Food Stamp Overpayments: Households in Different States Collect Benefits
for the Same Individuals (Letter Report, 08/06/98, GAO/RCED-98-228).

Pursuant to a congressional request, GAO provided information on
ineligible individuals who are improperly participating in the
Department of Agriculture's (USDA) Food Stamp Program, focusing on: (1)
determining how many individuals were included as members of food stamp
households in more than one state during 1996 and the estimated value of
the benefits that were improperly issued to those households; (2)
determining how these individuals could be improperly included without
being detected; and (3) identifying an option for detecting or
preventing future food stamp overpayments caused by such duplicate
participation.

GAO noted that: (1) among the four widely separated states GAO reviewed,
GAO identified over 20,000 individuals who were potentially improperly
included in food stamp households in at least two of those four states
at the same time during calendar year 1996; (2) while GAO cannot
estimate the potential amount of overpayments nationwide due to this
duplicate participation, the households in those four states improperly
collected an estimated $3.9 million in food stamp benefits; (3)
additional evidence of the scope of this problem is indicated by a
September 1997 Department of Health and Human Services computer match of
15 states and the District of Columbia, which found 18,000 potential
cases of duplicate participation in the public assistance programs,
including food stamps; (4) interstate duplicate participation goes
undetected because there is no national system to identify food stamp
participation in more than one state; (5) welfare reform legislation of
1996 contains work requirements for the Food Stamp Program and time
limits for Temporary Assistance for Needy Families that can be fully
enforced only by interstate tracking of participation in public
assistance programs; (6) however, the law does not require a national
tracking system to be established, and no federal agency is responsible
for creating a national system to facilitate such tracking; (7) although
the states have been working individually to modify their systems to
meet welfare reform requirements, few states have made progress in
developing automated systems to track participation outside their
borders; (8) while states may currently learn of some duplicate
participation from the Social Security Administration or through their
own matching efforts with neighboring states, they rely primarily on
applicants and clients to truthfully identify who resides in their
households; (9) in the absence of a comprehensive national database or
information system to track participants receiving public assistance,
creating a USDA-managed system to collect and disseminate information on
national participation in the Food Stamp Program could provide an
efficient and effective means to identify duplicate participation and
help prevent food stamp overpayments; (10) rather than relying on states
to individually develop programs to identify and prevent duplicate
participation, it would be more efficient for USDA to develop a single
nationwide system; and (11) once established, such a system could be
expanded to track additional information that would help states enforce
welfare reform provisions relating to food stamps.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  RCED-98-228
     TITLE:  Food Stamp Overpayments: Households in Different States 
             Collect Benefits for the Same Individuals
      DATE:  08/06/98
   SUBJECT:  Computer matching
             State-administered programs
             Overpayments
             Program abuses
             Management information systems
             Food relief programs
             Social security benefits
             Welfare benefits
             Eligibility determinations
IDENTIFIER:  Food Stamp Program
             USDA Temporary Emergency Food Assistance Program
             Aid to Families with Dependent Children Program
             Medicaid Program
             California
             Florida
             New York
             Texas
             AFDC
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to the Chairman, Committee on Agriculture, Nutrition, and
Forestry, U.S.  Senate

August 1998

FOOD STAMP OVERPAYMENTS -
HOUSEHOLDS IN DIFFERENT STATES
COLLECT BENEFITS FOR THE SAME
INDIVIDUALS

GAO/RCED-98-228

Food Stamp Overpayments

(150283)


Abbreviations
=============================================================== ABBREV

  EBT - electronic benefit transfer
  FNS - Food and Nutrition Service
  HHS - Health and Human Services
  OMB - Office of Management and Budget
  RSDI - Retirement, Survivor and Disability Insurance
  SSA - Social Security Administration
  SSI - Supplemental Security Income
  TANF - Temporary Assistance for Needy Families
  USDA - U.S.  Department of Agriculture

Letter
=============================================================== LETTER

RCED

B-280311

August 6, 1998

The Honorable Richard G.  Lugar
Chairman, Committee on Agriculture,
 Nutrition, and Forestry
United States Senate

Dear Mr.  Chairman: 

During the last 2 years, we have responded to the Committee's
interest in identifying ineligible individuals who are improperly
participating in the Food Stamp Program administered by the U.S. 
Department of Agriculture (USDA).  In 1997 and 1998, we reported that
millions of dollars in food stamp overpayments in four states were
caused by counting thousands of inmates of correctional institutions
and thousands of deceased individuals as household members.\1 In
these reports, we recommended several actions, including the use of
automated systems, to identify such ineligible participants.  In
response to your request, this report focuses on individuals who were
included as members of recipient households in more than one state
during the same period, a problem referred to as "duplicate
participation." Because many states already compare information with
neighboring states to identify duplicate participation, we focused on
duplicate participation in widely separated states. 

Specifically, we (1) determined how many individuals were included as
members of food stamp households in more than one state during 1996
and the estimated value of the benefits that were improperly issued
to those households, (2) determined how these individuals could be
improperly included without being detected, and (3) identified an
option for detecting or preventing future food stamp overpayments
caused by such duplicate participation. 

To identify individuals concurrently counted as members of households
receiving food stamps in more than one state, we conducted a computer
match comparing calendar year 1996 food stamp rolls of four states
(California, Florida, New York, and Texas) against one another.\2
These four states account for 35 percent of the nation's participants
in the Food Stamp Program.  (See app.  I.) To provide the highest
level of confidence that our analyses resulted in valid matches, we
used only those matches that (1) met the most stringent criteria used
by the Social Security Administration (SSA) to verify Social Security
numbers and (2) showed that food stamp benefits were issued to the
households on behalf of the same person during the same month or
months. 


--------------------
\1 Food Stamps:  Substantial Overpayments Result From Prisoners
Counted as Household Members (GAO/RCED-97-54, Mar.  10, 1997) and
Food Stamp Overpayments:  Thousands of Deceased Individuals Are Being
Counted as Household Members (GAO/RCED-98-53, Feb.  13, 1998). 

\2 We used the 1996 participation information obtained in our earlier
review of food stamp participation by deceased household members
because it was readily available and the time required to obtain more
current data could have delayed our review significantly.  Because
California does not maintain statewide participation information, we
used state eligibility information in our match.  For California, we
determined that eligibility for the program is representative of
actual participation (see app.  III). 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Among the four widely separated states we reviewed, we identified
over 20,000 individuals who were potentially improperly included in
food stamp households in at least two of those four states at the
same time during calendar year 1996.\3 While we cannot estimate the
potential amount of overpayments nationwide due to this duplicate
participation, the households in those four states improperly
collected an estimated $3.9 million in food stamp benefits. 
Additional evidence of the scope of this problem is indicated by a
September 1997 Department of Health and Human Services computer match
of 15 states and the District of Columbia, which found 18,000
potential cases of duplicate participation in the public assistance
programs, including food stamps. 

Interstate duplicate participation goes undetected because there is
no national system to identify food stamp participation in more than
one state.  Welfare reform legislation of 1996 contains work
requirements for the Food Stamp Program and time limits for Temporary
Assistance for Needy Families that can be fully enforced only by
interstate tracking of participation in public assistance programs. 
However, the law does not require a national tracking system to be
established, and no federal agency is responsible for creating a
national system to facilitate such tracking.\4 Although the states
have been working individually to modify their systems to meet
welfare reform requirements, few states have made progress in
developing automated systems to track participation outside their
borders.  While states may currently learn of some duplicate
participation from the Social Security Administration or through
their own matching efforts with neighboring states, they rely
primarily on applicants and clients to truthfully identify who
resides in their households. 

In the absence of a comprehensive national database or information
system to track participants receiving public assistance, creating a
USDA-managed system to collect and disseminate information on
national participation in the Food Stamp Program could provide an
efficient and effective means to identify duplicate participation and
help prevent food stamp overpayments.  Rather then relying on states
to individually develop programs to identify and prevent duplicate
participation, it would be more efficient for USDA to develop a
single nationwide system.  Once established, such a system could be
expanded to track additional information that would help states
enforce welfare reform provisions relating to food stamps, such as
recipients' work history and their length of time in the program. 


--------------------
\3 We did not conduct a field investigation to determine the specific
circumstances of each match and therefore cannot say with 100 percent
certainty that duplicate participation occurred.  However, for each
duplicate participant, the Social Security number, name, and date of
birth matched between states in their computerized benefit issuance
records for the same month(s). 

\4 Report to Congress on Data Processing and Case Tracking in the
Temporary Assistance for Needy Families Program, Department of Health
and Human Services, Dec.  1997. 


   BACKGROUND
------------------------------------------------------------ Letter :2

The Food Stamp Program is designed to promote the general welfare and
to safeguard the health and well-being of the nation's population by
raising the nutritional levels of low-income families.  Recipients
use their food stamp benefits to purchase allowable food products
from authorized retail food merchants.  USDA's Food and Nutrition
Service (FNS) manages the Food Stamp Program through agreements with
state agencies.  FNS approves the states' plans to operate the
program and ensures that states administer the program in accordance
with regulations.  The federal government pays all of the costs for
benefits and one-half of the administrative costs for each state.  In
fiscal year 1997, USDA provided over $21 billion in benefits to about
23 million participants and paid the states, U.S.  territories, and
the District of Columbia about $1.8 billion to administer the
program. 

Food stamps are issued to households, which can be an individual, a
family, or another group that lives together and customarily
purchases and prepares food in common.  Households applying for
benefits must provide a Social Security number for each member.  The
value of the food stamp benefits for a household is determined by the
number of eligible household members and their income, adjusted for
assets and such costs as shelter and utilities.  Therefore, a
household's monthly food stamp allotment increases with each
additional member, provided income limits are not exceeded.  The
average monthly benefit per household member in 1996, which varied by
state, was about $73 nationally, but the benefit could have been as
much as $122 for a single-person household. 

Eligibility workers in service centers work directly with applicants
or their representatives to certify household eligibility and
determine the amount of benefits at the time of the application and
at least annually thereafter.  Households that receive food stamps
are required to report changes in household membership, such as the
loss or the addition of a member, to the administering state or local
agency. 

Generally, the service centers maintain the clients' detailed case
records while the computerized eligibility and benefit issuance data
are maintained at the state or the local level.  State agencies in
Florida, Texas, and New York maintain both types of data, while
California's state agency maintains the information on eligibility
and its counties maintain the information on benefit issuance. 

Each state is required to establish a performance reporting system to
monitor its food stamp program, including a quality control review
process to help ensure that benefits are issued only to qualifying
households and that the amounts of these benefits are correct. 
Federal regulations (7 C.F.R.  Ch.  II, part 273.3) specify that no
individual may participate as a member in more than one food stamp
household or in more than one area in any month.\5 This regulation
also applies to participation in more than one state.  State agencies
are responsible for imposing penalties for violations of program
requirements, such as the receipt of multiple food stamp benefits,
and for recovering food stamp overpayments.\6


--------------------
\5 The only exception is an individual who is a resident of a shelter
for battered women and children and was a member of a household
containing the person who had abused him or her.  While some of the
duplicate participants we identified may have been in these
circumstances, no such cases were identified in our review of sample
files in each state. 

\6 The Personal Responsibility and Work Opportunity Reconciliation
Act of 1996, P.L.  104-193, doubled the disqualification period for
intentional program violations from 6 months to 1 year for the first
violation and from 1 year to 2 years for the second violation. 
Furthermore, individuals that fraudulently misrepresent their
identity or residence to receive multiple food stamp benefits are
disqualified for 10 years. 


   DUPLICATE PARTICIPATION MAY
   COST MILLIONS IN OVERPAYMENTS
------------------------------------------------------------ Letter :3

In the four states we examined, about $3.9 million in food stamp
benefits were provided during calendar year 1996 to households in
different states that may have simultaneously included the same
individuals as members.  Table 1 summarizes the number of potential
duplicate participants we identified in the six possible pairings of
the four states we reviewed.  The California and Texas pairing
produced the largest number of duplicate participants, while the New
York and Texas pairing produced the fewest.  In total, we identified
about 20,000 duplicate participants. 



                                Table 1
                
                     Extent of Potential Duplicate
                   Participation, Calendar Year 1996

                                            Total months     Estimated
                                 Number of  of duplicate  overpayments
                                 duplicate  participatio           (in
States matched                participants             n    thousands)
----------------------------  ------------  ------------  ------------
California and Texas                 7,744        16,022        $1,145
Florida and New York                 6,065        20,991         1,682
California and Florida               2,521         4,965           364
Florida and Texas                    2,016         4,026           316
California and New York                885         3,248           242
New York and Texas                     813         2,371           186
======================================================================
Total                               20,044        51,623        $3,935
----------------------------------------------------------------------
Sources:  For California, the California Department of Public Social
Services; for Florida, the Florida Department of Children and
Families; for New York, the New York State Department of Social
Services; and for Texas, the Texas Department of Health Services. 

The individuals we identified in our match were members of households
of varying sizes, some with multiple members and some with a single
memberï¿½the individual was the household.  Households with multiple
members may have (1) fraudulently listed an individual as a member
when applying or (2) neglected to report that the individual was no
longer in the household, resulting in the continued issuance of
benefits on the presumption that the individual was still present in
the home.  For single-member households, the individual or the
individual's authorized representative (or someone posing as the
individual or representative),\7 may have applied for and obtained
food stamp benefits in one or more states.  For our match, we did not
determine whether the duplicate participation was the result of
inadvertent or intentional actions by the household, nor did we
determine which state each individual actually resided in during the
months of duplicate participation. 

We found that duplicate participation was generally concentrated in a
relatively small number of counties in each state.  (See app.  II.)
Such information may help states determine where to focus their
efforts, such as intensifying the screening of applicants, to reduce
duplicate participation.  While counties with the most duplicate
participation were generally also those with the greatest number of
program participants in each state, we did find some exceptions.  For
example, New York City, which has about 65 percent of the state's
food stamp participants, had the fifth highest number of duplicate
participants, while Monroe County, New York, which has only about 4
percent of the state's food stamp participants, had the most
duplicate participants. 

While we focused our analysis on duplicate participation in the Food
Stamp Program, such participation may be indicative of similar
problems in other public assistance programs administered by the
states.  In our review of the sample case files in each of the four
states we visited, we found that some duplicate food stamp
participants were also participants in other benefit programs in
those states, including Aid to Families With Dependent Children (now
Temporary Assistance to Needy Families (TANF) and general assistance. 
Similarly, FNS reported in its 1996 study of the characteristics of
households receiving food stamps that 52 percent of the recipients
also received family aid and general assistance benefits.\8

The possibility of duplicate participation in multiple benefit
programs is consistent with a September 1997 analysis by the
Department of Health and Human Services (HHS).  In a one-time
interstate computer match of information provided by 15 states and
the District of Columbia, HHS identified 18,000 potential cases of
duplicate participation in various public assistance programs (the
match was based on participation information that was voluntarily
provided by state agencies for the single month of August 1997).  HHS
compared the Social Security numbers, surnames, and dates of birth of
the participants in each of the 15 participating states and the
District of Columbia against those of the other participating states. 
All cases in which at least the Social Security numbers matched
between states were considered to potentially indicate duplicate
participation.  HHS referred all such cases to the states for
follow-up and for each client listed the programs that provided
benefits, such as TANF, food stamps, general assistance, Medicaid,
and Supplemental Security Income (SSI), and indicated when the client
had participated in each program.  Because the states had not
reported the results of their follow-up efforts as of May, 1998, HHS
could not provide information on the actual extent of duplicate
participation identified by that computer match. 


--------------------
\7 All food stamp applicants have the right when they apply to
specify an authorized representative to act on their behalf,
including receiving their food stamp benefits. 

\8 Characteristics of Food Stamp Households:  Fiscal Year 1996
(Advance Report), USDA, Food and Consumer Service (now known as Food
and Nutrition Service), Office of Analysis and Evaluation, Oct. 
1997. 


   LACK OF COMPREHENSIVE
   INFORMATION ALLOWS DUPLICATE
   PARTICIPATION
------------------------------------------------------------ Letter :4

Individuals included in food stamp households in more than one state
escape detection because there is no comprehensive national system to
identify food stamp participation in more than one state.  Each state
is responsible for establishing a system to ensure that no individual
participates in more than one household simultaneously within its
borders.  The Congress, recognizing that the 1996 welfare reform
legislation, which contains work requirements for the Food Stamp
Program and time limits for TANF, would necessitate tracking
participation across states, asked HHS to study how these
requirements would be implemented.\9 In December 1997, HHS reported
to the Congress that most states had not begun planning how to track
participation from one state to another.  Furthermore, no federal
agency has overall responsibility for creating a national database or
information system to facilitate such tracking. 

In the absence of such a system, states must rely primarily on
applicants to truthfully report the individuals who are members of
their households, not include individuals physically residing in
another state or household, and notify them of any subsequent
changes.  SSA provides some information to the states under certain
circumstances when recipients of SSA's Retirement, Survivor and
Disability Insurance (RSDI) or SSI move to a different state;
however, few food stamp participants receive RSDI or SSI.\10 In
addition, 22 states conduct duplicate participation matches with
other states; however, these matches are not comprehensive, generally
including food stamp and other public assistance benefit programs of
only neighboring states. 


--------------------
\9 The Personal Responsibility and Work Opportunity Reconciliation
Act of 1996, P.L.  104-193. 

\10 RSDI is an insurance program, principally funded out of dedicated
employment taxes, that pays monthly benefits to retired and disabled
workers, their dependents, and survivors to replace income that is
lost to a family through the retirement, the death, or the disability
of a worker who has earned protection against these risks.  SSI
provides a minimum cash income to all aged, blind, or disabled
individuals with limited resources. 


      NO NATIONAL DATABASE OF FOOD
      STAMP PARTICIPANTS EXISTS
---------------------------------------------------------- Letter :4.1

There is no national database or information system from which states
can obtain comprehensive information on participation in public
assistance programs, including the Food Stamp Program.  Although
certain provisions of the welfare reform legislation of 1996 cannot
be fully enforced without interstate tracking of participation in
public assistance programs, most states have not begun planning how
to track and thus prevent duplicate participation, and no federal
agency has overall responsibility for creating a national database or
information system to facilitate such tracking. 

The Personal Responsibility and Work Opportunity Reconciliation Act
of 1996 (hereafter referred to as the Welfare Reform Act), among
other things, places a lifetime limit on TANF benefits, whether
provided in one or multiple states, that an individual may
receive.\11 To implement this provision, states must track
information on clients receiving these benefits, not only within
their own jurisdictions, but across state boundaries as well. 
Recognizing the potential impact of the legislation on information
systems and capabilities within and among the states, the Congress
directed HHS to prepare a report that would address (1) the status of
automated data-processing systems operated by the states to assist in
welfare administration and (2) the requirements to establish a system
to track participants in public programs over time and check state
records to determine whether individuals are participating in welfare
programs in two or more states. 

In December 1997, HHS reported to the Congress that significant
modifications to the states' automated systems were needed to respond
to the requirements of the Welfare Reform Act.  Furthermore, a
November 1996 HHS survey of all the states; the territories of Guam,
Puerto Rico, and the Virgin Islands; and the District of Columbia
found that little progress had been made to meet those requirements. 
Specifically, 78 percent of the survey respondents stated that no
progress had been made in developing an automated system to track and
help prevent duplicate payments of benefits among the states, 11
percent indicated that they were in the planning stage, and the
remainder indicated that they were in a more advanced (developmental
or operational) stage.  Several respondents also recommended
establishing a national database to facilitate interstate tracking. 

While HHS' report identified five potential systems that could be
developed for interstate tracking, the report did not recommend
developing any specific system.  Instead, the report offered several
approaches that the Congress could consider should it decide to
pursue developing such a system.  When a system might be implemented
is uncertain, because HHS' Director of State System Approvals told us
that the Department does not intend to independently implement a
national interstate tracking system for public assistance programs
without guidance and funding from the Congress.  Furthermore, as HHS
reported, no other agency, public or private, federal or state, has
been designated to pursue this goal. 


--------------------
\11 In addition to changing the requirements for the Food Stamp
Program, the Welfare Reform Act replaced Aid to Families With
Dependent Children with TANF, a program of limited-term financial
assistance, which is administered by HHS. 


      STATES RELY HEAVILY ON
      CLIENTS TO PROVIDE ACCURATE
      INFORMATION
---------------------------------------------------------- Letter :4.2

Most public assistance agencies rely primarily on food stamp
applicants to truthfully identify household members, not include
individuals that physically reside in another state or household, and
report subsequent changes, such as the change of residence of a
household member.  Controls over determining household composition
are not as rigorous as they could be because the Food Stamp Program
must balance the issues of the clients' convenience, administrative
simplicity, and payment accuracy.  A household that wishes to receive
benefits must present an application listing all the members of that
household and their Social Security numbers and provide information
about their income and other eligibility factors.  The applications
include questions designed to determine whether the applicant is
receiving benefits elsewhere, for example, another county or
state.\12 Eligibility workers review this information, interview a
household representative, obtain identification documents (such as a
driver's license from the representative), certify the eligibility of
household members, and determine the amount of benefits.  In
addition, they recertify the household at least annually.  At no time
are all household members required to appear and present
identification; however, clients are responsible for identifying
changes in household membership.\13

According to FNS, which identifies overall error rates for each state
by reviewing a random sample of cases, client errors or
misrepresentations contribute significantly to incorrectly issued
benefits, particularly when an overpayment occurs.  In 1996, FNS
reported that about 7 percent or $1.5 billion of the $22 billion of
benefits issued nationwide were overpayments and that 57 percent of
overpayments were attributable to intentional or unintentional
inaccuracies in client-provided information.  (Errors also resulted
in underpayments of about $518 million in fiscal year 1996.)

Nevertheless, FNS' regulations do not require verification of
client-provided information on household membership, unless the
caseworker deems the information "questionable." The regulations
allow each state agency to develop guidance for identifying
questionable information.  In the four states we visited, the
guidance defined questionable information as applicants' statements
that were contradictory or did not agree with information that was in
the case record or otherwise available to the eligibility workers. 
Consistent with its practice of allowing states flexibility in
administering the program, FNS does not require the states to match
client information such as Social Security numbers with those of
clients on the food stamp rolls in other states. 

When the eligibility workers in the states we visited suspected
questionable or fraudulent information, they could refer the
application to investigators before granting aid.  Investigators in
each state told us that they try to verify questionable information
on household composition by visiting homes and making collateral
contacts to confirm information with friends, neighbors, or
landlords.  According to the investigators, these techniques are
hit-or-miss, time-consuming, costly undertakings and provide
information that is only as reliable as its source.  Furthermore,
investigative resources are generally very limited. 


--------------------
\12 In addition, all but one of the states we visited had implemented
fingerprint imaging of certain clients to prevent duplicate public
assistance program participation within the state or specific
jurisdictions within the state. 

\13 Almost all households in California must submit monthly change
reports.  Households in Florida and Texas are required to report
changes in circumstances, including changes in household composition,
within 10 days of becoming aware of them.  Households with earned
income in New York must report changes quarterly; all others must
report changes within 10 days. 


      THE SOCIAL SECURITY SYSTEM
      PROVIDES LIMITED INFORMATION
      ON INTERSTATE MOVEMENT
---------------------------------------------------------- Letter :4.3

Currently, SSA is the only central national source to inform states
when an applicant or client may be participating in the Food Stamp
Program (or any other public assistance program) in another state. 
SSA verifies the Social Security numbers of applicants to public
assistance programs, including the Food Stamp Program, and notifies
the states if applicants are receiving SSA benefits so their incomes
can be verified.  States submit information about the applicants
(specifically, name, date of birth, and Social Security number) and
receive from SSA notification of verification (or nonverification) of
the Social Security number and the current status of SSA payments
associated with that number. 

After initial verification, SSA automatically notifies the state of
subsequent changes in the status of RSDI and SSI beneficiaries,
including a change in state of residence.  If an RSDI beneficiary
applies for public assistance programs in a new state of residence
and that state requests verification from SSA, the system will
automatically (1) notify the state that submitted the Social Security
number for verification that the number has previously been submitted
by another state and (2) notify the state that originally submitted
the number that another state has submitted the same number. 
However, this notification is provided only if the Social Security
number belongs to an individual already receiving RSDI payments. 
According to a report by FNS on the characteristics of fiscal year
1996 food stamp households, only about 13 percent of the members of
households receiving food stamps were also receiving RSDI benefits. 

When an SSI beneficiary moves from one state to another, SSA
automatically notifies the former state of residence and the new
state of residence.  However, the state that the beneficiary left
does not automatically receive the new address nor does the new state
of residence automatically receive the previous address.  As with
RSDI, SSA provides information only on individuals receiving SSI
payments.  According to FNS' report on the characteristics of
households receiving food stamps during fiscal year 1996, only about
19 percent of the members of those households were SSI recipients. 


      STATES' MATCHING EFFORTS,
      WHILE USEFUL, RELY ON
      LIMITED DATA
---------------------------------------------------------- Letter :4.4

In the absence of a comprehensive national database or information
system, some state agencies have successfully employed their own
computer matches to identify applicants or clients who are included
in food stamp households in other states; however, the effectiveness
of these matches is limited by the data they have access to. 
According to FNS, 22 states currently perform a routine computer
match between their food stamp rolls and the food stamp rolls of at
least one other state.  All of the states we visited--California,
Florida, New York, and Texas--have established a matching program
between the state's public assistance rolls (for all assistance
programs, not just food stamps) and the public assistance rolls of
selected states, usually those on their borders.  Of those states,
New York conducts the most extensive matching, comparing information
with five border states, as well as with Florida, Virginia, and
Puerto Rico. 

Texas' public assistance duplicate participation match is
characteristic of the neighboring-state matches conducted in the
states we visited.  Texas conducts a quarterly match for all the
public assistance programs it administers, including food stamps,
using computer tapes provided by the bordering states of Oklahoma,
Louisiana, and New Mexico, and is in the process of establishing a
match with Arkansas.  Texas sends a tape of participants to each of
the other states on the same quarterly schedule so that they can
conduct their own matches.  Texas established the data exchange for
each match through a separate memorandum of understanding negotiated
with each state.  Aside from some initial system compatibility and
data reliability problems, Texas officials have found the matches to
be simple, low-cost routines to maintain and are satisfied with the
results.  A Texas state official estimated that computer matches of
Texas and Oklahoma public assistance programs for the first two
quarters of fiscal year 1998 identified 715 potential duplicate
participants. 

While computer matching to identify duplicate participants in
neighboring states is useful, the effectiveness of such matches is
limited to those states that agree to share data.  States generally
establish cooperative agreements to exchange data only with border
states, because (1) they are the most accessible to clients and
therefore considered the most likely place for clients to be included
in another household and (2) establishing cooperative agreements
requires significant administrative effort.\14

While many states are able to establish data exchange agreements with
border states, state officials told us of several cases in which
states declined to participate in a duplicate participation matching
effort because they had higher-priority issues to work on. 

Furthermore, our findings of duplicate participation and HHS' similar
findings involving remote states show that matching between
neighboring states goes only part of the way towards identifying
improper payments.  For example, HHS' analysis identified more than
7,000 potential cases of participants in public assistance programs
in California who also participated in public assistance programs in
at least 1 of the 14 other states examined and the District of
Columbia.  None of the other states shares a border with California. 


--------------------
\14 To protect the confidentiality of information obtained from food
stamp applicants or recipient households, FNS' regulations (7 C.F.R. 
272.8 (a)(4)) require state agencies to enter into data exchange
agreements before exchanging food stamp data.  These agreements limit
the disclosure of such information to persons directly connected with
the administration or the enforcement of the Food Stamp Program and
other federal assistance programs (7 C.F.R.  272.1 (c)(ii)). 


   A NATIONAL DATABASE OF FOOD
   STAMP PROGRAM PARTICIPANTS
   WOULD MORE EFFECTIVELY PREVENT
   DUPLICATE PARTICIPATION
------------------------------------------------------------ Letter :5

Instead of establishing 50 state-matching programs, it would be more
efficient and practical to establish one comprehensive national
database of information on food stamp participation that could help
states determine if an applicant or recipient is already receiving
benefits in another state.  Until a national system to track all
participation in public assistance programs is established to assist
states in enforcing welfare reform provisions, a national database of
food stamp participation information could be used to detect and
prevent duplicate participation at the time of application, thereby
obviating the process of recouping overpayments. 

FNS already maintains a national database of disqualified food stamp
recipients to help states keep individuals who have previously
violated program restrictions from reenrolling.  Experience with
using this system could provide FNS with information to use in
developing a system that covers all food stamp recipients. 


      NATIONAL INFORMATION ON
      PARTICIPATION COULD HELP
      PREVENT OVERPAYMENTS
---------------------------------------------------------- Letter :5.1

Recent statements of the FNS Administrator emphasize the agency's
support for activities that improve the integrity of the Food Stamp
Program and that help preclude making overpayments.  In April 1998
testimony, the Administrator stated that FNS' focus is on
prevention--barring recipients who would abuse the program from
getting an opportunity to do so.\15 The Administrator also stated
that critical to the success of FNS' activities to combat fraud is
its interaction and exchange of information with states.  The
Administrator also noted that GAO's reports, most recently focusing
on using computer matching to identify imprisoned and deceased
individuals improperly included in Food Stamp Program households,
have identified cost-effective ways for states to use information
systems to improve the program's integrity. 

Consistent with the Administrator's comments, the Office of
Management and Budget's (OMB) report in January 1997 pointed out the
savings government benefit programs could achieve through improved
access to information.\16 According to the report, timely and
intelligent data sharing could significantly improve the ability of
federal and state agencies to make more accurate and faster initial
and ongoing eligibility determinations.  The report also found the
following: 

ï¿½Improved mechanisms to provide accurate information to eligibility
workers would help prevent overpayments and improve service to
clients. 

ï¿½Data sharing at the time of application could change enforcement
efforts from a "pay-and-chase" mode to one that would be more
proactive and efficient. 

ï¿½Such a proactive system could increase public confidence in the
administration of benefit programs. 

However, without a national database of food stamp participants,
preventing duplicate participation is difficult.  Usually,
eligibility has been certified and benefits have been issued before a
state discovers duplicate participation.  As a result, the state is
forced into the "pay-and-chase" mode to try to recoup overpayments. 


--------------------
\15 Testimony of Yvette S.  Jackson, Administrator, FNS, USDA, Before
the Senate Committee on Agriculture, Nutrition, and Forestry, Apr. 
23, 1998. 

\16 Strategies for Efficiency:  Improving the Coordination of
Government Information Resources was issued by the Benefit Systems
Review Team, an interagency effort to find ways to enhance the
integrity of federal and state benefit programs and reduce fraud and
administrative costs through better coordination and use of
information systems. 


      FNS HAS EXPERIENCE MANAGING
      A NATIONAL DATABASE OF
      DISQUALIFIED FOOD STAMP
      RECIPIENTS
---------------------------------------------------------- Letter :5.2

While FNS does not currently collect participation information
nationwide, it does make the records from USDA's Disqualified
Recipient Subsystem--a database that compiles the records of
individuals disqualified from participation for intentionally
violating the rules of the Food Stamp Program--available to help
states identify these individuals across state borders.  To determine
the correct penalty to assign violators, states must check this
database for their history.  The lessons learned in the development
and operation of this information system could be useful to FNS in
developing a system that covers all food stamp recipients. 

Creating a system to identify duplicate participation is most
appropriately within FNS' realm of responsibility, and, given the
potential scope of this problem that we and others have identified,
would go far to help ensure the integrity of its Food Stamp Program. 
Developing such a system would require FNS to develop the systems
software and to coordinate the flow of information to and from the
states, as well as to provide oversight to help ensure that the
states take the appropriate actions.\17 With regard to state
follow-up costs, a 1995 FNS study of the cost of states' computer
matching with government agency databases concluded that
data-processing costs for the comparison were approximately 2 cents
per case, and investigative follow-up and claim collection costs were
about $5 to $7 for all matches, whether or not they were
overpayments. 

While initially serving to identify duplicate participation, a
national database or information system on food stamp participation
could be expanded to also help the states track the requirements for
food stamp eligibility under the Welfare Reform Act (e.g., the
limitation on benefits to 3 months in a 36-month period for
able-bodied adults who do not work).  Furthermore, by connecting it
to other databases, such as those containing records on prisoners or
deceased persons, this national information system could evolve into
a comprehensive database to help state agencies verify eligibility
for food stamps. 

According to FNS officials, other than an ongoing USDA Office of
Inspector General review of six states,\18 the Department has not
conducted any reviews to determine if interstate duplicate
participation has occurred.  In the absence of evidence of the need
for a national database of food stamp participation information, USDA
has not considered creating such a system.  In our discussions with
FNS regarding how it would help prevent interstate duplicate
participation by food stamp beneficiaries, FNS was reluctant to
discuss any possible solutions, including the potential benefits or
drawbacks of creating a national database, until FNS has conducted a
detailed analysis of the issue. 


--------------------
\17 Such a system would also have to protect the confidentiality of
information obtained from food stamp applicant and recipient
households. 

\18 USDA's Office of Inspector General is currently reviewing
duplicate participation among six neighboring midwestern states; the
results of this review are not yet available. 


   CONCLUSIONS
------------------------------------------------------------ Letter :6

The duplicate participation by members of food stamp households in
more than one state undermines the credibility of the Food Stamp
Program and results in millions of dollars of overpayments. 
Conventional methods that have been used by state agencies to detect
such individuals have primarily focused on neighboring states and
thus have not been fully effective.  While the need for a national
information system to track recipients in all public assistance
programs, including food stamp recipients, among the states has been
recognized, the timing for implementing such a system is unknown. 

Given our findings of potential duplicate participation in four
widely separated states, as well as similar findings by HHS, we
believe that creating a national system to collect, analyze, and
disseminate information on participation in the Food Stamp Program
would be an effective way to provide the states with the information
they need to help prevent duplicate participation.  Such a system
should allow the states to identify duplicate participation in their
current rolls and provide a means for them to check on the
participation status of applicants, thereby preventing future
duplicate participation and overpayments.  Furthermore, this system
could also be expanded to help states better enforce the 1996 welfare
reform requirements for food stamps as well as identify ineligible
participants, such as prisoners and deceased individuals. 


   RECOMMENDATION TO THE SECRETARY
   OF AGRICULTURE
------------------------------------------------------------ Letter :7

In the absence of a comprehensive national information system on
participants in all public assistance programs, we recommend that the
Secretary of Agriculture direct the Administrator of FNS to consider
establishing a central system to help ensure that individuals
participating in the Food Stamp Program are not being improperly
included as household members in more than one state concurrently. 
As part of this effort, FNS should conduct a feasibility study to
identify options and provide a cost-benefit estimate for each option. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :8

We provided a copy of a draft of this report to the U.S.  Department
of Agriculture for review and comment.  Its comments and our
responses are in appendix IV. 

In commenting on the draft report, Department officials stated that
our work provided a foundation for further action in preventing
duplicate participation in the Food Stamp Program.  They noted that
despite the low rate of duplicate participation in relation to total
participation identified by our analysis, duplicate participation is
a violation of law and program regulations that must be detected and
prevented.  They agreed that exchanging information through computer
matching holds great promise, but noted that information systems that
will be used for computer matching must be implemented thoughtfully. 
FNS officials said they intend to conduct a feasibility study that
would identify design options and provide a cost-benefit estimate for
each option.  We agree with this approach and have revised our
recommendation to explicitly recognize the need for a feasibility
study. 

We also provided excerpts from the draft report to California,
Florida, New York, and Texas state officials for their review and
comment.  California officials noted that there were relatively small
numbers of duplicate participants in comparison with the total number
of participants in the Food Stamp Program.  Florida officials
commented that our findings demonstrated the need for a national
database to provide program participation information to states. 
Texas officials said that follow-up investigations of the match
results would be needed to determine the extent to which they
actually represented overpayments.  New York state officials
questioned the number and the validity of the matches we identified
between New York and Florida based on the results of their own
matching efforts.  Specifically, they identified fewer matches--fewer
potential duplicate participants--annually.  Furthermore, they
determined that many of these initial matches were invalid upon
additional investigation.  While we did not conduct a detailed
evaluation of New York's matching process, we are aware of
methodological differences, including our use of more stringent
matching criteria, which could account for differences in results. 
Because of our knowledge of New York's methodology, we continue to
believe that our work provides a valid picture of duplicate
participation.  See appendix III for our methodology. 


---------------------------------------------------------- Letter :8.1

We conducted our work from February 1998 through July 1998 in
accordance with generally accepted government auditing standards. 
Our detailed methodology is presented in appendix III. 

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report for 30 days. 
At that time, we will send copies of this report to appropriate
congressional committees, Members of Congress, and other interested
parties.  We will also make copies available to others on request. 

Major contributors to this report are listed in appendix V.  If you
have any questions about this report, please contact me at (202)
512-9692. 

Sincerely yours,

Lawrence J.  Dyckman
Director, Food and
 Agriculture Issues


INFORMATION ON THE FOOD STAMP
PROGRAM'S COSTS, PARTICIPANTS, AND
BENEFITS IN FOUR STATES
=========================================================== Appendix I

In fiscal year 1996, California, Florida, New York, and Texas
represented almost 36 percent of the cost of benefits in the Food
Stamp Program and approximately 35 percent of the nation's
participants. 



                               Table I.1
                
                The Food Stamp Program's Benefit Costs,
                   Participants, and Average Monthly
                Benefit Per Participant by State, Fiscal
                               Year 1996

                  (Costs and participants in millions)

                                                               Average
                                                               monthly
                                  Costs of         State   benefit per
State                          food stamps  participants   participant
----------------------------  ------------  ------------  ------------
California                          $2,555           3.1        $67.73
Florida                              1,296           1.4         78.72
New York                             2,054           2.1         81.56
Texas                                2,140           2.4         75.18
======================================================================
Total                               $8,045           9.0
Program total                      $22,441          25.5        $73.24
----------------------------------------------------------------------
Source:  U.S.  Department of Agriculture's Food and Nutrition
Service. 


INFORMATION ON COUNTY OF RESIDENCE
OF DUPLICATE PARTICIPANTS
========================================================== Appendix II

Identifying patterns of duplicate participation in the Food Stamp
Program and other benefit programs could help public assistance
agencies better target their resources to prevent fraud and abuse in
the programs they administer.  For example, we identified the
counties in four states--California, Florida, New York, and Texas--
where interstate duplicate participation occurred most frequently
among them.  In these four states, we found that the 10 most frequent
counties of residence for duplicate participation accounted for most
of such participants in each state --75 percent in California, 61
percent in Florida, 60 percent in New York, and 63 percent in Texas. 
While counties with the most duplicate participation were generally
also those with the greatest number of program participants in their
states, we also found some divergence.  For example, New York City,
which accounted for 65 percent of program participation in the state,
was fifth for duplicate participation, while Monroe County, New York,
which accounted for 4 percent of program participation was first for
duplicate participation. 



                               Table II.1
                
                Most Frequent Counties of Residence for
                 Duplicate Participants in Four States,
                  Ranked From Most to Fewest Duplicate
                    Participants, Calendar Year 1996

California                    Florida       New York      Texas
----------------------------  ------------  ------------  ------------
Los Angeles                   Dade          Monroe        El Paso

San Diego                     Hillsborough  Suffolk       Harris

San Bernardino                Orange        Erie          Dallas

Sacramento                    Broward       Westchester   Bexar

Riverside                     Pinellas      New York      Hidalgo
                                            City

Kern                          Polk          Onondaga      Cameron

Orange                        Palm Beach    Nassau        Tarrant

Fresno                        Duval         Orange        Travis

Alameda                       Volusia       Oneida        Bell

Santa Clara                   Osceola       Chautauqua    Nueces
----------------------------------------------------------------------
Sources:  For California, the California Department of Public Social
Services; for Florida, the Florida Department of Children and
Families; for New York, the New York State Department of Social
Services; and for Texas, the Texas Department of Health Services. 


OBJECTIVES, SCOPE, AND METHODOLOGY
========================================================= Appendix III

To determine how many individuals were included as members of more
than one household that received food stamp benefits during the same
time period and the estimated value of the benefits that were issued
to those households, we matched the food stamp records of each of the
four states with the largest benefit issuance in the Food Stamp
Program against each other.  We used the states' data as follows: 

  -- State welfare agencies in Florida, New York, and Texas provided
     us with computer files containing information on all members of
     households and the amount of food stamp benefits issued to those
     households during calendar year 1996.  The data provided
     personal identifiers, including name, Social Security number,
     date of birth, gender, and the months in which food stamp
     benefits had been issued to the household while each individual
     was a member.  The state agencies had verified the Social
     Security numbers for the data on food stamp beneficiaries
     through SSA's Enumeration Verification System.  Texas state
     officials provided electronic benefit transfer (EBT) data for a
     selected sample of 75 cases.\1

  -- In California, where issuance data is maintained at the county
     level, we determined that eligibility was predictive of
     participation in two counties, so we used the state's
     eligibility information in lieu of issuance data for our
     match.\2

We matched the verified Social Security numbers of members of food
stamp households in each state with those of members of food stamp
households in each of the other three states.  For each individual
identified as a member of households in more than one state, we
determined the periods during which food stamps were issued to both
(or, in a few cases, as many as three) households for that
individual.  For double participation, we estimated the dollar value
of the improperly issued food stamps by applying the average of the
two states' average monthly issuance per individual recipient from
fiscal year 1996 to each period in which issuance occurred
concurrently in the states.  (See table III.1.) Triple participants
were treated as double participants in each set of states where we
found them.  Our findings cannot be used to estimate potential
overpayments nationwide, because our methodology was not designed for
that purpose. 



                              Table III.1
                
                Average Monthly Benefit for State Pairs,
                            Fiscal Year 1996

                        California     Florida    New York       Texas
----------------------  ----------  ----------  ----------  ----------
California                $67.73\a      $73.23      $74.65      $71.46
Florida                      73.23     78.72\a       80.14       76.95
New York                     74.65       80.14     81.56\a       78.37
Texas                        71.46       76.95       78.37     75.18\a
----------------------------------------------------------------------
\a Single-state average. 

Source:  GAO's analysis of U.S.  Department of Agriculture's Food and
Nutrition Service data. 

Food stamp benefits are calculated for households, not for
individuals.  As such, it is difficult to determine the exact value
of benefits issued to an individual included in a household, unless
he or she is the only member of a household.  Even then, the amount
will vary from individual to individual, on the basis of such factors
as income, assets, and the cost of shelter.  Therefore, we relied on
the average monthly benefit issued per person in the locations we
reviewed, which ranged from a high of $82 in New York to a low of $68
in California.  We realize that the actual issuance may be higher or
lower than our estimates; for example, the maximum issuance to
single-member households who are included in our findings was $119 in
fiscal year 1996.  Our estimates are intended to show the general
magnitude of the problem. 

We considered every month of duplicate issuance in more than one
state to be an overpayment, in keeping with food stamp regulations (7
C.F.R.  Ch.  II, part 273.3), which specify that no individual may
participate as a member in more than one household or in more than
one area in any month.  However, our estimates of overpayments were
conservative in that we assumed that the individual was actually
eligible to participate in one of the states. 

Because of the quality control program operated by the Food and
Nutrition Service and the states' ongoing quality assurance efforts,
we accepted their computerized food stamp data as reliable.  To
provide additional confidence in the data's accuracy, we reviewed a
limited number of food stamp case files at social service centers in
four large metropolitan areas--Los Angeles County, California; Dade
County, Florida; El Paso County, Texas; and New York City.  We
compared the Social Security number and the date of birth of each
individual in the computerized database with information in the
relevant case files and found no significant differences.  To further
verify participation in the household, we reviewed copies of
available documents in the case files, such as birth certificates,
drivers licenses, and Social Security cards.  We attempted to obtain
additional verification of the periods of participation and identify
participation in other public assistance programs by reviewing
approved applications; however, many case files did not contain
complete information.  To evaluate the extent to which individuals
may be exempt from the prohibition on duplicate participation, we
reviewed applications and other relevant documentation to determine
if the individuals were residents of shelters for battered women and
children. 

For a sample of 75 Texas cases, we compared the computerized food
stamp months of participation with EBT data showing the months that
the benefit accounts were accessed to determine whether program
participation had actually occurred, i.e., benefits had in fact been
drawn out.  The resulting analysis showed that benefits had been
drawn from the EBT account within 3 months of being credited in 91
percent of the months for which Texas' computerized food stamp data
indicated participation.  This analysis attests to the accuracy of
Texas' computerized participation data and indicates a high
correlation between the availability of benefits and their use. 

To determine why the inclusion of an individual in a food stamp
household was not detected, we contacted state agency officials in
Sacramento, California; Tallahassee, Florida; Albany, New York; and
Austin, Texas, to discuss and review policy and procedures for
verifying an applicant's data and any subsequent changes, including
determining whether an applicant or client is a household member in
another state.  We discussed fraud detection and computerized
data-matching efforts, quality control and assurance efforts, and
methods of food stamp issuance with state officials.  In the four
large metropolitan areas we selected for review, we discussed local
fraud detection and computerized data-matching efforts with officials
at the state's social service agency. 

To identify the options for detecting or preventing future food stamp
overpayments caused by duplicate participation, we discussed with
agency officials in each of the states we visited their opinions
regarding the value of computer matching.  We contacted state
officials to determine the cost, the quality, the savings, and the
barriers to computer matching to identify duplicate participation. 
To determine the effort associated with data matching to identify
such individuals, we calculated the time used by our programmer to
develop and implement the match programs and reviewed studies
performed for FNS, OMB, and HHS regarding the costs and the
effectiveness of data-sharing and matching. 



(See figure in printed edition.)Appendix IV

--------------------
\1 Under EBT systems, the state agency issues access cards (similar
to debit cards) and personal identification numbers to clients, who
obtain benefits through point-of-sale terminals in stores. 

\2 In a match between the state eligibility information and the Los
Angeles County and Orange County, California, food stamp issuance
information, we found that more than 89 and 75 percent, respectively,
of the eligible individuals participated. 


COMMENTS FROM THE U.S.  DEPARTMENT
OF AGRICULTURE
========================================================= Appendix III



(See figure in printed edition.)


The following are GAO's comments on the U.S.  Department of
Agriculture's letter dated July 8, 1998. 

GAO COMMENTS

1.  We have revised our report to delete any implication that the
Department's Disqualified Recipient Subsystem might be modified to
encompass a system for identifying duplicate participation.  But we
continue to believe that the lessons learned by the Department in
developing and implementing the Disqualified Recipient Subsystem will
be useful in creating such a system. 

2.  We agree that the knowledge and views of the Department of Health
and Human Services regarding the development of a national database
to track participation in assistance programs across state lines
would be helpful to the Department in creating such a system for the
Food Stamp Program.  However, the Department should not delay its
efforts in anticipation of action by another agency.  A Department of
Health and Human Services official told GAO that it does not plan to
create a national system, nor is it responsible for doing so. 

3.  Because of privacy concerns, GAO does not usually release data on
individuals resulting from computer matches.  Most of GAO's computer
matches have involved federal or federal and state databases.  Under
the Computer Matching and Privacy Protection Act, executive branch
agencies must adhere to data integrity and privacy requirements
before such computer matches can take place.  Although not subject to
the act, GAO has denied requests for GAO computer matches because
providing the information would enable agencies to obtain data
without adhering to the kinds of requirements contained in the act. 
However, in the present case, under the federal food stamp
regulations, states can match food stamp recipient databases if there
is an agreement in effect between the states which, for privacy
concerns, limits their use of such information.  If any of the states
have matching agreements to share food stamp recipient information,
we would, in this instance, consider providing the computer matches
with a written undertaking that they would be subject to such
agreements.  This would ensure that confidentiality of such data
would be maintained. 


MAJOR CONTRIBUTORS TO THIS REPORT
=========================================================== Appendix V

Keith Oleson, Assistant Director
David A.  Moreno, Project Leader
Leo N.  Acosta
Brad L.  Dobbins
Donald Ficklin
J.  D.  Hall
Jonathan Silverman

OFFICE OF GENERAL COUNSEL

Alan R.  Kasdan
Oliver Easterwood


*** End of document. ***