Nutrition and food security in the food stamp program

              Oo7V-^ -e ( /lit D-'. /V 7 5 —>
UnitedMM
MMMMMM o(
Agrtculunt
Food and
Conwmor
Nutrition and Food Security
in the Food Stamp Program
Offleoof
AMRfW ond
Evaluation
January 1996
^
c^
United Stares
Creparfrrrentoi
Agriculture
Food and
Consumer
Sorvtc*
Offlcoof
Analysis and
Evaluation
Nutrition and Food Security
in the Food Stamp Program
Edited by Daryl Hall and Mike Stavrianos
Mathematica Policy Research, Inc.
The research performed for the papers in this monograph was sponsored by the
Food and Consumer Service (FCS) Office of Analysis and Evaluation as part
of its ongoing research agenda. Opinions expressed in the papers are those of
the authors and do not necessarily reflect the positions of FCS.
MPR Project Number:
FCS Contract Number:
FCS Project Officer:
8156-020
53-3198-3-038
Alana Landey
z-
< HfflZ MI
Contents
Acknowledgments
vii The Authors
1 Introduction
9 Rates Up: Trends in FSP Participation Rates, 1985-1992
Carole Trippe
25 Evaluation of Expedited Service in the Food Stamp Program
Susan Bartlett
47 Access of FSP Participants to Food Retailers
Richard Mantovani. Lynn Daft, James Welsh, and Theodore
Macaluso
75 Measuring the Dietary Quality of Americans' Food
Consumption: The Healthy Eating Index
Eileen Kennedy, James Ohls, Steven Carlson, and Kathryn
Fleming
91 Measuring Food Security in the U.S.: A Supplement to the
CPS
Gary Bickel, Margaret Andrews, and Bruce Klein
in
V HfflKw
Acknowledgments
The editors wish to express their appreciation to the many persons who
contributed to this monograph. We thank the authors for their excellent
research and sustained efforts. We also thank Steven Carlson. Alana
Landey, Gary Bickel. Pat McKinney, Barbara Murphy. Ken Offerman.
Ted Macaluso. Harold Beebout. and Thomas Fraker. who provided
technical consultation and review, and Sharon Clark, who prepared the
manuscripts.
v\ mmwn
The Authors
Margaret Andrews
Dr. Andrews is an analyst in the Office of Analysis and Evaluation (OAE)
at the Food and Consumer Service (FCS). In addition to her involvement
with the FCS food security measurement effort, she is currently serving as
project officer for a study in which nationally representative data will be
collected and analyzed to assess customer service issues in the Food
Stamp Program (FSP). Dr. Andrews has also directed evaluations of
electronic benefit transfer (EBT) demonstrations in Maryland and Ohio,
and she has been involved in various research efforts to measure and
evaluate FSP participants' access to food retailers
Susan Bartlett
A senior policy analyst at Abt Associates, Dr. Bartlett has been involved
in evaluating a variety of FSP-related initiatives. She recently completed
the study of expedited service summarized in this compendium. Before
that, she directed a study of the food stamp application process, exploring
factors that encourage and discourage persons from completing the
process. She also analyzed state-to-state variation in food stamp
certification costs and evaluated the impact of the system on food stamp
recipients for the initial EBT demonstration.
Gary W. Bickel
Dr. Bickel, an economist at FCS, has studied the phenomenon of poverty
in the U.S. from several different settings. For the past three years, he has
played a major role in FCS' development of the first direct measures of
food insecurity and hunger in the U.S. At FCS, he was also responsible
for the initial feasibility study of the EBT system for food stamps and for
the development of a model to estimate the size of the WIC-eligible
population. Previously, Dr. Bickel served on a U.S. Senate subcommittee
concerned with poverty issues, held associate professorships at Cornell and
Colorado universities, and provided social science research support to the
original Legal Services Program.
Steven Carlson
Director of the Family Programs staff in OAE at FCS, Mr. Carlson has
devoted his career to policy research and to the analysis and evaluation of
vii
The Authors
domestic food assistance programs, primarily the FSP. He currently leads
a multidisciplinary staff with research interests in welfare reform and
coordination, EBT systems, program operations, and nutrition education
and monitoring.
Lynn Daft
A senior partner of Abel, Daft, Earley & Ward International, Dr. Daft has
conducted research for more than 20 years in farm policy, economic
development, dairy markets, and domestic food programs. As associate
director of the White House Domestic Policy staff during the Carter
Administration, he was the president's principal staff advisor and policy
coordinator for agricultural issues. He also supervised analysis of policy
and budget issues for the Congressional Budget Office, the Office of
Management and Budget, and the office of the Secretary of Agriculture.
Kathryn H. Fleming
Dr. Fleming is a senior nutritionist at Technical Assessment Systems
(TAS). She is an expert in food consumption research, survey design,
dietary data collection methodology, and data analysis. At TAS, Dr.
Fleming was the principal investigator for the Food Guide Pyramid
component of the Healthy Eating Index. In this project for USDA, the
dietary intake of foods as reported in national food consumption surveys
was translated into servings of foods based on Food Pyramid groups. Dr.
Fleming was also principal investigator for the extant data analysis
component of the FCS-funded Barriers to Good Nutrition project—an
examination of the relationship between household and individual
characteristics, food expenditure patterns, diet quality, and nutrient intake.
Eileen Kennedy
Dr. Kennedy is executive director of the Center for Nutrition Policy and
Promotion, USDA. Before joining USDA, Dr. Kennedy was involved in
research on the food security, nutrition, and health effects of a range of
government food and agriculture policies in developing countries; much of
Dr. Kennedy's international work focused on Sub-Saharan Africa. Dr.
Kennedy worked with colleagues at Mathematica Policy Research in
developing the Healthy Eating Index.
Bruce Klein
An economist at the Center for Nutrition Policy and Promotion, Dr. Klein
is a member of an interdisciplinary team that disseminates new scientific
information and evaluates nutrition promotion. In his previous position at
FCS, he analyzed food assistance policy and programs, and he led the
design team for the new USDA food security and hunger questionnaire.
viu
The Authors
Dr. Klein is co-author, with William Spriggs, of Raising the Floor: The
Effect of the Minimum Wage on Low-Wage Workers. His other
publications deal mainly with issues of food security and the working
poor.
Theodore F. Macaluso
Dr. Macaluso is chief of the Evaluation Research Branch in the OAE at
FCS. He supervises the staff who design, award, and manage from 10 to
15 survey research, econometric, and experimental-design research studies
a year. His recent work concentrates on the agency's relationship with the
food retailing industry (including efforts to improve participants' access to
retailers that stock quality food at affordable prices and to reduce taxpayer
exposure to retailers who engage in food stamp trafficking). Dr. Macaluso
is also an expert in technology (EBT, automated case management, and
statistical risk-targeting systems) and in the characteristics and needs of
the low-income elderly.
Richard Mantovani
Dr. Mantovani, a technical director at Macro International Inc., directed
the Retailer Characteristics Study, which addresses availability and access
for F.SP participants. He has also contributed to a variety of other human
needs-related studies, including the current evaluation of Automated
Finger Imaging in New York State and an income quality control study of
tenants in housing subsidized by the Department of Housing and Urban
Development.
James C. Ohls
Dr. Ohls is a senior fellow at Mathematica Policy Research, Inc. He has
directed several major studies of food and nutrition policy, including
evaluations of the SSI/E!derly Food Stamp Cashout Demonstration, the
Food Stamp Simplified Application/Standardized Benefits Demonstration,
and the San Diego Food Stamp Cashout Demonstration. Dr. Ohls is the
co-author, with Dr. Harold Bcebout, of The Food Stamp Program: Design,
Tradeoffs, Policy and Impacts, published in 1993.
Carole Trippe
A researcher at Mathematica Policy Research, Inc., Ms. Trippe specializes
in the use of microsimulation models to conduct research on FSP policy
issues. Ms. Trippe directs a project for FCS in which these models are
developed to estimate FSP participation rates, trends in these rates, and to
examine the impact of proposed changes to the FSP.
IX
The Authors
James B. Welsh
President of Geosocial Resources, Inc., Mr. Welsh pioneered the
adaptation of geographic information systems (GIS) to human services.
From 1989 to 1991, he demonstrated uses of GIS in the FSP under a
small grant from FCS. Since then, he has extended the uses of this
technology to public assistance, Medicaid, and other programs by creating
systems for administrators, analysts, and caseworkers.
Introduction
THE FOOD STAMP PROGRAM The Food Stamp Program (FSP). administered by the U.S. Department of
Agriculture's Food and Consumer Service (FCS), is a central component
of America's national nutrition safety net, providing monthly benefits to
about 27 million participants in nearly 11 million households nationwide.
Through the FSP, over $23 billion in annual benefits is distributed to
nearly all low-income households that meet income, asset, and
employment-related eligibility requirements. Unlike many other welfare
programs, the FSP has few categorical requirements for eligibility, such as
the presence of children, or elderly or disabled individuals in a household.
As a result, the program supports a wide range of needy persons, many of
whom are not eligible for other forms of assistance.
The goals of domestic food assistance policy have evolved substantially
since the first food stamp plan operated from 1939 until 1943, during the
Depression. While a goal of this plan was to alleviate hunger, its primary
purpose was to stabilize agricultural prices by stimulating consumption of
surplus farm commodities. In the 1960s, the program was revitalized—
first through a 1961 Executive Order by President Kennedy, which created
a number of pilot food stamp projects, and then with the passage of the
Food Stamp Act of 1964. Still, participation in the program remained
optional for states and localities; benefit levels viried widely across states;
and there was a purchase requirement: participants had to pay a portion of
their income in order to receive benefits.
"This volume of research papers offers
an overview of current research on .. .
improving access to and participation
in the FSP. and measuring nutrition in
the FSP-The
passage of the Food Stamp Act of 1977 marked the beginning of the
modern FSP. Through the act, the purchase requirement was eliminated,
making the program more accessible to low-income households because
they no longer had to contribute in advance to participate. This recent
reform diminished the FSP's role in supporting agricultural prices and
made the program's primary goal, as stated in the Food Stamp Act, to
permit low-income households to obtain a more nutritious diet by
increasing their food-purchasing power.
This volume of research papers offers an overview of current research on
two topics critical to the program's success in achieving its goals:
Nutrition and Food Security in the Food Stamp Program
improving access to and participation in the FSP, and measuring nutrition
in the FSP. Of the rive papers, three focus on access and participation.
The first in this set presents the latest trends in FSP participation rates
among those eligible for benefits; the second examines the effectiveness of
specific provisions to expedite service to those in greatest need; and the
third investigates access of FSP participants to food retailers. The two
remaining papers address the measurement of food security and nutrition.
This introduction provides the policy context for the questions and issues
raised in the research papers.
PARTICIPATION IN AND
ACCESS TO THE FSP
The FSP has few categorical eligibility requirements and hence offers
assistance to a large and diverse population of low-income households.
The size of the eligible population is influenced by many factors,
including changes in program rules, the economy, and demographics. At
any given time, some percentage of these eligible households will
participate in the FSP. The ratio of participants to eligibles, or
participation rate, provides a useful measure of the program's success at
reaching its target population.
Participation rates can reveal other useful information as well. For
example, not all subgroups of the eligible population participate at the
same rate—demographic and economic factors can influence a
household's decision as to whether or not to participate. Comparing
participation rates across subgroups can therefore help program
administrators identify unmet needs and focus program outreach efforts.
Participation rates vary not only by subgroup but also over time. These
trends offer insight into the impact of outreach efforts, and into changes in
program rules and the economy on FSP participation.
Measuring FSP Participation The paper in this volume written by Carole Tri- * presents the latest
trends in FSP participation rates among those eligible for benefits. This
study used data from the Survey of Income and Program Participation
(SIPP) on food stamp eligibles and FSP administrative data on food stamp
participants. The results show that overall participation rates have risen
substantially since 1989 because of a surge in participation among the
eligible population. Almost 6 million new participants joined the program
between January 1989 and January 1992, compared with an increase of
only 1.9 million eligibles during that time. As a result, the participation
rate rose from 59 to 74 percent.
Participation rates and trends in rates varied across demographic and
economic subgroups. Nearly all eligible children participated in the FSP in
1992, but only one-third of eligible elderly persons did so. Households
Introduction
headed by African Americans were more likely to participate, as were
households that received AFDC and households that were eligible for a
relatively large food stamp benefit. Trends in participation rates for
subgroups tended to follow trends for the entire FSP population,
increasing by about 15 percentage points between 1989 and 1992.
However, rate increases for some groups—such as young children, single
adults, and households with no income—outstripped the average.
Conversely, participation rates increased only modestly among elderly
persons, households with income above the poverty level, and households
eligible for only a small FSP benefit.
Identifying and Overcoming
Barriers to FSP Participation
Throughout the history of the FSP, policymakers have varied their
emphasis on identifying the reasons that certain groups participate at lower
rates and on developing outreach strategies to boost these rates. In 1971,
Congress passed legislation requiring state food stamp agencies to inform
low-income households of their potential eligibility for the program. But
Congress weakened this mandate in the late 1970s and. in 1981,
eliminated federal funding for outreach efforts.
Surveys conducted in the 1980s indicate that most eligible,
nonparticipating households did not participate for one of three reasons:
they did not need food stamps, they were unaware that they were eligible,
or the perceived costs associated with participation outweighed the
benefits. These findings, however, were too general to inform the
development of outreach efforts; they did not explain the specific
motivations and constraints that shape a household's decision to
participate. Still, it was clear that groups such as the elderly, disabled, and
homeless were underrepresented in the FSP. The Hunger Prevention Act
of 1988 renewed funding for outreach activities in the FSP, and outreach
efforts have increased since then.
Expedited Service Policymakers have also developed administrative mechanisms to ease the
application process for those with special needs. One such mechanism,
examined in the paper by Susan Bartlett. is expedited service, through
which the FSP provides immediate assistance to households that have
insufficient resources to purchase food in the month they apply for
benefits. First implemented in 1979, expedited service was expanded by
the Stewart B. McKinney Homeless Assistance Act of 1987 to cover
homeless households and households in danger of becoming homeless. Dr.
Bartlett addresses several issues surrounding expedited service, including
the size and composition of the expedited service caseload, the impact of
expedited service on FSP administration, and ways in which expedited
service operations may be improved.
Nutrition and Food Security in the Food Stamp Program
In fiscal year 1992, nearly 2.5 million households received expedited
service—35 percent of the 7.1 million households approved to receive
I'ood stamps during that period. Some 90 percent of expedited service
households were eligible for the service because their income and
resources were below the established limits. Compared with other food
stamp households, these expedited service households were less likely to
contain children, less likely to be employed, and tended to have lower
income and assets. The remaining 10 percent of expedited service
households qualified solely because of the provisions of the McKinney
Act.
More than three-quarters of these expedited service applicants in fiscal
year 1992 received benefits within five days (as mandated by federal law)
compared with only 60 percent in the early 1980s. Despite this progress,
there is still room for improvement in the timeliness of expedited service.
Some food stamp offices have addressed the issue by screening applicants
to determine their eligibility for expedited service before scheduling the
certification interview and by interviewing expedited service cases on the
same day the applicant first appears in the office. Contrary to the concerns
of many policymakers, overpayment errois resulting from expedited
service are modest.
Food Stamp Participants' Access to Researchers have identified several methods to boost participation in the
Food Retailers FSP, but even among participants, certain barriers can diminish the
program's impact on food-purchasing power. For example, participants
may not have access to food retailers that accept food stamps. The
importance of access is acknowledged in that the FSP authorizes over
200,000 stores nationwide to accept food stamps. The paper written by
Richard Mantovani and colleagues examines whether these authorized
retailers are actually accessible to FSP participants.
The study documented in the paper covers eight sites, three in highly
urbanized areas, three in smaller metropolitan areas with adjacent rural
areas, and two in rural areas with small central cities or towns. In the
highly urbanized areas, over 90 percent of recipients lived within one-half
mile of a supermarket or large grocery store at the time of the study, and
all participants lived within one mile. In two of the three small
metropolitan areas and in both rural areas, about 40 percent of participants
lived within one-half mile, and about three-quarters lived within one mile
of a supermarket or large grocery store. In the remaining small
metropolitan area, the distance between participants and retailers was
substantial; fewer than half of the participants were within one mile of a
retailer, and fewer than two-thirds were within five miles. The authors
Introduction
conclude that most food stamp recipients live close to an authorized food
retailer, but that an inadequate infrastructure can exacerbate the access
problem, particularly in geographically remote sections of rural areas.
MEASURING NUTRITION
IN THE FSP
Increasing access to and participation in the FSP is a crucial first step in
achieving the program goal established by the Food Stamp Act of 1977—
to permit low-income households to obtain a more nutritious diet by
increasing their food-purchasing power. As mentioned, however, the use
of food stamps may not improve a household's nutritional status. For
example, a household may reduce its cash food expenditure by the exact
amount of the food stamp benefit, yielding no net impact on food
purchases. Research has revealed that while food stamps prompt
households to reduce cash spending on food, the benefit increases food
expenditures on the margin. One study indicates that each additional dollar
of food stamps produces a 20- to 45-cent increase in food expenditures
(Fraker 1990).
Measuring Dietary Quality The link between the receipt of food stamps and improved nutrition could
also be severed if increased food expenditures do not translate into
improved nutritional status. This could occur if, for example, the
household uses the extra money to purchase more pre-prepared food or to
eat more meals away from home, as these foods tend to have lower
nutritional quality than foods prepared at home.
To empirically assess the impact of the program on nutrition, researchers
need a reliable measure of dietary quality. In their paper on the Healthy
Eating Index, Eileen Kennedy and her colleagues propose such a measure.
The Healthy Eating Index combines 10 quantifiable components of a
healthy diet into a unidimensional measure of nutritional quality. The
index's components include overall fat consumption, saturated fat
consumption, cholesterol intake, sodium intake, dietary variety, and the
degree to which a person's diet conforms to USDA Food Pyramid serving
recommendations for each of the five major food groups.
The authors used the Healthy Eating Index to assess dietary quality in
households that participated in the 1989 and 1990 Continuing Survey of
Food Intakes by Individuals, in which three days of detailed dietary intake
are reported. The results of the assessment indicate that dietary quality is
associated with various socioeconomic characteristics. Dietary quality was
above average among young children and the elderly, and women tended
to score higher than men on the index. In addition, persons with income
above 200 percent of poverty scored significantly higher than those with
lower incomes, and those with higher levels of education outscored
Nutrition and Food Security in the Food Stamp Program
persons without a high school diploma. The authors suggest that a
simplified version of the index, which would not require access to large
computer databases, could be used to measure the effects of policy
interventions such as nutrition education programs.
Measuring Food Security and
Hunger Through the Current
Population Survey
The paper in this volume written by Gary Bickel and colleagues presents a
method of empirically measuring the prevalence and severity of poverty-linked
food insecurity and hunger in the United States. Such a measure
could be used to identify the neediest segments of the population in order
to investigate the causes and consequences of hunger; to examine the
relationships between hunger, malnutrition, and other health problems; and
to monitor efforts to reduce poverty-linked hunger.
Building on existing research and advice from experts in the field, the
authors developed a survey instrument that collects the data needed to
mi asure food insecurity and hunger. The questionnaire elicits information
on food shopping patterns, food expenditure levels, and participation in
food assistance programs. It also allows analysts to define and measure
broader concepts such as food sufficiency and methods of coping with
food scarcity. This survey was administered as a supplement to the April
1995 Current Population Survey, so it yields a large, representative
national sample. Data from this sample are being used to construct scaled
measures of food insecurity and hunger, and to measure the prevalence
and severity of food insecurity and hunger in the U.S. population.
RESEARCH AND POLICY This monograph highlights two aspects of FCS' broad research agenda: to
increase access to the FSP, and to measure nutrition and food security.
The research on these topics offers lessons on how the FSP can come
closer to its goal of improving the nutritional status of all low-income
households. The papers that follow are intended to stimulate discussion
and elicit new ideas about alternative approaches to achieving this and
other objectives of the FSP.
BLANK
PAGE
7
BLANK
PAGE
y
Rates Up:
Trends in FSP Participation Rates, 1985-1992
Carole Trippe
INTRODUCTION
• 'Almost 6 million new participants
joined the [FSPj between January
1989 and January 1992, causing the
participation rate to increase from 59
to 74 percent.
The Food Stamp Program (FSP) is the largest food assistance program in
the country, serving over 27 million persons in a typical month and
distributing $23 billion in benefits in fiscal year 1994. No other public-assistance
program reaches more poor individuals over the course of a
year. Because the FSP does not limit eligibility to persons meeting certain
categorical restrictions, such as the disabled, eiderly. or families with
children, food stamp benefits reach a much wider universe of persons than
other programs, providing assistance to some who might otherwise "fall
through the cracks" in the social welfare system.
Since food stamp benefits are available to any low-income persons who
meet the eligibility criteria, policymakers want to know how well the
program is reaching its intended population. The participation rate, which
measures the proportion of those eligible for food stamps who actually
apply for and receive them, provides this information. It can also indicate
how well the program is reaching certain subgroups of the eligible
population, such as children, the elderly, or the working poor.
This paper presents the latest FSP participation rates and trends in rates
since 1985 using Survey of Income and Program Participation (SIPP) data
for eligibles and FSP administrative data for participants. The results show
that overall participation rates have risen substantially since 1989 due to a
surge in participants among the eligible population. Almost 6 million new
participants joined the program between January 1989 and January 1992,
compared with an increase of only 1.9 million eligibles. causing the
participation rate to rise from 59 to 74 percent.
We first review the data and methods used for estimating FSP
participation rates in 1985, 1988, 1989. and 1992.' We then present
participation rates for January 1992 and set these estimates in the context
of past participation rates, thus revealing trends in rates over time. We
also present findings on which subgroups participated at higher or lower
rates than others and look at their trends over time. Finally, we examine
the characteristics of eligible nonparticipants.
C. Trippe
DATA AND METHODS:
ESTIMATING PARTICIPATION
RATES
The participation rate is the number of participants divided by the number
of eligibles. The number of participants is based on FSP administrative
caseload data. The number of eligibles, however, is approximated using
data from SIPP and a simulation designed to replicate the eligibility
process.
Participants We know how many persons and households participated in the FSP at a
given time as well as what their benefits and characteristics were because
food stamp offices collect and track this information. The total number of
participants and total food stamp benefits is based on the FSP Statistical
Summary of Operations data, or Program Operations data. The Program
Operations database contains the total caseload and dollar value of benefits
issued each month, but not information on the characteristics of FSP
participants. The characteristics reported here are based on a sample of
food stamp case records for each month from the FSP Integrated Quality
Control System, or IOCS data.
Eligibles We do not know explicitly how many persons and households were
eligible for food stamps at a given time or what the potential benefits were
because there is no record of eligible persons unless they actually apply
for food stamps. Therefore, we use SIPP data and a microsimulation
model to simulate FSP eligibility and potential benefits. To determine
participation rates in January 1992, for example, we simulated eligibility
for all households in the SIPP universe in that month based on Wave 7 of
the 1990 Panel and Wave 4 of the 1991 Panel.
' 'The model applied FSP eligibility
criteria .. . to each household on the
[SIPP-based data file/ to determine
whether the household to OS eligible for
food stamps.
A large part of estimating the number of eligibles involves preparing a
SIPP file that contains all the information needed to closely replicate FSP
eligibility criteria. A series of 30 programs was used to gather information
from various SIPP data products. The core questionnaire of SIPP provides
most of the information needed to model FSP eligibility, and the topical
module questionnaire and the initial Wave 1 questionnaire provide the
rest. The file preparation process begins by selecting all households that
were present in January 1992 from Wave 7 of the 1990 Panel and Wave 4
of the 1992 Panel. From these waves, we extracted most of the data
necessary for our simulation: household composition, earned and unearned
income, asset income, and participation in the various government
assistance programs. We then compiled the following remaining
information from the topical module and initial Wave I questionnaire:
disability status, financial asset balances, medical expenses, shelter and
dependent care expenses, and nonfinancial assets (vehicle ownership).3
Once the data file was created, the model applied the FSP eligibility
criteria in effect in January 1992 to each household on the file to
10
FSP Participation Kales
RESEARCH FINDINGS:
JANUARY 1992 PARTICIPATION
RATES AND TRENDS OVER
TIME
Participation Rates Rose
Substantially Between January
1989 and January 1992
determine whether the household was eligible for food stamps. For
households that were eligible, .'he model determined the value of the food
stamp benefits for which they qualified.
In January 1992. food stamp benefits reached more individuals than ever
before. The FSP provided benefits to 74 percent, or 24 million, of the 33
million persons eligible for benefits (Table 1). FSP participants received
$1.6 billion, or 82 percent, of the total potential food stamp benefits, and
they occupied 9.6 million households, or 69 percent of the total eligible
households. In terms of the total U.S. population of 252 million. 13 of
every 100 persons were eligible for food stamps, and 10 of every 100
persons received food stamp benefits in January 1992.
There was a 15 percentage point increase in the FSP participation rate for
individuals between January 1989 and January 1992 (Figure 1).' This
substantial increase—from 59 to 74 percent—occurred because of a surge
in new participants (up 32 percent) along with only a modest increase in
new eligibles (up 6 percent) (Table 2).
Almost 6 million new participants joined the program between January
1989 and January 1992. About 77 percent, or 4.6 million of the increase
was due to a higher participation rate among eligibles (Figure 2).4 Only 19
percent, or 1.1 million, of the increase was due to an increase in the
number of eligibles. The remaining 4 percent increase was due to the
interaction between the two factors: additional eligibles participating at a
higher participation rate/
The surge in participants and thus participation rates after 1989 can be
attributed to a worsening economy and other factors such as expansions in
Medicaid, increased access to FSP offices, and outreach. Expansions in
Medicaid began as early as 1988, and the effects of the recession were felt
in many areas of the country before the recession was indicated by
national measures.*
In January 1992, FSP participation rates reached their highest point since
the beginning of the series in August 1985. Between August 1985 and
Table 1. January 1992 FSP Participation Rates
Participants
(thousands)
Persons
Households
Benefits
24.291
9.631
$1,615,320
Eligibles
(thousands)
32,n3l
13.483
$1,981,717
Participation
Rate (percent)
74
69
82
11
C. Trippe
Figure 1. Trends in FSP Participation Rates, 1985-1992
90% Participation Rate
40%
August
1985
January January
1988 1989
Month/Year
Source: Food Stamp Program Operations data, SIPP data for the years shown.
Figure 2. Increase in Participants Due to a Higher Participation Rate, 1989-1992
Thousands
35
30 -
25 -
20 -
15
August
1985
[£] Increase in participants due to a higher
participation rate
Increase in participants due to an increase in
the number pf eligibles
January January
1988 1989
Month/Year
Source: Food Stamp Program Operations data, SIPP data for the years shown.
Benefits
Persons
Households
January
1992
Eligibles
- Participants
January
1992
12
FSP Participation Rates
Trends Are Consistent Across
Different Data Sources
January 1988, the individual participation rate declined slightly, from 64
percent to 59 percent. It remained constant though 1989 before surging to
74 percent in January 1992. The decline in rates between 1985 and 1988
was largely a result of legislative changes authorized under the 1985 Food
Security Act (FSA). Although the act expanded the number of persons
eligible to receive food stamps, most of the newly eligible persons did not
participate in the FSP.
These trends, identified using SIPP data to estimate eligibles. are
consistent with those identified in a study in which Current Population
Survey (CPS) data were used to estimate eligibles, as shown in Figure 3.
The SIPP-based estimates show a 5-point drop in the individual
participation rate from 1985 to 1988, no change from 1988 to 1989, and a
15-point increase from 1989 to 1992. The CPS-based estimates show a
similar drop (4 points) in the individual participation rate from 1984 to
1986, no change in the rate (less than 1 point) from 1986 to 1988, and a
12-point rise in the rate from 1988 to 1993.
Although these trends are based on different data sources, and each covers
a slightly different period, they are remarkably similar. The rates based on
SIPP data are more accurate than those based on CPS data because the
Figure 3. Comparison of SIPP-Based and CPS-Based Participation Rates, 1884-1993
Participation Rate
80%
70% -
60%
50%
40%
30%
SIPP-Based
Participation
Rates
CPS-Based
Participation
Rates*
T r 1 1 1 i i i i r
8/84 1/85 8/85 1/86 8/86 1/87 8/87 1/88 8/88 1/89 8/89 1/90 8/90 1/91 8/91 1/92 8/92 1/93
Month/Year
Source: Food Stamp Program Operations data, March CPS data, SIPP data for the years shown.
* There is a break in the CPS-based time series in 1992 due to revisions
to the weighting process introduce i in the March 1993 CPS.
13
C. Trippe
SIPP database contains more of the information needed to estimate food
stamp eligibility, thus allowing us to more closely replicate the actual
eligibility determination process. However, the SIPP data do not cover as
long a period, and certain types of SIPP data needed to estimate eligibles
are available for only a limited number of years. Nonetheless, the CPS-based
study supports the slight drop in rates during the mid-1980s and the
substantial increase in rates since 1989.
Rates for Subgroups As summarized below, and as shown in Tables 3 and 4, some groups of
eligibles tended to participate in January 1992 at a higher or lower rate
than others:
Most Eligible Children Participated. The FSP served almost every
eligible child younger than 5 (95 percent) and most eligible children
younger than 18 (86 percent).
One in Three Eligible Elderly Persons Participated. One-third (33
percent) of eligible elderly persons participated in the FSP. Eligible
elderly persons living alone were more likely to participate than those
living with others.
Table 2. Comparison of FSP Participation Rates Over Time, 1985-1992
August 1985 August 1988 January 1989 January 1992
Percent Change
(1989 to 1992)
THOUSANDS
Eligibles
Persons
Households
Benefits
28,884
11,604
$1,072,262
30,973
12.292
1,334,779
31,041
12.689
1.405.636
32.931
13.983
1,981,717
6.1 %
10.2%
41.0%
Participants
Persons
Households
Benefits
18,560
6,894
$807,265
18,286
6,882
890,158
18.344
7.037
927,391
24,291
9.631
1,615,320
32.4%
36.9%
74.2 %
PERCENT
DIFFERENCE
(l989to 1992)
Participation Rates
Persons
Households
Benefits
64.3
59.4
75.3
59.0
56.0
66.7
59.1
55.5
66.0
73.8
68.9
81.5
14.7 points
13.4 points
15.5 points
Source: Estimates for eligibles are from the FOSTERS model, using data from SIPP.
Food Stamp Program Operations data, adjusted for issuance errors.
Numbers of participants a. from the
14
FSP Participation Rates
African Americans Participated at Higher Rates Than Other
Racial/Ethnic Groups. Eligible households headed by African
Americans were more likely to participate (92 percent) than
households headed by Hispanics (61 percent) or white non-Hispanics
(59 percent).7
The Lower the Income, the Higher the Participation Rate. The
FSP participation rate for households with a monthly income below
the poverty line was 86 percent, compared with 21 percent for
households with an income above the poverty line." As income
increased, households were less liftjly to participate.
Table 3. Participation Rates By Selected Demographic Characteristics, January 1992
Characteristics
Individual
Elderly age 60 or older
Living alone
Living with others
Disabled under age 60
Living alone
Living with others
Children under age 18
Preschool (under Age 5)
School-age (Age 5 to 17)
Adults ages 18 to 59
Living alone (not disabled)
Gender
Male
Female
Total
Household
White non-Hispanic head
African American head
Hispanic head
Total
Number of Number of Participation
Participants Eligibles Rate
(thousands) (thousands) (percent)
1,707 5.137 33.2
1,129 3.113 36.3
578 2.023 28.6
951 1.419 67.0
446 380 117.5
504 1,039 48.5
12,357 14,455 85.5
4.695 4,954 94.8
7.662 9,500 80.6
10,214 13,340 76.6
1,527 1.358 112.4
10.014 13,475 74.3
14,276 19,456 73.4
24,291 32,931 73.8
4,570 7,803 58.6
3,334 3,612 92.3
1,300 2,117 61.4
9,631 13,983 68.9
Sources: January 1992 Food Stamp Program Operations data adjusted for issuance error. Special tabulations from IOCS data
for January and February 1992.
January 1992 FOSTERS model. Wave 7 of the 1990 Panel and Wave 4 of the 1991 Panel of SIPP.
Note: Participation rates exceeding 100 percent are due to reporting and measurement errors in SIPP and in IQCS data.
Although IQCS data is a survey of FSP participants, it has some, although fewer, reporting and measurement errors,
particularly for characteristics not needed to determine eligibility.
15
C. Trippe
• The Higher the Benefit, the Higher the Participation Rate. The
participation rate was higher for households eligible for large benefits
than for households eligible for small benefits. For example, the
participation rate was 57 percent for those eligible for $150 in benefits
or less, and 89 percent for those eligible for benefits of more than
$150. The average benefit for eligible households was $142 in
January 1992.
• Households Receiving AFDC Benefits Were More Likely to
Participate Than Those with Earnings or Unemployment
Compensation. The participation rate for households with AFDC
exceeded 100 percent, the rate for households with earnings was 41
Table 4. FSP Participation Rates Tor Households by Income, Income Sources. . nd Benefit Amounts, January 1992
Number of Number of Household
Participating Eligible Participation
Households Households Rate
(thousands) (thousands) (percent)
8.870 10.288 86.2
761 3,695 20.6
1,910 3,959 48.2
1,755 2,393 73.4
876 1,372 63.8
879 1,020 86.2
4,574 3,783 120.9
3,754 3,129 120.0
885 744 118.8
267 648 41.2
353 1.828 19.3
1,606 2,973 54.0
2.942 3.856 76.3
4,729 5.326 88.8
Income as a Percentage of Poverty
Total £ 100%
Total > 100%
Source of Income
Earned income
SSI
Elderly in the unit
No elderly in the unit
Public assistance
AFDC
Other welfare
Unemployment compensation
Benefit Amount
$10 or less
$11 - 75
$76 - 150
$151 or more
Total 9,631 13,983 68.9
Sources: January 1992 Food Stamp Program Operations data adjusted for issuance error. Special tabulations from IQCS data
for January and February 1992.
January 1992 FOSTERS model, Wave 7 of the 1990 Panel and Wave 4 of the 1991 Panel of SIPP.
Note: Participation rates exceeding 100 percent are due to reporting and measurement errors in SIPP and in IQCS data.
Although IQCS data is a survey of FSP participants, it has some reporting and measurement errors, particularly for
characteristics not needed to determine eligibility.
16
FSP Participation Rates
percent, and the rate for households with unemployment compensation
was 48 percent.''
Trends in Rates for Subgroups
Tend to Follow Overall Trends
Trends in participation rates for subgroups tend to follow trvnds for the
general FSP population. Between 1989 and 1992. rates for most subgroups
increased substantially—by about 15 points. However, rates for some
groups increased more or less relative to other groups, as existing trends
continued in most cases. Figures 4 through 7 illustrate the trends in rates
over time and the patterns in rates by different characteristics of eligibles.
Rates for Children, Especially Preschoolers, Surged. Although
participation rates for preschool-age children fell slightly from 1985 to
1989, they surged between 1989 and 1992 by 21 points. By comparison,
participation rates for school-age children also fell slightly from 1985 to
1989 and rose between 1989 and 1982 by about the same as overal!
rates—15 points (Figure 4).
Rates for Elderly Persons Showed Little Change. Rates for elderly
persons fell slightly between 1985 and 1989. This drop in rates is largely
Figure 4. Trends In Participation Rates By Age Groups, 1985-1992
Participation Rate
August January January
1985 1988 1989
Month/Year
Source: Food Stamp Program Operations data, SIPP data for the years shown.
Preschool
Children
School-Age
Children
Adults
Total
Elderly
January
1992
17
C. Trinpe
due to greater numbers of elderly made eligible under the 1985 FSA.
These rates rose by only 5 points between 1989 and 1992 (Figure 4).
During this period, rates for eldeily per n rose less than rates for
children, which may reflect the fact that the income of elderly persons
tends to be relatively constant in real terms Figure 4 also shows that
participation rates for elderly persons EU ently lower than rates for
children.
Rates for Adults Increased, and Rates for Single Adults Surged.
Participation rates for adults (ages 18 to 59) increased substantially—by
17 points—and rates for single (nondisabled) adults soared—by 28
points—between 1989 and 1992 (Figure 4).10 The latter increase may be
partly a result of the large number of single persons applying for food
stamps after other forms of assistance, such as General Assistance, were
terminated or cut back."
Rates for African Americans Continued to Outpace Rates for Other
Groups. Participation rates for households headed by African Americans
were consistently higher than rates for other racial/ethnic groups over the
1985 to 1992 period (Figure 5). Rates increased slightly more for African
Americans than for other groups between 1989 and 1992—by 15 points
Figure 5. Participation Rates by Race/Ethnicity of Household Head, 1985-1992
Participation Rate
100%
20%
August
1985
1 1
January January
1988 1989
Month/Year
African
American
Total
Hispanic
White
January
1992
Source: Food Stamp Program Operations data, SIPP data for the years shown.
18
FSP Participation Rates
compared with 13 points for whites and 11 points for Hispanics. For
Hispanic households, there was a larger-than-average increase in the
number of participants and eligibles (46 and 20 percent, respectively)
between 1989 and 1992, reflecting increases in immigration.
Rates Increased More for Those with Lower Incomes Than for Those
with Higher Incomes. Participation rates increased much more for
households with gross income below the poverty level (14 points) than for
households with gross income above the poverty level (6 points) between
1989 and 1992 (Figure 6). As gross income rose, the rate grew less
dramatically. Participation rates for households with zero gross income
grew the most, by 23 points. Rates for households with gross income less
than 50 percent of poverty grew by 16 points, and rates for households
with income between 50 and 100 percent of poverty grew by 11 points.
Rates for the Working Poor Climbed. After dropping slightly between
1985 and 1989, the participation rate for households with earnings
increased by 16 points between 1989 and 1992 (Figure 6). This rise may
reflect the growing need for food assistance by the working poor as their
real income falls.
Figure 6. Participation Rates by Selected Economic Characteristics, 1985-1992
<nnM Participation Rate
UU 70
90%
Households
80% -
70%
in Poverty
Total
60% - ,
50% - Households with
Earned Income
40% - ..
30% -
Households
20% - Not in Poverty
10% -
fU)7%0
Aufi
19
ust
85
i i
January January
1988 1989
Jam
191
jary
)2
Month/Year
Source: Food Stamp Program Operations data, SIPP data for the years shown.
19
C. Trippe
Rates Increased More for Those Eligible for High Benefits Than for
Low Benefits. Since those with the lowest income are eligible for the
highest benefits, it is not surprising that the increase in rates for those with
the lowest income is associated with an increase in rates for those eligible
for the highest benefits. The participation rate for households eligible for
the highest benefits (between 76 percent and 99 percent of the maximum
benefit) increased by 25 points, while the rate for households eligible for
the lowest benefits (between I percent and 25 percent of the maximum)
increased by only 2 points between 1989 and 1992 (Figure 7).'2
Characteristics of Eligible
Nonparticipating Households in
January 1992
Although 24 million persons participated in the FSP in January 1992, 8.6
million (26 percent of all eligible persons) did not participate. These
eligible nonparticipants occupied 4.3 million households and were eligible
for $36 million in benefits. The groups of eligibles with the largest
proportion of nonparticipants included:
Households with elderly persons (70 percent) and households headed
by a white non-Hispanic (74 percent)
Households with the highest gross income (67 percent for households
with a gross income above poverty) and eligible for the lowest food
Figure 7. Participation Rates by FSP Benefit Levels as a Percentage of Maximum Benefit, 1985-1992
Participation Rate
110%
76-99%
51-75%
Total
28-50%
1-25%
August
1985
January
1988
January
1989
January
1992
Month/Year
Source: Food Stamp Program Operations data, SD?P data for the years shown.
20
FSP Participation Rales
stamp benefits (50 percent for those eligible for less than 25 percent
of the maximum benefit)
Households with earnings (47 percent)
SUMMARY AND CONCLUSIONS The FSP participation rate—the ratio of the number of participants to
eligibles—provides a good measure of the program's success in reaching
its intended population. It can also indicate how well the program is
reaching certain subgroups of the eligible population. Estimates based on
recent SIPP data for eligibles and food stamp administrative data for
participants indicate that the FSP is reaching more eligible persons than
ever before. In January 1992, the program reached 74 percent of eligible
individuals—up 15 points from January 1989.
The substantial rise in the participation rate since 1989 is due to a surge in
new participants (up 32 percent) along with only a modest increase in new
eligibles (up 6 percent). Almost 6 million new participants joined the
program between January 1989 and January 1992. About 77 percent of
this increase was due to a higher participation rate among eligibles, 19
percent was due to an increase in eligibles, and the remaining 4 percent
was due to additional eligibles participating at a higher participation rate.
■Results based on 1993 CPS data
indicated that participation rates are
likely to continue to grow after 1992,
but at a slower pace.''
Participation rates for most subgroups increased by the same amount—15
points—as for the general FSP population between 1989 and 1992.
However, rates for some subgroups increased more or less relative to other
groups. For example, rates surged for preschool-age children—by 21
points—but showed little change for elderly persons—rose by 5 points.
Rates for total adults increased by about the average— 17 points—but
rates for single adults surged—by 28 points. Rates continued to increase
more for those with incomes below the poverty level—14 points—than
for those with incomes above the poverty level—6 points. Since the
dramatic jump in participation rates from 1989 to 1992, the number of
participants has continued to rise, climbing from 24.3 million persons in
January 1992 to 26.8 million persons in January 1993, and to 27.7 million
in January 1994. The number of participants has fallen slightly since 1994,
with 27.0 million participants in January 1995. As new SIPP data are
released, allowing us to estimate the change in eligibles since 1992, we
will be able to measure whether the participation rate has continued to
rise, or if it leveled off after 1992. Results based on 1993 CPS data
indicated that participation rates are likely to continue to grow after 1992,
but at a slower pace.
21
C. Trippe
REFERENCES Dan/.iger, Sandra K . and Sherrie A. Kossoudji. "What Happened lo Former GA
Recipients? The Second Interim Report of the General Assistance Termination Project."
University of Michigan School of Social Work, April 1994.
Martini, Alberto. "Participation in the Food Stamp Program: A Multivariate Analysis." In
Current Perspectives on Food Stamp Program Participation. Alexandria. VA: Food and
Consumer Service, U.S. Department of Agriculture, 1992.
McConnell. Sheena. "The Increase in Food Stamp Participation Between 1989 and 1990: A
Report to Congress." Alexandria, VA: Food and Consumer Service. U.S. Department of
Agriculture, 1991.
Sykes, Julie. "Creation of the January 1992 FOSTERS Microsimulation Model and
Database: Technical Working Paper." Washington. DC: Mathematica Policy Research.
1994.
Trippe. Carole. "Trends in FSP Participation Rates: Focus on August 1993." Washington.
DC: Mathematica Policy Research, Inc., 1995.
Trippe, Carole, and Julie Sykes. "Food Stamp Program Participation Rates: January 1992."
In Current Perspectives on Food Stamp Program Participation. Alexandria. VA: Food
and Consumer Service. U.S. Department of Agriculture. 1994.
Trippe, Carole, and Pal Doyle. "Food Stamp Program Participation Rates: January 1989."
In Current Perspectives on Food Stamp Program Participation. Alexandria. VA: Food
and Consumer Service. U.S. Department of Agriculture. 1992.
NOTES I. The paper is based on a report on January 1992 participation rates (Trippe and Sykes
1994). which is the latest in a series that provides participation rates for 1985. 1988. and
1989. The years are selected on the basis of availability of SIPP Panel data needed to
estimate eligibility, since necessary waves of SIPP data are only available in certain years.
2. More detailed information on (he creation of the model database is in Sykes (1994).
3. The rate also rose for eligible households (up 13 poinls) and for potential benefits (up 16
points)
4. If there had been no increase in the size of the eligible population between 1989 and
1992. but participation rales had been as observed in 1992. the increase in the number of
participants would have been 4.6 million persons, or 77 percent of the observed 6 million
increase—just from an increase in the participation rale. However, if the participation rate
had remained constant between 1989 and 1992. (he increase in participants would have been
only I.I million, or 19 percent of the observed 6 million increase—just from ;he increase in
the number of eligibles. The remaining 4 percent increase is likely due to interactive effects
of additional eligibles participating at a higher rale.
5. Figure 2 sums the 19 and 4 percent increases into a 23 percent increase due to additional
eligibles (participating at the 1989 rate and higher rates).
6. See also McConnell (1991). The recession officially began in July 1990 and ended in
March 1991. according to the National Bureau of Economic Research.
7. Multivariate analyses of the relationship between race/ethnicity and participation rates
revealed that when other household characteristics are held constant, there is a much smaller
gap between participation rates of households headed by African Americans and whites
22
FSP Participation Rules
(Martini 1992). This suggests lhat most of the difference between participation rates of
African Americans and whites is not a result of race per se, but of factors correlated with
race.
8. Households in poverty are defined as households with gross income less than the federal
poverty guideline by household size.
9. When adjusted for known levels of underreporting of AFDC program participation in
SIPP, the participation rate for households with AFDC was 88 percent. This more realistic
rate for households with AFDC is still much higher than for other groups of eligibles.
Multivariate analyses of program participation have shown that FSP eligible households
with public assistance are much more likely to apply for food stamps than other households,
regardless of their income, household size, or other characteristics.
10. Participation rates for single nondisabled adults are not shown in the figure.
11. See Danziger et al. (1994) for a discussion of the extent to which former General
Assistance participants in Michigan have used other public assistance since General
Assistance was terminated.
12. Participation rates for households eligible for the maximum benefit (households with
zero net income) increased by 14 points, but these households seem to behave differently
from other households. A study of zero-income households is currently being conducted by
Mathematica Policy Research.
23
M u
Evaluation of Expedited Service
in the Food Stamp Program1
Susan Bartlett
INTRODUCTION Expedited service is the administrative mechanism through which the
Food Stamp Program (FSP) provides immediate assistance to households
that do not have enough resources to purchase food in the month they
apply for benefits. Applicants processed under expedited service
procedures are entitled to receive food stamps within five calendar days of
filing their application, instead of within the standard processing time of
30 days. Under current law, four categories of households qualify to
receive expedited service:
Households that have less than $150 in gross monthly income and
$100 or less in liquid resources
• Households composed of destitute migrant and seasonal farm workers
with liquid resources of $100 or less
• Households in which all members are homeless
• Households judged at risk of becoming homeless because their
combined gross monthly income and liquid resources are less than
their monthly housing and utility costs.
Expedited service procedures were first implemented in 1979, and in 1987,
the Food and Consumer Service (FCS) published a comprehensive study
of expedited service that used data from 1981 through 1984.2 Washington,
DC: U.S. Department of Agriculture, Food and Nutrition Service; and
SRA Technologies, Inc., April 1987. We refer to this as the "1987
Study" throughout this paper. Since that time, important legislative,
economic, and legal developments have prompted FCS to sponsor the
evaluation of expedited service provisions, which is the subject of this
paper. First, the Stewart B. McKinney Homeless Assistance Act of 1987
(McKinney Act) expanded expedited service entitlement to homeless
households and to households judged in danger of becoming homeless
because they lacked the resources to cover their shelter expenses. States
have raised concerns that the proportion of approved applicants entitled to
expedited service increased markedly because of the McKinney Act.
creating a substantial burden on local offices. In addition, some have
25
S. Bartlell
expressed concern that expedited service is no longer targeted to
households with the most urgent need for emergency assistance. Many
also believe that it is more difficult to process the groups added by the
McKinney Act than others who qualify for expedited service.
Second, the economic downturn of the late 1980s led to a 34 percent
increase in the food stamp caseload between 1988 and 1992, as well as
state budgetary cutbacks, both of which affected the local administration
of the FSP. Many states have argued that expedited service policy adds a
difficult administrative burden to an already over-stretched system.
Finally, because states have been faced with legal challenges for failing to
meet expedited and general food stamp delivery standards, FCS is
interested in examining the factors that affect the timeliness of benefit
delivery.
"The overall objective of this study is
to provide information that FCS can
use to assess current expedited service
operations.''
In light of these events and issues, the overall objective of this study is to
provide information that FCS can use to assess current expedited service
operations at the national level. This project updates previous findings and
examines the impact of recent legislative and regulatory changes designed
to improve expedited service. The study also responds to states' concerns
with expedited service and presents the perspectives of state officials, local
food stamp officials, and the advocacy community on the problems they
are having with the implementation of expedited service and suggestions
for ways to improve operations. Specifically, the study:
• Provides information on the number and characteristics of expedited
service households and the nature of expedited service operations
• Examines the impact of the McKinney Act on expedited service
households and operations
• Evaluates the extent to which expedited service operations achieve the
intent of federal laws and regulations
• Examines the impact of expedited service on overall FSP
administration
• Identifies ways in which expedited service policy operations may be
improved
The approach to these tasks, including study design, sampling, and data
collection activities, is explained in the next section. This is followed by a
26
Expedited Service in the FSP
discussion of the research findings, which presents the number and
proportion of food stamp applicants that receive expedited processing and
examines the impact of the McKinney Act on the expedited service
caseload. We also analyze the demographic and economic characteristics
of regularly processed and expedited service cases, and we explore two
key aspects of the implementation of expedited service policy—timeliness
of benefit delivery and accuracy of screening—to assess how well current
operations achieve the intent of the provisions. Also examined is the
extent of payment error associated with expedited service. The paper
concludes with a summary of findings and attempts to draw implications
for further public policy discussions.
RESEARCH APPROACH: The study is designed to provide precise and valid national estimate^ of
DATA AND METHODS the size and characteristics of the expedited service population and
selected subgroups, and to compare key characteristics and processing
outcomes for approved expedited service and regularly processed
applicants.' In order to meet these goals we developed a two-stage
national probability sample of expedited and regularly processed
households.
In the first sampling stage, we selected 59 local food stamp offices to
participate in the study by stratifying all local offices in the 48 contiguous
states and the District of Columbia along two dimensions.4 First, offices
were grouped into three categories of caseload size—small, medium, and
large- defined so that one-third of all local offices fell within each
category.' The second stratifier also divided offices into three groups:
metropolitan offices with a high concentration of homeless applicants,
other metropolitan offices, and nonmetropolitan offices.6 Within each
stratum, offices were selected using systematic probability proportional to
size (PPS) sampling, with monthly caseload as the measure of size.
Because of this methodology, larger offices within a stratum had a greater
probability than smaller offices of being included in the sample.
In the second sampling stage, two nationally representative samples of
approved food stamp applicants were drawn from the 59 local offices
selected for the study. The first sample included households that applied
for food stamps between October 1, 1991 and September 30, 1992 and
were approved to receive benefits. The states (and in some cases, the
counties) participating in the study supplied these data. The second sample
included those that applied during August and September 1993. The time
frame for this sample was restricted because we used a self-administered
survey of food stamp applicants (described in more detail below) to
provide the sampling frame.
27
5. Barllett
Approved applicants were stratified into three categories—homeless,
expedited but not homeless, and regularly processed. Within each local
office, we then drew a systematic random sample from each category. The
final analysis sample included 4,497 approved households that applied
during fiscal year (FY) 1992, and 3,695 approved households that applied
during the 1993 period. Although the applicant samples were designed to
reduce variation in the probability of selection, the samples were not fully
self-weighting. All applicant data were therefore weighted to account for
their selection probabilities.
Data for those in the final analysis sample were abstracted from case file
records of food stamp applicants. These data provided information on
whether applicants received expedited service, the characteristics of
expedited and regularly processed food stamp applicants, and details of the
application process necessary to assess expedited service operations.
All persons applying for food stamps in the small and medium-sized
offices during August and September 1993, and a sample of persons in the
large offices were asked to complete a self-administered applicant survey,
which was attached to the application form. The eligibility workers
reviewed the survey during the certification interview and entered
information on whether the household was approved for food stamps and
whether it received expedited service. Using this information, we selected
the 1993 sample of approved applicants for the case file record
abstractions. The survey also provided information about applicants'
circumstances immediately prior to applying for food stamp benefits,
which helped us to better measure households' need for emergency food
assistance. Finally, we surveyed state and local food stamp directors,
workers involved in the application process, and food stamp advocacy
groups in the selected local officers to obtain information on state and
local policies and procedures concerning expedited service. We were also
interested in'their perspective on the issues and problems associated with
current expedited service policy and changes that would improve the
policy.
RESEARCH FINDINGS
Prevalence of Expedited Service
Among Approved Applications
In the 12 months from October 1991 to September 1992, 7.1 million
households were approved to receive food stamp benefits nationwide
(Table I). Thirty-five percent of them received expedited service. In
August-September 1993, the proportion receiving expedited service was
43 percent. Although the difference between these two estimates appear*
to suggest that the proportion of expedited service cases rose between
1992 and 1993, further analysis suggests that no important increase
occurred. The expedited service rate in August-September 1992 was also
28
Expedited Service in the FSP
relatively high at 38 percent. The difference of five percentage points in
the two August-September periods is not statistically significant.
Therefore, expedited service rates do not appear to have changed greatly
between 1992 and 1993.
The proportion of applicant households that receives expedited services in
larger offices is somewhat higher than in smaller offices. Similarly,
metropolitan offices have somewhat higher expedited service rates than
non-metropolitan offices. Most of the observed differences, however, are
not statistically significant.
The previous national study of expedited service in the FSP found that the
proportion of applicants receiving expedited service during the February
1983-May 1984 period was 34 percent. This does not differ significantly
from the current study's estimated expedited service rate for FY 1992.
Although the data suggest that the percentage of food stamp applicants
receiving expedited service has not increased in the past decade, the actual
number of applicants processed under expedited procedures has increased
quite substantially. During FY 1992, approximately 2.5 million households
received expedited service. During the early 1980s, somewhat over 2
million households received expedited processing annually. Thus, the
number of actual applications that were processed under expedited service
increased by 22 percent during the past decade. This observed increase is
mainly a result of the dramatic growth in the overall food stamp caseload
Table 1. Expedited Service Status of Approved Food Stamp Applicant Households
2/83-5/84 10/91-9/92 8/93-9/93
Total number of approved 7,960,000 7.132,380 794.904
food stamp households
Number receiving expedited service
(standard error)
2.710,000
(349,000)
Percent receiving expedited service
(standard error)
34.0
(2.5)
Number regularly processed
(standard error)
5,250,000
(501.000)
Percent regularly processed
(standard error)
66.0
(2.5)
2.485.603 338,744
(289.184) (30,780)
34.9% 42.6%
(2.9) (3.0)
4,646,777 456,159
(658.916) (47.825)
65.2% 57.4%
(2.9) (3.0)
Unweighted N 2434 4497 3695
Source: 1987 study; unweighted N refers to sample size for detailed case tile abstraction.
29
S. Bartlett
and does not reflect significant increases in the proportion of applicants
receiving expedited service.
Expedited Service Entitlement
Criteria
Households may qualify for expedited service under any of four criteria.
Table 2 shows the distribution of expedited cases by entitlement criteria
for 1991-1993.7 The first column shows a duplicated distribution in which
some households appear in more than one category. The total thus sums to
more than 100 percent. The second column is unduplicated. Each case
appears in only one category—the first identified category in the order
listed. The total of this column sums to 100 percent.
Fully 90 percent of all expedited service cases qualify because their
income and resources are below the established limits. Destitute migrants
and seasonal farmworkers make up only 2 of the expedited service
caseload. However, most also qualify for expedited service on the basis of
income/ resources.
Of special interest are the households qualifying for expedited service
because of the McKinney Act. Overall, a substantial number of expedited
service cases are homeless (24 percent) or appear to be at risk of
becoming homeless (56 percent). Most of these households, however, also
have income and resources below the established guidelines and so would
have qualified for expedited service before the McKinney Act. Only 10
percent of households qualifying for expedited service can be regarded as
"McKinney" cases, meaning that they qualify for expedited service only
because of the McKinney Act provisions. Most of these cases are not
homeless (2.5 percent) but are considered to be in danger of becoming
homeless (7.1 percent) because their shelter costs exceed their income and
liquid resources. We conclude that the McKinney Act added a substantial,
though not overwhelming, number of cases to the expedited service
Table 2. Approved Applicants Receiving Expedited Service, by Entitlement Criteria
Percentage Distribution of Approved Applicants
Receiving Expedited Service
Entitlement Criteria Duplicated Unduplicated
Monthly income/resources below limits
Destitute migrant/seasonal worker
McKinney criteria
Homeless
Shelter expenses exceed income/resources
Total
Unweighted N
90.3%
1.7
24.3
55.8
> 100.0%
5234
90.3%
0.1
2.5
7.1
100.0%
5234
30
Expedited Service in the ESP
caseload. In fact, the McKinney Act appears to be responsible for the
stability in the expedited service rate observed over the past decade. In the
absence of the McKinney Act, the FY 1992 rate would have been lower
than observed in the early 1980s.
Characteristics of the Expedited
Service Caseload
Expedited service cases are expected to have characteristics that differ
from those of regularly processed cases, reflecting their differential needs
for program assistance. This section explores these differences. It also
examines how the characteristics of expedited households differ from one
another depending on the criteria under which they qualify for expedited
service. In particular, we are interested in whether McKinney cases differ
from other expedited cases in ways that suggest they are more or less
needy than other cases.
Demographic Characteristics. The typical expedited service household is
strikingly different from the typical regularly processed case, as shown in
Table 3. A majority of expedited service applicants (56 percent) are one-person
households, whereas most regularly processed households include
two or more people. Expedited service households do not typically include
children. Only 38 percent of expedited service households include
children, compared with 61 percent of regularly processed cases.
Given this basic difference in household types, several differences in the
demographic characteristics of the heads of household are not surprising.
Compared with the heads of regularly processed households, these in
expedited service households are:
• More likely to be men
• More likely never to have married
Less likely to be elderly
' 'Expedited service households do not
typically include children. . . . and
expedited service applicants are less
likely than regularly processed
applicants to be employed.
Expedited applicants are also far less likely than regularly processed
applicants to be employed when they apply for benefits. This difference is
to be expected, given that the basic objective of expedited service is to
serve applicants with the most pressing needs. Around 60 percent of both
types of applicants, however, are fairly recently attached to the labor
force, as indicated by the percentage who report having worked at some
time within the past year.
The patterns of differences between expedited service and regular cases
observed in this study and in the 1987 study are very similar. Although
31
Table 3. Demographic Characteristics of Approved Food Stamp Applicants, by Expedited Service Status and Entitlement Criteria
Expedited Service Criteria
Monthly Income/ Shelter Expenses
Regularly Processed Expedited Service Resources Below Exceed Income/
Characteristic Applicants (%) Applicants (%) Limits i'v i Homeless (%) Resources {%)
Household composition characteristic
One-person households 32.7 56.2 ttt 58.4 68.9 29.3 ***
Households with children 60.7 38.0 ttt 35.8 29.1 62.9 ***
Female-headed with children 32.9 24.4 ttt 22.8 18.9 42.0 ***
Characteristics of household head
Age (mean years) 36 33 ttt 33 31 35 **
Female 73.5 55.0 ttt 53.3 39.3 *** 76.3 •**
Non-white 44.7 44.2 43.8 46.6 47.3
Never married 29.7 43.3 tt 45.2 44.4 22.9 **
Currently employed 24.7 7.5 ttt 6.2 22.7 ** 16.6 ***
Employed within past year" 63.2 56.7 t 55.7 56.6 66.8
Disabled 15.8 8.3 ttt 7.8 16.9 ** 12.3 *
Received food stamps previously 47.5 44.9 44.1 55.8 51.5 *
Received expedited service previously 9.8 21.8 ttt 22.4 21.2 14.4 *
Unweighted N 2885 5307 4718 147 360
'Undu|)licated criteria. Excludes cases with missing data on criteria (n = 73). Destitute migrants and seasonal farmworkers not shown separately, as sample
size in = 9) was too small to produce valid estimates.
hData from self administered applicant survey, and thus only available for 1993 sample (n = 3,426).
* Significantly different from monthly income/resources below limits category at 0.10 level.
** Significantly different from monthly income/resources below limits category at 0.05 level.
*** Significantly different from monthly income/resources below limits category at 0.01 level.
t Significantly different from regularly processed at 0.10 level.
tt Significantly different from regularly processed at 0.05 level.
ttt Significantly different from regularly processed at 0.01 level.
M>
Expedited Service in the FSP
' Expedited sen ice households have
lower income and asset levels than
households receiving regular
processing.
the demographic characteristics of food stamp cases as a whole have
changed over time, the demographic characteristics of expedited cases
relative to regular cases have changed little. For example, among all
recipients, the proportion of female-headed households has increased,
reflecting a nationwide trend, but in both studies the proportion of female-headed
households is significantly lower among expedited cases than
among regular cases.
Differences in Income, Resources, and Expenses. Because expedited
service is intended for applicants with the most urgent need for assistance,
one would expect expedited service households to have lower income and
asset levels than households receiving regular processing. The data bear
out this expectation (Table 4).
Applicants receiving expedited service have an average monthly gross
income of $154, or a little over one-quarter of the average recorded for
regularly processed cases. Regular applicants are clearly poor, with
incomes averaging 59 percent of the federal poverty level, but expedited
service applicants are in even more difficult circumstances, as their
incomes average just 19 percent of the poverty level.
The lower earnings of these households account for 60 percent of the
difference in average gross income. Expedited service cases also have
lower amounts of unearned income from almost every source, including
Social Security, AFDC. SSI. and unemployment compensation. General
Assistance is the only source providing similar amounts of income to
expedited service and regularly processed applicants. These patterns of
income reflect differences in the demographic composition of cases
described earlier.
Neither expedited service nor regularly processed cases have substantial
assets; assets for these cases average $60 and $170. respectively. Both
groups have liquid resources that, on average, are below the $100 limit for
expedited service. Even if nonliquid resources were included, most
expedited service cases would still fall below the $100 limit.
The average shelter expenses of expedited service cases are 72 percent of
the average for regular cases, despite the fact that they have only about a
quarter as much income as regular cases. This pattern reflects the presence
of two quite different kinds of households in the expedited service
caseload: homeless households with little or no housing expense and
households with shelter costs that exceed their combined income and
assets.
33
BLANK
PAGE
Table 4. Income, Resources, and Expenses of Approved Food Stamp Applicants, by Expedited Service Status and Entitlement Criteria r
Regularly Processed Expedited Service
Expedited Service Criteria'
Monthly Income/ Shelter Expenses Exceed
Applicants Applicants Resources Below Limits Homeless Income/Resources
Monthly gross income
Mean amount" $532 $154 ttt $125 $225 $394 ***
Percent with zero income 14.3 53.9 ttt 58.4 46.0 8.6 ***
Income relative tc poverty line 0.59 0.19 ttt 0.16 0.30 * 0.47 ***
Earnings
Mean amount" $272 $47 ttt $33 $114** $139 ***
Percent receiving 37.9 13.0 ttt 10.1 29.3 ** 36.0 ***
Unearned income
Mean amount" $259 $107 ttt $92 $111 $255 ***
Percent receiving 57.6 35.2 ttt 33.2 27.9 60.4 ***
Total assets
Mean amount" $170 $60 ttt $57 $76 $77
Percent reporting zero assets 64.6 80.9 ttt 82.2 75.2 68.7 **
Liquid resources
Mean amount" $96 $22 ttt $19 $42 $44 **
Percent holding 32.1 17.0 ttt 15.8 20.3 29.2 **
Shelter expenses
Mean amount" $354 $256 ttt $238 $69 *** $543 ***
Percent reporting no shelter expense 11.2 34.7 ttt 36.7 65.3 *** 0.6 ***
Unweighted N 2,885 5,307 4,718 147 360
'Unduplicated criteria. Excludes cases with missing data on criteria (n = 73). Destitute migrants and seasonal farmworkers not shown separately, as sample
size (n = 9) was too small to produce valid estimates.
"Averaged across all cases; includes those reporting no incomes/assets/expenses.
* Significantly different from monthly income/resources below limits category at 0.10 level.
** Significantly different from monthly income/resources below limits category at 0.05 level.
*** Significantly different from monthly income/resources below limits category at 0.01 level.
t Significantly different from regularly processed at 0.10 level,
tt Significantly different from regularly processed at 0.05 level.
ttt Significantly different from regularly processed at 0.01 level.
9f
Expedited Service in the FSP
Differences in Characteristics by Expedited Service Criteria. Many
officials and policymakers have expressed concern that the groups added
by the McKinney Act do not urgently need emergency assistance. This
analysis shows that although the 10 percent of applicants who were
granted expedited service by the McKinney Act have higher average
incomes than other expedited service cases, they do have a greater need
for this service than the households that receive regular processing (see
Tables 3 and 4).
Households qualifying for expedited service on the basis of income and
resources account for 90 percent of all expedited service cases, so their
profile closely resembles that of the entire expedited service caseload.
Their income and resources are extremely limited—average monthly
income is $125, and liquid resources average $19." Their shelter expenses
average $238, which is almost $100 more than their combined monthly
income and liquid resources.
Households qualifying for expedited service solely on the basis of
homelessness live in somewhat less severe circumstances than do
households that fall below the income and resource thresholds. This is
because the homeless category includes only those homeless households
that were not recorded as qualifying for expedited service under the
income and resource criteria.1' Compared to the income/ resource group,
the homeless:
• Have roughly twice as much total monthly income ($225 versus $125)
• Are much more likely to have earnings (29 percent versus 10 percent)
• Are more likely to have unearned income from Social Security and
SSI, and are less likely to receive AFDC and General Assistance.
Although the homeless group has a higher income than the income/
resource group, their income is less than half the average for regularly
processed cases.
The demographic profiles of homeless applicants and those who meet the
income/ resource criteria are quite similar. Both groups have a large
proportion of one-person households (roughly 60 percent, compared with
33 percent of regularly processed cases). Both groups include a high
proportion of male applicants and relatively few female-headed households
with children.
35
S. Bartlell
Not surprisingly, households qualifying for expedited service only because
their shelter expenses exceed their income and resources are distinguished
by exceptionally high shelter expenses—$543 per month, on average. This
far exceeds the average monthly expenditures of other expedited service
households and those of regularly processed applicants, who report
monthly shelter expenses averaging $354.
Data suggest that recent job loss may have triggered the food stamp
application for a substantial number of households with shelter expenses
that exceed their income and resources. Although 36 percent reported
some earnings in the past month, only 17 percent of the households were
employed when they applied. A quite comparable 38 percent of regularly
processed applicants reported earnings, but 25 percent of the household
heads continued to be employed when they applied for benefits.
With respect to unearned income, households qualifying for expedited
service only because their shelter expenses exceed their income and
resources look more like regularly processed households than other
expedited service households. Approximately 60 percent of the households
report receiving some unearned income, and the monthly average of $255
is virtually identical to the average for regularly processed cases.
The resources of households with shelter expenses that exceed their
income and resources are quite limited. On average, they report total
assets of $77, fairly similar to the average for other expedited service
households and less than the $170 average reported by regularly processed
households.
The demographic profile of the households with shelter expenses that
exceed their income and resources is quite similar to the profile of
regularly processed cases, and thus very different from other expedited
service cases. Most are multi-person households, three-quarters are female-headed,
and more than one-third are female-headed households with
children.
Application Processing Timeliness of benefit delivery and the accurate designation of cases for
expedited service are two aspects of FSP application processing that can
be used to measure how well current expedited service operations achieve
the intent of federal laws and regulations.
Timeliness of Benefit Delivery. A central indicator of the effectiveness of
expedited service policy is the percentage of expedited service cases that
receive their initial food stamp benefits within the 5 days mandated by
36
Expedited Service in the FSP
' 'Local offices are doing a substantially
better job of delivering expedited
service benefits within the mandated
time period than in the early 1980s."
federal law. The data show that 76 percent of expedited service cases were
authorized for benefits within 5 days, and 85 percent were authorized
within 10 days. The average time between application and authorization
was 5.7 days.
These data indicate that local offices are doing a substantially better job of
delivering expedited service benefits within the mandated period than in
the early 1980s, when data for the 1987 study were collected. That study
found that approximately 60 percent of all expedited service cases
received their benefits within 5 days. The average processing time was 7
days at that time.
A substantial number of regularly processed applicants also receive their
benefits fairly quickly. More than one third were authorized within 5 days
of application, and roughly half, within 10 days. On average, regularly
processed cases were authorized for benefits 14.8 days after they filed
their applications.
The percentage of expedited applications processed within 5 days varies
substantially by office. Some offices processed all of their expedited
applications within five days, whereas others processed only 30 to 40
percent within this period. In order to understand why some offices are
more successful than others in processing expedited applications quickly,
we examined how office-level characteristics are related to the timeliness
of benefit delivery.
Two aspects of office and workflow organization appear to positively
affect an office's ability to process expedited cases within five days. Some
offices conduct the certification interview for expedited cases on the same
day that the applicant first appears at the office. These offices authorize
benefits for 87 percent of their expedited cases within five days. Beginning
the certification process quickly seems advantageous for these offices.
Offices that screen applicants to determine their eligibility for expedited
service before scheduling the certification interview are also more likely
than other offices to meet the five-day standard for expedited cases. These
offices identify expedited cases quickly and place them on a "fast track"
for processing.
The data also suggest that smaller expedited service caseloads may help
offices meet the five-day deadline. Offices with the highest proportions of
expedited service cases (exceeding half of all applicants) approve 73
percent of applications from eligible households within five days.
37
5. Bartlett
"Postponed verification, which allows
workers to suspend normal
requirements for verifying eligibility,
appears to increase the timeliness of
benefit delivery:
compared with roughly 78 percent in offices with lower proportions,
though the observed difference is not statistically significant. Also,
supporting the theory that an office's expedited service rate affects
timeliness, small and medium-sized offices and offices in nonmetropolitan
areas, which tend to have below-average proportions of expedited service
cases, process relatively more of their expedited cases within five days
than other types of offices.
In addition to these office-level characteristics, postponed verification
appears to increase the timeliness of benefit delivery because it allows
workers to suspend normal requirements for verifying items of eligibility
in order to meet the processing standard for expedited cases. Although
offices postpone verification to different degrees, postponed verification
cases are somewhat more likely to receive benefits in five days than the
cases for which verification is not postponed (81 percent versus 71
percent), though the difference is not statistically significant.
Accuracy of the Expedited Service Designations. The effectiveness of
expedited service policy depends on the accuracy of the expedited service
designation as well as timeliness of benefit delivery. The case file records
of food stamp applicants contain information on whether or not they were
identified as being qualified to receive expedited service. Using other
information from the case file record, including income, resources, and
housing expenses, we independently determined the expedited service
status of all applicants and compared this to the status designated by the
food stamp worker.
The expedited service status of 82 percent of all applicants was correctly
determined.1" Only 6 percent of clients who received expedited service
were not actually qualified to receive it. On the other hand. 12 percent of
all applicants who appeared to be qualified for expedited service were
shown in the case record to have been processed regularly. Results from
the 1987 study are similar. That study found that 4.5 percent of all
applicants received expedited service even though they did not meet the
entitlement criteria, and 15.7 percent qualified for expedited service but
received regular processing.
The data suggest that the criterion granting expedited service to
households whose shelter expenses exceed their income and resources may
be misunderstood in some systematic fashion. Of those cases that qualified
for expedited service solely because of excessive shelter expenses, 42
percent did not actually receive expedited service. In contrast, only 15
percent of all other cases that qualified for expedited service did not
38
Expedited Service in the F5P
Payment Error Associated with
Expedited Service
' 'Expedited service cases in general
are much more likely than regular
cases to close within three months of
initial certification.
receive the appropriate designation. Either food stamp workers are
substantially less likely to understand the criterion related to high shelter
costs, or they are more likely to ignore the criterion in the belief that these
cases do not urgently need assistance. Apart from this issue, the data
suggest that random human error accounts for much of the remaining
misdesignation.
Since expedited service policy was first implemented in 1979, officials and
policymakers at all levels of government have been concerned about the
effect of the provisions on program integrity, particularly on the potential
for error and fraud. The 1987 study found that expedited service policy
did not lead to more payment errors. Concerns about program integrity
surfaced again, however, with the passage of the McKinney Act, as many
argued that the groups added by the act are particularly difficult to
process.
The most accurate way to estimate issuance errors associated with
expedited service would be to measure the errors directly. This approach
would be quite costly, however, and would require extensive federal-state
planning, as it would involve special quality control reviews of a
nationally representative sample of expedited service cases." The approach
in this study has been to examine measures that serve as indicators of
potential error. Specifically we examined patterns of benefit change in the
early months after initial issuance. If expedited cases are more likely than
regular cases to experience early terminations or decreases in their
benefits, this would suggest that issuance errors are more likely to occur in
association with expedited service cases than with other cases.
Expedited service cases in general are much more likely than regular cases
to close within three months of initial certification. As shown in Table 5,
the overall or unadjusted three-month termination rate is 8.1 percent for
regular cases and 16.0 percent for expedited service cases. At 21.3
percent, the rate for expedited service cases with postponed verification is
still higher. The patterns are similar for combined rates of termination and
benefit decrease. Rates are higher for expedited cases compared with
regular cases, and postponed-verification cases have the highest rates of
all.
We cannot conclude from this evidence alone that the higher rates for
expedited cases are due to errors in initial allotments. Benefit reductions
and terminations can also reflect changes in household circumstances or
failure to follow procedures (e.g., for not providing all verification). Some
types of households are more likely than others to experience these
39
5. Bartlell
changes, and we thus expect that some of the observed difference in the
rates is due to differences in the caseload composition of expedited versus
regular cases. To control for these differences, we estimated multivariate
models of the likelihood of early termination or benefit reduction that
control for case characteristics.12 The models were used to estimate the
adjusted termination and benefit reduction rates shown in Table 5.
The differences in the adjusted rates between expedited service and
regular cases are substantially smaller than the differences in the
unadjusted rates. Nevertheless, even after office and case characteristics
are controlled for. the benefits of expedited service cases are still
significantly more likely to be terminated early or reduced, compared with
regular cases.
These differences stem entirely from postponed verification. The expedited
service cases with postponed verification are especially prone to early
termination and early benefit reduction. The benefits for 30 percent of
expedited cases with postponed verification were either terminated early or
decreased, compared to benefits for 21 percent of regular cases."
Impact on Overpayment Error. Given the evidence that expedited
service with postponed verification leads to at least some payment error in
initial issuances, it would be useful to know the magnitude of the error in
order to gauge whether this presents a relatively large or small concern for
expedited service policy. We can use the rates of termination and benefit
decrease (calculated above), along with our estimates of the expedited
service rate and the postponed verification rate to estimate the potential
national overpayment error attributable to postponed verification. We refer
Table 5. Early Termination and Benefit Decrease Rates by Expedited Service Status: Unadjusted and Adjusted for
Household Characteristics, Certification Period, and Site
Unadjusted Adjusted
Termination
Termination or
Benefit Decrease Termination
Termination or
Benefit Decrease
Expedited service cases
Verification postponed
Verification not postponed
All
21.3 %*»*
12.0*
16.0 *••
34.6 %•**
20.7
26.5 **
16.2 %***
8.6
12.0 **
29.5 ♦*•
19.0
23.8 ***
Regular cases 8.1 19.4 10.3 21.1
* Significantly different from regular cases at the 0.10 level.
•* Significantly different from regular cases at the 0.05 level.
*** Significantly different from regular cases at the 0.01 level.
40
Expedited Service in the FSP
SUMMARY AND CONCLUSIONS
Expedited Service Caseload
to this as "presumptive overpayment error," as it represents our best
estimate of the dollar value of the additional error that results from the use
of i stponed verification.14
Our calculations show that the amount of error attributable to postponed
verification is fairly small. For expedited cases with postponed
verification, the estimated national presumptive overpayment error ranges
from $14 million to $30 million per year (depending on whether one uses
the adjusted or unadjusted rates). Perhaps large in absolute terms, these
estimates of error are fairly small relative to the amount of food stamp
benefits issued, making up only 0.1 to 0.2 percent of total issuances to all
active cases. This compares to the cost of overall overpayments which for
FY 1992 was 8.2 percent of total issuances.
All food stamp officials surveyed in this study voiced support for
expedited service policy— they consider it successful in quickly
alleviating the problems faced by people with inadequate access to food.
However, these individuals as well as others who have been involved with
expedited service since it was first implemented in 1979 have expressed a
number of reservations about the policy, particularly the provisions
legislated by the McKinney Act. Their concerns tend to focus on the
burden of expedited service on local offices, the entitlement criteria, and
the policy's potential to encourage fraud and error.
Many food stamp officials felt that the McKinney Act, by expanding the
criteria under which applicants qualify for expedited service, added a
large pool of applicants to the expedited sennce caseload. Others felt that
the proportion of expedited cases grew during the rapid rise in the overall
food stamp caseload during the late 1980s and early 1990s. In addition to
the size of the expedited caseload, many expressed concerns about the
burden placed on local offices by requiring that expedited cases be
processed within five days.
The expedited service rate, or the proportion of applicant households
receiving expedited processing, remained essentially constant over the past
decade despite the McKinney Act and the expansion of the food stamp
caseload. We found that fully 90 percent of applicants qualified for
expedited service because their income and resources were below the
established limits. Only 10 percent of applicants qualified solely because
of the provisions of the McKinney Act.
Because the food stamp caseload grew substantially over the past 10
years, the actual number of expedited service applicants increased by more
41
5. Bartlett
than 20 percent. This may explain why many perceived the McKinney Act
to have had a sizeable impact on the expedited service rate. Yet, despite
the size of the expedited service caseload, local offices are now more
successfully processing applications within the required timeframe than
they did a decade ago. More than three-quarters of expedited applicants
received their benefits within five days, compared to 60 percent in the
early 1980s.
Inequities in Expedited Service Many program officials are concerned about the criteria added by the
Criteria McKinney Act that provide expedited service to the special populations—
homeless and those judged in danger of becoming homeless. They argue
that these criteria dilute the effectiveness of expedited service and create
inequities because they include households whose needs are less urgent
than those of some applicants who do not qualify for expedited processing.
Expedited service households generally face quite severe economic
conditions— their average income is approximately 30 percent of the
average of regularly processed cases. By definition, the income and
resources of households qualifying solely under the McKinney Act do not
fall below the basic thresholds. As a result, their average monthly income
is considerably above that of other expedited service households, though
still well below the average for regularly processed cases.
These findings suggest that on average, the McKinney Act does give
expedited service to households more urgently in need of assistance than
those who do not qualify for expedited service. Because there is no simple
and universal way to measure the urgency of a household's need,
however, any criterion for expedited service that attempts to approximate
need will introduce some inequities into the system—that is, some
applicants who receive expedited service will seem to a reasonable
observer to be less urgently needy than some applicants who do not
qualify. Because the pure McKinney households generally face less severe
circumstances than those who meet the traditional income and resource
criteria, it is practically inevitable that the number of such inequities has
increased since the act was passed.
The two groups added by the McKinney Act represent ruite different
types of households. The homeless households added by the act look very
similar to the homeless households that also qualify under the income/
resources criterion; the obvious exception is that the pure McKinney
households have somewhat higher average incomes. In contrast, the
demographic profile of households with excessive shelter expenses is very
similar to that of regularly processed cases. Their distinguishing features
42
Expedited Service in the FSP
are high shelter expenses which are higher on average than regularly
processed cases, and a lower average monthly income to support them.
Thus, it seems likely that, to the extent that inequities have been created,
they mainly concern households that qualify for expedited service because
of excessive shelter expenses.
Overpayment Errors Since expedited service policy was first implemented, officials and
policymakers have been concerned that it posed a potential threat to
program integrity. Their concerns have focused on three issues. Many
have argued that, because households can obtain benefits with very little
verification of their circumstances, this would lead some to either
intentionally or unintentionally misrepresent their situation. The potential
for error also increases, according to some arguments, because workers
must process applications in a relatively short time frame, which can lead
to careless mistakes. Finally, some have suggested that processing
expedited cases detracts from the office's efforts to process regular cases,
causing more error in that portion of the caseload.
Overpayment errors resulting from expedited service do not pose a
substantial problem for the FSP. The one aspect of expedited policy that
apparently does lead to some overpayment errors concerns the use of
postponed verification. Patterns of benefit change in the early months after
initial issuance showed that, after we control for differences in caseload
composition, expedited service cases with postponed verification had
higher rates of termination and benefit decrease than did either regularly
processed cases or expedited cases that did not have postponed
verification. If all of these "extra" early terminations represented
eligibility errors—that is. if none of these cases should have received any
food stamp benefits at all—this would imply that each year, erroneous
payments of $14 million are issued nationwide because of postponed
verification. If these errors were counted by the quality control system,
they would add about 0.1 percentage point to the overpayment error rate
for food stamps, raising the 1992 error rate from 8.2 to 8.3 percent.
Unanswered Questions There is general agreement that expedited service policy fulfills an
important function—it gets benefits to clients in need relatively quickly,
allowing them to meet basic food needs without delay. However, this
study did reveal two problems with current operations. First, not all who
are entitled to expedited service receive their initial benefits within the five
days mandated by law. Second, postponed verification does lead to some
apparent overpayment in the initial food stamp issuance.
43
5. Bartlett
Future research might seek strategies for improving timeliness while
decreasing the use of postponed verification. We know from the current
study that some local offices are able to process virtually all their
expedited applications within five days. We also know that at least some
of these offices are able to do so without extensive use of postponed
verification. Understanding the policies and practices of the more
successful offices could allow others to improve their delivery of
expedited service.
NOTES I. The findings presented in this paper are condensed from a larger report: Bartlett. Susan.
Nancy R. Burstein, and Elsie C. Pan. "Evaluation of Expedited Service in the Food Stamp
Program." Cambridge. MA: Abt Associates Inc.. June 1995. In addition to my co-authors
on the report, William Hamilton at Abt and Barbara Murphy and Christine Kissmer of the
Office of Analysis and Evaluation at FCS provided invaluable guidance and support
throughout the entire study.
2. Esrov. Linda. James Hersey, John Mitchell, John Moeller. and Mary Dent. "Evaluation
of Expedited Service in the Food Stamp Program." Washington, DC: U.S. Department of
Agriculture, Food and Nutrition Service; and SRA Technologies. Inc.. April 1987. We refer
to this as the "1987 Study" throughout this paper.
3. This study includes only those applicants who were approved to receive food stamp
benefits. Applicants denied benefits were excluded whether or not they were initially
processed under expedited procedures.
4. We initially selected and recruited 60 local offices for the study. One dropped out just
prior to the start of data collection activities.
5. We excluded offices with monthly caseloads below 300 because they could not support
the necessary cluster sizes of applicants. These offices accounted for only 0.81 percent of
the national total caseload.
6. Local offices located within a metropolitan statistical area (MSA) are considered
metropolitan, and those outside an MSA are non metropolitan areas.
7. The analyses presented in this section are based solely on the classification*s) recorded in
the case file. The analysis does not "second guess" the worker's classification by
examining, for example, the recorded amounts of income and resources. We examined the
distributions for 1991-1992 and 1993 separately, and found that they were virtually
identical.
8. As Table 4 shows, 58 percent had zero gross income. Of the 42 percent with some
income, 13 percent have reported incomes of less than $150, and 29 percent have reported
incomes of $150 or more. The latter groups should not be entitled to expedited service on
the basis of their income, suggesting that they were erroneously classified in the case file
records. Some, however, may qualify for expedited service according to other criteria.
9. It is possible that some workers recorded some cases as qualifying under only the
homeless criterion even though they met other criteria as well.
10. Includes cases qualifying for expedited service (8 percent) that were designated as
regular cases but received their benefits within five days, thus in effect receiving expedited
service.
44
Expedited Service in the FSP
11. This extensive data collection would be required because quality control reviews
currently conducted on expedited service cases apply more lenient standards (because of the
use of postponed verification) than are applied to regular cases. The more stringent reviews
would be required to estimate total payment error accurately.
12. The models also include site indicators and the length of the initial certification period,
an indicator of the eligibility workers assessment of the stability of the households
circumstances.
13. Cases that qualify for expedited service solely on the basis of the McKinney Act exhibit
patterns similar to those of other expedited service cases.
14. Stales are not currently held liable for any payment error to expedited service cases that
are processed according to policy, nor do these payment errors count in the calculation of
quality control error rates.
45
% m.im MS
Access of FSP Participants to Food Retailers
Richard Mantovani, Lynn Daft, James Welsh, and
Theodore Macaluso
INTRODUCTION AND
BACKGROUND
"At the basic level of physical
proximity, are food stamp-authorized
retailers located where food stamp
participants live?"
The Food Stamp Act of 1977, as amended, declared it the policy of
Congress "to safeguard the health and well-being of the Nation's
population by raising levels of nutrition among low-income households."
To alleviate hunger and malnutrition. Congress authorized "a food stamp
program... which will permit low-income households to obtain a more
nutritious diet through normal channels of trade by increasing food
purchasing power for all eligible households who apply for
participation."'
The ability of the Food Stamp Program (FSP) to meet these health and
nutrition goals depends on the nature and characteristics of the "normal
channels of trade" actually accessible to program participants. Retail food
stores are one of these channels. Because access to retail food stores is
critical to the success of the FSP, a broad base of over 200,000 stores has
been authorized to accept food stamps. The magnitude of this number
becomes more meaningful in light of the total number of supermarkets in
the United States, which is about 30,000. In addition to supermarkets,
FSP-authorized stores include large and small groceries, convenience
stores, gas/grocery stores, food delivery routes, general stores, and health
food and other specialty stores (such as meat and fish markets).
While this strategy of broad authorization is likely to increase access to
food stores, the following questions related to access remain:
• At the basic level of physical proximity, are food stamp-authorized
retailers located where food stamp participants live?
If so, do the retailers accessible to participants stock foods that
support a nutritious diet? What is the quality and variety of food sold
by food retailers and accessible to participants?
• What is the cost of a market basket at the stores to which participants
have access?
The Food and Consumer Service (FCS) has initiated a research program to
address these issues. This paper, which presents the result of part of this
effort, documents an initial exploration of the first two issues above. We
47
Mantovani et al.
DATA AND METHODS
address them by answering the question: how much distance must food
stamp participants travel to reach an authorized supermarket or large
grocery store? We use distance to a supermarket or large grocery store,
rather than distance to any authorized store, because supermarkets and
large grocery stores are the mostlikely types of stores to stock a wide
variety of food.2 These two types of authorized food retailers became a
proxy measure for the true topic of interest: access to a variety of high-quality
food at competitive prices. Subsequent studies funded by FCS will
measure food quality and price directly, rather than through this proxy
measure of store type.
The sites reported on in this paper were selected from the 40 primary
sampling units (PSUs) used in a nationally representative study on retailer
characteristics sponsored by FCS. We selected 5 of the 40 sampling units
on a purposive basis to develop a cross-section of areas that differ in
terms of urbanization, income, and ethnic characteristics. The five PSUs
comprised eight sites for the analysis.' Three of the sites were highly
urbanized areas: the South East area of Los Angeles City; Baltimore,
Maryland; and Pasadena, California. Another three were smaller
metropolitan statistical areas (MSAs) adjacent to rural areas: Kanawha and
Boone counties. West Virginia; Dona Ana County, New Mexico; and
Antelope Valley, Los Angeles County, California. The remaining two sites
were non-MSA (or rural) areas with small central cities or towns: Dillon
and Marion counties in South Carolina, and Otero and Lincoln counties in
New Mexico.
RESEARCH FINDINGS
We derived measures for proximity using geographic information systems
(GIS) software to locate and map the street address of authorized food
retailers and FSP participants in each community.
Communities were described according to information obtained from site
visits and census demographics. We also used information on food stamp
issuances to participants residing in a particular ZIP code area and
redemptions at all authorized stores within the ZIP code to calculate an
inflow/outflow measure. This measure indicates where food stamps are
flowing throughout the area and identifies areas in which retailers capture
more food stamps than are locally issued. The major sources of data were
the FCS authorized retailer tracking system (Store Tracking and
Redemption Subsystem), participant data files obtained from the respective
state or county jurisdictions, 1990 census data, and interviews conducted
with local food system experts during a set of visits to the sites.
Tables 1A and IB present an overview of findings. For the three types of
areas. Table 1A shows the proportion of food stamp participants living
48
Access to Food Retailers
Table 1A. Distances of Food Stamp Recipients to Authorized Supermarkets and Large Grocery Stores
Total Under .25 Under .5 Under 1 Under 2 Under 5 Median Mean
Area Recipients Mile Mile Mile Miles Miles Distance Distance
Highly Urbanized Areas
Baltimore 13.393 44.5% 95.7% 100.0% 100.0% 100.0% .27 .28
South East LA 28.319 47.3% 90.4% 100.0% 100.0% 100.0% .26 .28
Pasadena 6,324 52.3% 93.0% 100.0% 100.0% 100.0% .24 .26
Smaller MSAs
Kanawha and Boone Counties 14.129 16.0% 39.2% 68.7% 82.1% 97.2% .60 1.13
Palmdale 4.325 16.9% 38.3% 75.7% 85.6% 95.7% .57 1.16
Dona Ana County 9.843 3.1% 12.5% 44.0% 57.4% 63.7% 1.23 5.52
Non-MSAs
Dillon and Marion Counties 4.987 16.4% 44.2% 84.3% 87.1% 92.2% .54 1.51
Otero and Lincoln County 3.009 27.9% 48.8% 71.0% 80.7% 93.0% 0.52 2.06
Source: Macro International Inc. The Authorized Food Retailers Characteristics Study. Contract No. 53-3198-3-007. USDA/
Food and Consumer Service, Office of Analysis and Evaluation. 1994.
Table IB. Food Stamp Dollar Flows Within Study Areas
Total
Hi
Redemptions
Million of
Dollars)
Redemption to
Issuances Ratio
(Study Areas)
No. of Sub-Areas
with Redemption to
Issuances Ratio*
Number of Areas With
Redemption to
Issuance Ratio
Area 0-0.74 0.75-1.25 > 1.25
Highly Urbanized Areas
Baltimore
South East LA
Pasad^rsa
70.4
47.5
9.8
0.94
0.93
0.94
6
4
3
12
1
4
10
1
1
2
Smaller MSAs
Kanawha and Boone Counties
Palmdale
Dona Ana County
34.1
7.0
18.0
1.24
0.74
0.72
14
3
4
4
2
3
4
1
1
6
0
0
Non-MSAs
Dillon and Marion Counties
Otero and Lincoln County
10.7
64
1.17
1.03
15
5
7
3
32
50
Source: Macro International Inc. The Authorized Food Retailers Characteristics Study. Contract No. 53-3198-3-007. USDA/
Food and Consumer Service, Office of Analysis and Evaluation, 1994.
* Sub-areas are generally defined as ZIP codes but in some cases may include several contiguous ZIP code areas.
49
Mantovani et al.
within a given distance of a supermarket or large grocery with over
$500,000 in sales.4 Table IB presents information on food stamp dollar
flows within each of the study areas using the redemptions-to-issuances
ratio. This table presents the overall redemption-to-issuances ratio for the
study area and the number of sub-areas with a lower, a higher, or roughly
the same level of redemptions as issuances.
Highly Urbanized Areas In the three highly urbanized areas in Table I A, over 90 percent of food
stamp recipients live within one-half mile of a supermarket or large
grocery, and all participants live within one-mile of such a retailer. The
median distance is a quarter of a mile. The three areas vary little in terms
of distance of participants from retailers, although there are distinct
differences among the three cities with regard to retailer availability and
participant utilization.
Baltimore, Maryland. The study area in Baltimore City is near the center
of the city (Figure I). Overall, approximately 74,000 households and
207,000 individuals live within this area.5 African Americans constitute
approximately two-thirds of the population, ranging from 64 percent in the
Union Square area to 93 percent in the Harlem Park area. Hispanics
constitute less than 1 percent of the population. The 1990 census indicates
that about one-fifth to one-half of the households are under the poverty
line. With regard to transportation, there is an extensive light rail and bus
system and a new subway system that facilitates travel within the central
city area. Automobile access is somewhat limited, since the 1990 census
reported that from one-half to two-thirds of the households do not have
access to an automobile.
FSP participants reside largely in two large clusters within the area. The
first cluster is in the eastern part of the city, generally in the Greenmount
and Johns Hopkins Hospital areas. The second cluster is located on the
west side of the city and covers the areas from Bolton Hill to Union
Square. Lexington Market, Downtown, and Clifton Park comprise areas
with fewer participants.
' 'Because Isupermarkets and large
grocery stores] are well distributed
within the Baltimore study area, nearly
all the participants live within one-half
mile of a large retailer."
It is noteworthy that the study area has proportionately fewer supermarkets
and large grocery stores than ether areas in Baltimore City. Supermarkets
and large grocery stores account for less than 4 percent of the retailers in
the core area, which is lower than the percentage in the Baltimore City
area as a whole and much less than the national percentages for
supermarkets and large grocery stores. However, because these stores are
well distributed within the area, nearly all the participants live within one-half
mile of a large retailer (Figure 2). Of the participants who are not
50
Access to Food Retailers
Figure 1. Baltimore study area, general orientation map
Tha location of tin Baltimora
Study Araa within tha City
It shown In tha map on tha laft.
<s^
Streets
City Limits
Railroads
Shoreline
Study Area
51
Mantovani et al.
within one-half mile of a supermarket or large grocery, many are clustered
in several pockets (Figure 2).
An analysis of the inflow and outflow of redemptions and issuances by
ZIP code areas and other information on redemptions indicates that the
use of large stores varies considerably.6 In Clifton Park, supermarkets are
an important source of food for FSP participants, while in Greenmount,
they are a minor source. Interviews with local food system experts suggest
that this pattern is influenced by widely dispersed "market places" within
the city. These urban indoor farmers markets (Lexington, Lafayette,
Hollins, and Northeast Markets, for example) are used to a large degree
by local residents. They supply a wide range of perishable foods
throughout the year and serve as alternatives to supermarkets for certain
types of foods.
' More than 90 percent offood stamp
households [in the Pasadena-Los
Angeles study area] are located within
one-half mile of a supermarket or large
grocery.''
Pasadena, California. Pasadena, a city of 23 square miles with a
population of 131,591, is located at the southern foot of the San Gabriel
Mountains, approximately 15 minutes by car from downtown Los Angeles
(Figure 3). The city is economically diverse, with some relatively affluent
areas (toward the eastern end of the city) and some low-income areas. The
analysis in Pasadena focused on an area northwest and north of the
Foothill Freeway and Colorado Boulevard areas. Total population is
80,685, and no one ethnic group is notably dominant. In the northwest
portion of the area, approximately 40 to 50 percent of the population is
below 125 percent of the poverty level.
More than 90 percent of food stamp households are located within one-half
mile of a supermarket or large grocery, and more than 50 percent are
located within one-quarter of a mile. Figure 4 indicates that FSP
participants farther than one-half mile from a large food store are scattered
except for those in a diagonal pocket in the northwest corner and a
vertical stretch east of South Los Robles Avenue.
There are several large national chain branch food stores in the area,
although none is in the lower-income northwest section of the city.
Analysis of redemption data indicates that residents of the northwest part
of the city, although they live near larger stores, tend to use their food
stamps mainly in other areas; food stamps they do use in the northwest go
toward purchases from small retailers in that area. (It is important to
remember that this paper explores proximity to supermarkets/large grocery
stores, not the quality, variety, or price of food available in these stores.
Further research will explore whether these redemption patterns reflect
quality, variety, price of food available in different stores.)
52
Access to Food Retailers
Figure 2. Baltimore study area, half-mile access to FSP SM/GS with annual sales over 500,000
H Recipients Beyond 1/2 Mile
El AH Other Recipient Households
City Limits
ZIP Codes
Shoreline
Study Area
53
Mantovani el al.
Figure 3. Pasadena-Los Angeles study area, general orientation map
:WB8&; <Ci
■ ■>"■;■
The location of the Pasadena
component of the Los Angeles
Study Area is shown in this
ZIP Code map.
ZIP Codes
Highways
Streets
Railroads
Study Area
• .
- - '
■ I
■ - -. i—J» - * i- -^ Ir-UJ ! : "• '- ■ ±7 j . •
f.
' j , . . .
L \V Z
r, A\ • ■ - • ...
■•."■•vA •' ' I ' '■ . : • i
•- ■
/
.
flu urtillf fmai
■ ■
025 0.25 0.5
54
Access to Food Retailers
Figure 4. Pasadena-Los Angeles study area, half-mile access to FSP SM/GS with annual sales over $500,000
H Recipients Beyond 1/2 Mile
GO A// Of/iar Recipient Households
ZIP Codes
Highways
Study Area
2
o •
OD • O
0 o o
o a °
8o
o o O
'oo °
■ o
*>o e
0<°
00
o o
Son
o
o o
'Oo
O O O O Oo
o
o
o
o
o
fo o o o o 8 o
o°0 * °
8°°0o o0-° °
O O 0° °0
is o
o o
§ OD
o o
Q O 000
o
o o
o
h
\
55
Mantovani et al.
"The 1990 census indicates that one-quarter
to one-third of the population
in (the South East Los Angeles study
area] has no access to a car.
South East Los Angeles. This area, south of downtown Los Angeles,
contains the sections of Florence and Watts in the east and what is known
as South Central Los Angeles (Figure 5). The study area, which is low-income
ano highly urban, has a population of 240,444, 60 percent of
whom are Hispanic. In most neighborhoods, the range of persons living
below 125 percent of the poverty line is from 40 to over 50 percent. The
study area is surrounded by other areas with similar or higher rates of
poverty, except to the southeast. Residents must shop within the area or in
similar areas unless they are willing to travel long distances. The bus
system is extensive and provides adequate access, although there is some
question concerning its safety and the ease of making connections. There
is also a transit subway stop on the eastern edge of the area. The 1990
census indicates that one-quarter to one-third of the population in these
areas has no access to a car.
Ninety percent of the food stamp participants live within a half-mile of an
authorized supermarket or large grocery store. In general, these stores
redeem about three-quarters of the total food stamps redeemed in the area.
It is clear that food stamp participants do their major shopping at large
stores. Figure 6 shows where FSP participants live. While most reside
within one-half mile of a supermarket or large grocery store, participants
outside this distance are not scattered but live in a few distinct clusters.
The pattern of redemptions within the area shows that food stamps flow
from some communities to others within the study area. For instance, the
redemption-to-issuance ratio in the Florence area is greater than that in
other areas in the city.
Smaller Metropolitan Statistical
Areas
Small MSAs are defined by a moderately sized city surrounded by
sparsely populated areas. The three small MSAs in this study were
Kanawha and Boone counties, West Virginia; Dona Ana County, New
Mexico; and Palmdale, California, which is part of the Los Angeles MSA
but distinct from the southern parts of the county just described. Unlike
participants in the highly urban areas, few participants in small MSAs
have a supermarket or large grocery store within one-half mile of their
residence. Over two-thirds of FSP participants in Kanawha and Boone
counties live within one mile of a larger store, and 82 percent live within
two miles. In Palmdale, 76 percent live within one mile and 86 percent
live within two miles of a larger store. In Dona Ana County, 44 percent
live within one mile, and 58 percent live within two miles of a large
retailer. Because of the variation in the size of these areas and the
dispersion of participants within them, we have geo-mappcl proximity
using criteria that differ by area: one mile in the West Virginia and New
56
Access to Food Retailers
Figure 5. Southeast Community-Los Angeles study area, general orientation map
■ . -
3 ■-»•-.
ft
w • ■
The location of the Southeast
Community component of the
Los Angeles Study Area is
shown in this ZIP Code map
ZIP Codes
Highways
Streets
Railroads
Study Area
57
Mantovani et al.
Figure 6. Southeast Community-Los Angeles study area, half-mile access to FSP SM/GS with annual sales over
$500,000
0 Recipients Beyond 1/2 Mile
E) All Other Recipient Households
7IP Codes
Highways
Study Area
58
Access to Food Retailers
Mexico counties, and one-half mile in the Palmdale area, where
participants are largely located in the city. Like the highly urbanized areas,
the small MSAs differ distinctly in terms of basic demographic/geographic
characteristics and access patterns.
Kanawha and Boone Counties. Kanawha and Boone counties are located
in south central West Virginia (Figure 7). Kanawha County, the most
populous in the state, is the home to Charleston, the capital and the central
city of the Charleston MSA. Boone County, although contiguous to
Kanawha County, is not part of this MSA and is not considered to be
economically integrated with Charleston. Kanawha, the larger county,
covers more than 901 square miles and had a population of approximately
207,000 in 1990. Over one-quarter of the population (58,000) is located in
the city of Charleston, and a large portion of the remaining population
resides in areas along the Kanawha River. Other population centers (e.g.,
Elk View and Clendenin) lie along the Elk River in the northern portion
of the county.
Boone County has a population of only 25,870 spread over 503 miles.
Madison and Danville, located near each other in the northwest portion of
the county, had populations of 5,000 each in 1990. Other small towns
within the county lie along or just off the state highways. Other
differences between the counties relate to the level of urbanization—71
percent in Kanawha County and 12 percent in Boone County—and the
poverty level, which is almost two times as high in Boone County as it is
in Kanawha County (30 percent and 15 percent, respectively).
Both the mountains and the Kanawha River have been instrumental in the
location of and access to communities. The majority of the population of
Kanawha County is dispersed in the mountain hollows and along the
northern and southern banks of the Kanawha River in mid-sized and small
communities. Because much of Kanawha and all of Boone County are
rural and isolated, people depend on the automobile for much of their
daily activities. Data from the 1990 census indicate that 85 percent of the
households in Kanawha County and Boone County have access to an
automobile.
"Within Charleston . .. 50 percent of
FSP households live within a half-mile
of an authorized large retailer. Outside
of Charleston, 30 percent live within a
half-mile.
Geographically, the patterns of access can best be described in terms of
several subareas (Figure 8). As expected, in Kanawha County, Charleston
and the surrounding towns have the majority of large stores and account
for the largest proportion of r