in high-poverty areas as among those in non-poverty
areas. The mix of stores in high-poverty urban areas is characterized by an extraordinarily high
abundance of small groceries with less variety and higher prices than supermarkets. However,
supermarkets exist in those same high-poverty urban areas and, based on actual food stamp redemption
data, food stamp participants frequent those supermarkets in high numbers and appear to save about four
cents on the dollar compared to supermarkets in other urban areas.
I. Many observers point out that access to transportation may be problematic for low-income families and it is possible
that the transportation burden of living in rural areas falls more on the poor than the non-poor. As a recent report has
indicated (R. Mantovani and J. Welsh; Authorized Retailer Characteristics Study: Technical Report III. February 1996,
Food and Consumer Service, USDA), most individuals live in or near to towns or small cities in rural areas.
30
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
APPENDIX
Reference Tables
3)
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Figure A-l
Full Market Basket Used for Data Collection
Fresh/Perishables
Apples
Bananas
Green Beans
Broccoli
Cabbage
Carrots
Cantaloupe
Celery
Corn
Cucumbers
Orange Juice
Lettuce
Onions
Oranges
Peaches
Potatoes, White
Squash
Tomatoes
Bacon
Roast Beef
Ground Beef
Cold Cuts
Chicken Parts
Chicken Whole
Fish Filets
Frankfurters
Ham, Not Canned
Pork Chops
Pork Sausage
Fresh Turkey
Butter
Cheddar Cheese
Cottage Cheese
Eggs
Soft Tub Margarine
Skimmed Milk
Yogurt
Donuts/Pastry
White Bread
Other Bread
Fresh Meat Pot Pie
Fresh Pizza
Frozen
Green Beans
Broccoli
Carrots
Com
Fruit
Orange Juice
Onions
Peas
Potatoes
Ground Beef
Chicken
Fish Filets
Breaded Fish
Ham
Pork Sausage
Turkey
Yogurt
Bread, Any Type
Grain Based Breakfast Foods
Sweet Baked Goods
Chicken/Beef Dinner
Meat Pot Pie
Ice Cream
Macaroni and Cheese
Macaroni & Meat
Frozen Pizza
Canned/Bottled
Applesauce
(ireen Beans
Cabbage or Sauerkraut
Carrots
Corn
Apple Juice
Orange Juice
Tomato Juice
Onions
Oranges
Peas
Peaches
Potatoes. White
Squash
Tomatoes
Fish
Frankfurters
Ham
Poultry
Pork Sausage
Tuna
Baked Beans
Canned Beans
Catsup
Macaroni and Sauce
Dry Roasted Peanuts
Peanut Butter
Soup, with Meat
Soup, Non-Meat
Soup Chicken Noodle
Spaghetti Sauce, Meatless
Hydrogenated Vegetable Fat
Jelly
Mayonnaise
Canned Whole Milk
Milk(Skim/Lowfat/Evaporat
ed
Salad Dressing
Diet Soft Drinks
Non Diet Soft Drinks
Map t or Corn Syrup
Vinegar
Dried
Fruits
Peas, Beans
Potato Chips
White Potatoes
Fish
Eggs
Skim/l.owfal Milk
Bran/Wheat Cereal
Rich/Corn Cereal
Presweetencd Cereal
Unsweetened Corn Flakes
Cookies
Cornmeal
Soda Crackers
All Purpose Flour
Whole Wheat Flours
M & M Type Candy
Oatmeal
Macaroni
Popcorn
Salt
Spaghetti Dry
White Sugar
White Rice
Macaroni and Cheese Dinner
Pizza Mixes
Soup Mixes
Coffee Regular
Coffee Instant
Peeper/Spices
Powdered Ades
Salad Dressing Mixes
Tea
* This market basket was assembled to assess availability along several dimensions of variety. Although the
market basket is not representative, it does include the item categories accounting for a large proportion of food
purchases. It should be noted that in some cases, items may be difficult to find in many stores (dried fish).
Including such items allow us to distinguish between different levels of variety across different store types. For
more information on how information on these products were used for the analysis, see Appendix B of Authorized
Food Retailer Characteteristics Study: Technical Report IV. February 1997. The Food and Consumer Services,
USDA.
Source Food Retailers in the Food Stamp Program Characteristics & Service to Participants, February 1997.
32
Table A-1
Geographic, Population, and Retailer Characteristics of ZIP Code Areas,
by Urbanization Level and Poverty Quintile
Urbanization
Level
Poverty
Quintile
Median Area
(Square Miles)
Median
Population
Mt jian Density
(Persons P«r Sq.
Mile)
Median
Poverty
Level
Supermarkets Large Stores Total Stores
Median
Number
Median
Density*
Median
Number
Median
Density"
Median
Number
Median
Density*
Urban 0-20% 9 16804 1911 29 2 0 09 2 0 10 4 0 22
20-40% 8 20604 2543 53 2 0 09 2 011 7 0 35
40-60% 8 22969 2719 83 2 0 10 3 013 11 0 53
60%-80% 7 24174 3166 133 2 0 10 3 015 17 0 77
80%-100% 4 20242 4064 265 2 0 09 3 017 23 1 26
Mixed 0%-20% 27 8203 311 42 1 0 07 1 010 3 0 33
20-40% 47 8382 172 80 1 012 1 0 16 5 0 58
40-60% 79 8348 106 11 8 1 0 14 2 0 20 6 0 75
60%-80% 106 8835 83 163 2 017 2 0 24 9 0 98
80%-100% 134 8013 54 24 2 2 0 18 2 0 26 12 1 43
Rural 0-20% 36 1698 54 69 0 0 00 0 0 00 1 0 53
20-40% 51 1252 26 11 1 0 0 00 0 0 00 1 0 86
40-60% 68 1075 18 14 6 0 0 00 0 0 00 1 1 00
60%-80% 75 1082 17 190 0 0 00 0 0 00 2 1 36
80%-100% 68 1123 19 27 5 0 0 00 0 0 00 *» 2 02
* Density is the number of stores per 5 000 persons
N=29 073
Source: Food Retailers m me Food Stamp Program Characteristics & Service to Participants February 1997
S3
Food Retailers in the Food Stamp Program: Characteristics & Service to Participants
Table A.2
Regression Coefficients Demonstrating
the Effects of Selected Demographic and Housing Measures on Store Location
Measure Large Stores Supermarkets
Coefficient t-Value Coefficient t-Value
Conslanl -0 6831 -5 29 -0 8971 -5 83
Area 0 0001 10 34 0 0001 8 02
Population 0 0831 63 85 0 0916 73 85
Population Squared -0 0006 -44 19 -0 0008 -41 21
Popuia'ion <20 Years ■ 1012 1 92 0 0049 -1 51
Population >65 Years 28 2020 6 89 0 0058 3 18
Household size -1 8624 -7 90 -0 4596 -1025
Female Headed Households 13 8450 5 79 0 0076 3 49
Male Headed Households 30 4617 -4 39 -0 0540 •8 63
Households with Children 0 0075 4 08 0 0095 4 23
Single Persons 0 0001 5 30 0 0000 1 79
Hispanics 0 0000 •0 67 0 0000 4 51
Blacks -0 0018 •3 63 -0 0014 2 24
Asian 0 0033 2 39 0 0012 0 68
Foreign 0 0057 5 37 0 0012 0 86
High School Graduate -0 0081 -8 64 -0 0061 -5 13
Average No ol Rooms 00219 1 81 0 0507 3 48
Median Rent 0 0001 1 22 0OOO3 3 40
Absence ol Plumbing -0 0257 -9 00 ■0 01O0 -4 66
Employed in Services -0 0008 -0 73 -0 0006 0 43
Employed as an Operator 0 0104 8 95 00122 8 18
Average T,me o! Commute -0 C042 -3 21 -0 0042 2 56
Use of Public Transportaton -0 0017 -1 42 -0 0008 -0 55
Access to Vehicle 0 0040 3 08 0 0019 1 15
Urban High-poverty 0 2374 7 10 0 6757 15 i,:
Mned High-poverty 0 7404 28 20 1 3052 38 62
Rural High-poverty -0 050b -1 86 0 0055 0 13
Urban Othei 0 3405 1 3 85 0 9099 28 73
Mixed Other 0 6186 30 46 1 1484 42 18
Source Food Retailers in the Food Stamp Program Characteristics & Service to Participants. February 1997
34