Public Assistance: PARIS Project Can Help States Reduce Improper 
Benefit Payments (06-SEP-01, GAO-01-935).			 
								 
Public assistance programs make millions of dollars in improper  
payments every year. Some of these improper payments occur	 
because state and local agencies that administer the programs	 
lack adequate, timely information to determine recipients'	 
eligibility for assistance. This inability to share information  
can result in both federal and state tax dollars being needlessly
spent on benefits for the same individuals and families in more  
than one state. In 1997, staff at the Administration for Children
and Families (ACF) within the Department of Health and Human	 
Services (HHS) began a project to help states share eligibility  
information with one another. The public assistance reporting	 
information system (PARIS) interstate match helps states share	 
information on public assistance programs, such as Temporary	 
Assistance for Needy Families (TANF) and Food Stamps, to identify
individuals or families who may be receiving benefit payments in 
more than one state simultaneously. Officials in almost all of	 
the 16 states and the District of Columbia that participated in  
PARIS reported that the project has helped identify improper	 
TANF, Medicaid, or Food Stamp payments in more than one state.	 
Despite its successes, the project also has several limitations. 
First, the opportunity to detect improper duplicate payments is  
not as great as it could be because only one-third of the states 
participate in the project. Second, participating states do not  
have adequate protocols or guidelines to facilitate critical	 
interstate communication. As a result, some states have reported 
problems that compromise the effectiveness of the project, such  
as difficulty determining whether an individual identified in a  
match is actually receiving benefits in another state. Third,	 
state administrators for the TANF, Medicaid, and Food Stamp	 
programs have not always placed adequate priority on using PARIS 
matches to identify recipients who are residing in other states. 
As a result, individuals may continue to receive or have benefits
paid on their behalf in more than one state even after they were 
identified through the matching process. Finally, because the	 
PARIS match is only designed to identify people after they are	 
already on the rolls, it does not enable the states to prevent	 
improper payments from being made in the first place.		 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-01-935 					        
    ACCNO:   A01737						        
  TITLE:     Public Assistance: PARIS Project Can Help States Reduce  
Improper Benefit Payments					 
     DATE:   09/06/2001 
  SUBJECT:   Erroneous payments 				 
	     Information systems				 
	     Internal controls					 
	     Program abuses					 
	     Public assistance programs 			 
	     Food Stamp Program 				 
	     HHS Public Assistance Reporting			 
	     Information System 				 								 
	     HHS Temporary Assistance for Needy 		 
	     Families Program					                                                                 
	     Medicaid Program					 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** 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.                                         **
**                                                              **
******************************************************************
GAO-01-935
     
Report to the Ranking Minority Member, Committee on Governmental Affairs, U.
S. Senate

United States General Accounting Office

GAO

September 2001 PUBLIC ASSISTANCE PARIS Project Can Help States Reduce
Improper Benefit Payments

GAO- 01- 935

Page i GAO- 01- 935 PARIS Project Can Help States Letter 1

Results in Brief 3 Background 4 PARIS: A Potentially Cost- Beneficial Means
of Controlling

Improper Payments 7 Limitations of PARIS 15 Conclusions 22 Recommendations
for Executive Action 22 Agency Comments and Our Evaluation 23

Appendix I Data and Assumptions Used in Our Analysis 26

Appendix II Comments From the Department of Health and Human Services 31

Appendix III Comments From the Food and Nutrition Service 36

Tables

Table 1: Programs We Examined That Are Covered by PARIS 7 Table 2: Number of
Potential Improper Payments for Selected

States, February 2001 10 Table 3: Estimated Net Savings Given Varying Valid
Hit Rates 14 Table 4: Estimated Net Savings Given a Varying Number of Future

Benefit Payment Months Avoided 14 Table 5: Scenario With Assumptions Used in
Report 26 Table 6: Estimated Net Savings Given Varying Valid Hit Rates 29
Table 7: Estimated Net Savings Given a Varying Number of Future

Benefit Payment Months Avoided 30

Figure

Figure 1: States Participating in Recent PARIS Matches 9 Contents

Page ii GAO- 01- 935 PARIS Project Can Help States Abbreviations

ACF Administration for Children and Families CMS Centers for Medicare &
Medicaid Services DMDC Defense Manpower Data Center FNS Food and Nutrition
Service HHS Department of Health and Human Services MCO managed care
organization OMB Office of Management and Budget PARIS Public Assistance
Reporting Information System SSA Social Security Administration SSI
Supplemental Security Income SSN social security number TANF Temporary
Assistance for Needy Families USDA Department of Agriculture

Page 1 GAO- 01- 935 PARIS Project Can Help States

September 6, 2001 The Honorable Fred Thompson Ranking Minority Member
Committee on Governmental Affairs United States Senate

Dear Senator Thompson: Every year, public assistance programs make millions
of dollars in improper payments. 1 Some of these improper payments are made
because state and local agencies that administer the programs lack adequate,
timely information to determine recipients? eligibility for assistance. For
example, in a report prepared for the Senate Committee on Governmental
Affairs in September 2000, we noted that state agencies responsible for
administering Temporary Assistance for Needy Families (TANF) lacked a
comprehensive mechanism to share vital eligibility information with their
counterparts in other states. 2 This inability to share information can
result in both federal and state tax dollars being needlessly spent on
benefits for the same individuals and families in more than one state.

In 1997, staff at the Administration for Children and Families (ACF) within
the Department of Health and Human Services (HHS) initiated a project to
help states share eligibility information with one another. The Public
Assistance Reporting Information System interstate match (hereafter referred
to as PARIS) helps states share information on public assistance programs,
such as TANF and Food Stamps, to identify individuals or families who may be
receiving benefit payments in more than one state simultaneously. 3 PARIS
also allows states to share eligibility information

1 See Financial Management: Billions in Improper Payments Continue to
Require Attention (GAO- 01- 44, October 27, 2000) and Financial Management:
Increased Attention Needed to Prevent Billions in Improper Payments (GAO/
AIMD- 00- 10, Oct. 29, 1999). Improper payments include payments that should
not have been made or were made for incorrect amounts irrespective of
whether the agency had effective controls in place. Specifically, improper
payments would include inadvertent errors, such as duplicate payments and
calculation errors; payments for unsupported or inadquately supported
claims; payments for services not rendered or to ineligible beneficiaries;
and payments resulting from outright fraud and abuse.

2 See Benefit and Loan Programs: Improved Data Sharing Could Enhance Program
Integrity (GAO/ HEHS- 00- 119, Sept. 13, 2000). 3 PARIS also includes
information- sharing efforts between states and federal agencies, such as
the Department of Veterans Affairs.

United States General Accounting Office Washington, DC 20548

Page 2 GAO- 01- 935 PARIS Project Can Help States

about Medicaid, the federal- state health financing program for certain
lowincome individuals, which may become more important as states continue to
enroll more beneficiaries in managed care organizations (MCOs). In a
Medicaid capitated managed care environment, states make prospective monthly
payments to MCOs for each beneficiary enrolled to provide or arrange for all
needed health services. Thus, if a Medicaid beneficiary moved out of state
and the state did not register the change, it would continue to make the MCO
monthly payment until the discrepancy was discovered. 4

PARIS is a voluntary project in which the participating states sign a
uniform agreement that governs the interstate exchange of data. Recipient
lists for all participating states are matched (compared) with one another
quarterly at a central location, using individuals? social security numbers
(SSN). Each state subsequently receives a list of individuals who may be
receiving duplicate TANF, Medicaid, and Food Stamp benefits in another
state. All participating states are expected to verify whether individuals
identified in the match are eligible for benefits in their state and remove
them from the rolls if they are not eligible. Some states use their local
benefit office staff to follow up on the matches, while others use fraud
investigators or some combination of the two.

Because of its potential for identifying improper benefit payments, you
asked us to review the PARIS project to determine (1) how effective PARIS
has been in identifying improper payments and (2) the limitations of PARIS.
You also asked us to identify what, if any, improvements can be made to
PARIS.

To answer these questions, we interviewed and collected data from federal,
state, and local officials responsible for administering the TANF, Medicaid,
and Food Stamp programs. At the federal level, ACF administers TANF, and
HHS? Centers for Medicare & Medicaid Services (CMS) administers Medicaid,
while the Department of Agriculture?s (USDA) Food and Nutrition Service
(FNS) administers the Food Stamp program. 5 Our interviews included
officials from the 16 states and the District of Columbia that had
participated in PARIS at least once since August 2000.

4 Under a traditional fee- for- service environment, payments in two states
for the same beneficiary would occur if a beneficiary intentionally used
medical services in two different states.

5 Until June 2001, CMS was known as the Health Care Financing
Administration. Other federal agencies we contacted included the Office of
Management and Budget (OMB) and the Social Security Administration (SSA).

Page 3 GAO- 01- 935 PARIS Project Can Help States

We also conducted site visits to seven jurisdictions. 6 In addition, we
spoke with officials from seven states that were not participating in the
project at the time of our review and visited two jurisdictions in one of
these states. 7 Finally, we performed our own analysis using available data
from five states to identify key factors that can influence the extent to
which participation in PARIS can result in program savings. 8 Our analysis
is based on a limited number of states, and thus, the results are not
generalizable to other states. Nonetheless, the results may provide useful
insights to other states that are considering participating in PARIS.

Officials in almost all of the 16 states and the District of Columbia that
participated in PARIS reported that the project has helped identify improper
TANF, Medicaid, or Food Stamp payments in more than one state. However, few
states have collected data on the actual savings resulting from their
participation in PARIS. Four states and the District of Columbia documented
about $16 million in savings, attributed mostly to the prevention of future
improper payments. In particular, these states reported significant savings
in the Medicaid program. In addition, three of these states collected data
on costs as well as savings from their participation in PARIS, and all three
reported that their savings were greater than the costs they incurred to
participate in the project. Moreover, our own analysis also suggests that
PARIS could help other states save program funds by identifying and
preventing future improper payments.

Despite the success that some states have reported with PARIS, the project
has several limitations. First, the opportunity to detect improper duplicate
payments is not as great as it could be because only one- third of the
states participate in the project, and thus, a substantial portion of the
national public assistance recipient population is not included in the PARIS
matching process. This is partly because the federal agencies responsible
for these programs have done little to encourage state participation in
PARIS through outreach efforts or to systematically address state concerns
that have contributed to limited participation.

6 We visited the following jurisdictions participating in PARIS: Tallahassee
and Miami, FL; Baltimore, MD; Harrisburg and Philadelphia, PA; New York
City, NY; and Washington, D. C. 7 We interviewed officials from Arkansas,
California, Colorado, Louisiana, Nevada, Texas, and Utah, and conducted site
visits to Sacramento and San Francisco. 8 The following states provided us
with data for our analysis: Kansas, Maryland, New York, Pennsylvania, and
Texas. Results in Brief

Page 4 GAO- 01- 935 PARIS Project Can Help States

Second, participating states do not have adequate protocols or guidelines to
facilitate critical interstate communication. As a result, some states have
reported problems that compromise the effectiveness of the project, such as
difficulty determining whether an individual identified in a match is
actually receiving benefits in another state. Third, state administrators
for the TANF, Medicaid, and Food Stamp programs have not always placed
adequate priority on using PARIS matches to identify recipients who are
residing in other states. As a result, individuals may continue to receive
or have benefits paid on their behalf in more than one state even after they
were identified through the matching process. Finally, because the PARIS
match is only designed to identify people after they are already on the
rolls, it does not enable the states to prevent improper payments from being
made in the first place.

Several actions could improve PARIS and help the project save additional
federal and state program funds. Although these actions would require the
states to carry the primary responsibility for improving the day- to- day
operations of the project, the federal government could take a greater
leadership role in helping facilitate their efforts. This leadership role
would have to balance the need for adequate state flexibility to administer
the programs with the need for accountability to ensure that federal and
state program funds are spent only on eligible individuals. We are making
recommendations to the Secretary of HHS that will improve the PARIS project
while taking these needs into account.

In commenting on a draft of this report, HHS officials agreed with the
overall intent of our recommendations, but consistently stressed the need
for additional funding and staff resources to increase their PARIS
activities. FNS officials noted that the report is a balanced and fair
description of the PARIS project, but they expressed concern that several
passages suggest that FNS should have a more formal role in PARIS.

The PARIS interstate match program was initiated to help state public
assistance agencies share information with one another. Its primary
objective is to identify individuals or families who may be receiving or
having duplicate payments improperly made on their behalf in more than one
state. In this voluntary project, the participating states agree to share
eligibility data on individuals who are receiving TANF, Food Stamps,
Medicaid, or benefits from other state assistance programs. The
participating states are primarily responsible for the day- to- day
administration of PARIS, and each state designates a coordinator for the
project. Background

Page 5 GAO- 01- 935 PARIS Project Can Help States

PARIS uses computer matching to identify improper benefit payments involving
more than one state. This process entails comparing participating states?
benefit recipient lists with one another, using individuals? SSNs. Other
items of information are included in the files that the states share, such
as the individual?s name, date of birth, address, case number, public
assistance benefits being received, and dates that benefits were received.
Matches are conducted by the Defense Manpower Data Center (DMDC) in
February, May, August, and November of each year. 9 DMDC produces a file of
all the SSNs on the list submitted by the participating state that are the
same as the SSNs appearing on the list of some other state and provides the
matched records, known as match hits, to ACF, who forwards them to the
appropriate states. To be considered a working member of PARIS, states agree
to participate in at least the August match each year.

Once the participating states receive the file of matched SSNs from DMDC,
they are expected to send the match hits to the appropriate staff for
follow- up or investigation. The staff may take a number of steps to verify
information that affects individuals? eligibility for benefits. These steps
include requiring an individual to come into the office to show proof of
residency and contacting other states to verify whether the individual is
still receiving benefits from those states. Improper benefit payments may be
made because of client error, agency administrative error, or fraud and
abuse. A client error might occur when an individual receiving program
benefits in one state moves to another state but fails to report the move to
program authorities. An administrative error could occur when a local
benefit worker is informed that the recipient is moving out of the state but
fails to update the record. Without PARIS matching, such errors might not be
detected until the individual is asked to reverify program eligibility,
which could occur as much as a year later. Additionally, the reverification
of eligibility might not detect fraud or abuse when a person deliberately
obtains benefits in more than one state by providing false information to
program authorities.

If after investigating the match hit, state or local officials determine
that an individual is improperly receiving public assistance benefits in
their state, they may initiate action to cut off benefits. In general, to
protect the rights of the recipients, administrative due process
requirements must be followed before benefits can be cut off. For example,
an individual may be given up to 30 days to respond to a formal notice that
benefits will be

9 DMDC acts as an independent contractor for ACF and the states, and it
provides computer matching services free of charge. How PARIS Operates

Page 6 GAO- 01- 935 PARIS Project Can Help States

stopped. Moreover, if the recipient can demonstrate that he or she is
residing in the state and is eligible for assistance, then benefits may be
continued or reinstated.

To ensure the confidentiality of the records that are identified in a match,
states must agree to the terms specified in the uniform computer matching
agreement that all states sign as a condition of their participation in
PARIS. Section three of this agreement requires that the states

?? will ensure that confidential recipient information received pursuant to
this Agreement shall, as required by law, remain confidential and be used
only for the purpose of the above described match and for verifying
eligibility and detecting and preventing fraud, error and abuse in [their]
respective Programs.? 10

Our review focuses on three benefit programs covered by PARIS interstate
matches: TANF, Medicaid, and Food Stamps. Benefits for TANF and Food Stamps
are provided directly to recipients; however, Medicaid payments are made
directly to those who provide health care services, such as MCOs and other
health care providers. All three programs are generally administered at the
state and local level, but are funded with federal money or a combination of
federal and state money. Depending on the state, the same staff in local
offices may determine eligibility and benefit levels for all three programs.
However, some states administer the TANF and Food Stamp programs separately
from the Medicaid program. Table 1 provides a brief description of the three
programs.

10 The agreement also requires the states to take reasonable steps to ensure
the security and confidentiality of information obtained under the
agreement, including having ?? locked files or other devices reasonably
calculated to prevent unauthorized copying or removal of manually held data;
passwords, access logs, badges or other methods of safeguarding
electronically or mechanically held data; limited physical access; limited
access to input documents and output documents; and design provisions which
avoid unnecessary use of names or other personal identifiers of data
subjects.? About the Three Programs

We Examined

Page 7 GAO- 01- 935 PARIS Project Can Help States

Table 1: Programs We Examined That Are Covered by PARIS Federal program
Implementation TANF a Medicaid Food Stamps Authorizing law Personal
Responsibility and Work

Opportunity Reconciliation Act (1996) Title XIX of the Social Security Act
(1965) Food Stamp Act (1964)

Purpose Assistance to needy families and to end dependence on government
benefits by promoting job preparation, work, and marriage.

Medical assistance for certain categories of needy individuals, including
families with low incomes and resources.

Monthly allotments of coupons or electronic benefits for low- income
households redeemable for food at retail stores.

Funding In fiscal year 2000, the program was funded by a total of $23.5
billion in federal and state expenditures, including 12. 5 billion in
federal block grants, and $11.1 billion in state funds. The program served
about 5.8 million individuals as of September 2000.

Funded jointly by the states and the federal government. State expenditures
trigger federal matching funds at a rate, that is based on a statutory
formula, that varies by state. The federal matching rate averages 57 percent
nationally. In 2000, the program covered health services for 34 million
persons at a cost of about $190 billion.

Funded by the federal government. In 2000, about 17 million individuals
received Food Stamps at a cost of $17.1 billion. b

Administration States have flexibility over the administration and design of
their welfare programs, but must impose several federal requirements,
including time limits on aid.

The Centers for Medicare & Medicaid Services (CMS) provides broad oversight
of Medicaid at the federal level; however, each state is responsible for
managing its own program. Many states contract with managed care
organizations to provide primary care and sometimes other services to
beneficiaries for a set monthly payment.

USDA?s Food and Nutrition Service manages the program through agreements
with state agencies. The states administer the program through local benefit
offices. a Temporary Assistance for Needy Families.

b These figures are preliminary and subject to change. Source: GAO.

A fourth program- the SSI Program administered at the federal level by SSA-
is indirectly related to the PARIS interstate matches. In many states, SSI
recipients are automatically qualified to receive Medicaid and are therefore
included in PARIS matches.

PARIS coordinators in most of the 16 participating states and the District
of Columbia told us they believe the interstate match is effective in
identifying improper TANF, Medicaid, and Food Stamps benefit payments in
more than one state. By eliminating these duplicate recipients from the
rolls, states can prevent future improper payments and save program dollars.
However, few states tracked the actual savings realized from the PARIS
match. Four states and the District of Columbia reported a total of about
$16 million in estimated savings from various PARIS matches PARIS: A
Potentially

Cost- Beneficial Means of Controlling Improper Payments

Page 8 GAO- 01- 935 PARIS Project Can Help States

conducted in 1997, 1999, and 2000. 11 A substantial proportion of the
estimated savings was attributed to the Medicaid program. 12 While officials
in only three states have compared the costs to the benefits that result,
their studies indicate that the matching is cost- beneficial. We prepared
our own analysis, which also suggests that PARIS may help states save
program funds by identifying and preventing future improper payments.

PARIS helps to identify improper benefit payments in bordering and
nonbordering states according to most of the PARIS coordinators we spoke
with. The February 2001 match identified almost 33,000 instances in which
improper payments were potentially made on behalf of individuals who
appeared to reside in more than one state. Of these, 46 percent of the
potential hits involved Medicaid benefits alone, while the remaining 54
percent involved some combination of TANF, Medicaid, and Food Stamps.
However, most of the states did not maintain detailed records on the number
of potential match hits that were, in fact, found to be instances of
improper payments. Nor were they able to tell us what proportion of improper
TANF and Food Stamp payments were due to client error, administrative error,
or fraud and abuse. 13

Some PARIS coordinators believe that fraud and abuse may be more common in
areas where two states share an urban border. For example, one coordinator
told us that individuals living in the District of Columbia metropolitan
area could travel in minutes to Maryland and Virginia and apply for benefits
in each place. Figure 1 depicts participating states and the District of
Columbia and their shared borders.

Independent of PARIS, some states conduct interstate matches with bordering
states to prevent improper payments caused by either error or potential
fraud and abuse. Many of these states now participate in the PARIS match.
PARIS coordinators told us that the PARIS approach offers significant
advantages over single state- to- state matches.

11 All estimated savings attributed to matches conducted in 1999 and 2000,
except for the District of Columbia, which reflects savings from one match
in 1997. 12 Duplicate Medicaid payments may occur because recipients who are
covered by MCOs do not notify a state that they are moving to another state.
The state may continue to pay premiums to the MCO because it is unaware the
recipient has moved. Savings can be realized by identifying recipients who
move and stopping future payments.

13 It appears that most Medicaid improper payments were due to
administrative error. PARIS Detects Improper

Payments in Bordering and Nonbordering States

Page 9 GAO- 01- 935 PARIS Project Can Help States

Figure 1: States Participating in Recent PARIS Matches

Source: DMDC records of states that participated in at least one PARIS
interstate match from August 2000 through February 2001.

For example, PARIS makes it possible for a state to match with numerous
other states by simply submitting a file to a central agency. In addition, a
uniform data- sharing agreement covers the exchange, and the DMDC adjusts
for incompatibilities between different computer systems. This unified
approach can be more efficient than individual state- to- state matches and
can help to reduce the expense of matching.

In addition to simplifying matches with bordering states, PARIS also
facilitates data sharing with nonbordering states. This is important because
even when two states do not share a border, improper payments can still be
made, whether due to error or deliberate deception, such as fraud and abuse.
For example, PARIS officials in New York discovered a woman receiving TANF
benefits in New York for five children who were actually living with
relatives and receiving benefits in Illinois. Table 2

States participating in recent PARIS interstate matches AZ

WA OR

CA NV

NM TX LA

OK CO

UT ID

MT WY

ND SD

NE KS

MN IA

MO AR

MS TN

KY IL WI

MI IN OH

WV VA

NC SC GA AL

FL AK HI

PA NY

VT NH

ME NJ

RI DE

DC MA

CT MD

Page 10 GAO- 01- 935 PARIS Project Can Help States

shows the results from the February 2001 PARIS match for selected states,
including nonbordering states. 14

Table 2: Number of Potential Improper Payments for Selected States, February
2001 Matching states (and D. C.) District of Columbia Florida Illinois
Maryland New York North Carolina Pennsylvania

District of Columbia 30 18 778 88 179 37 Connecticut 4 48 42 22 369 97 19
Florida 30 883 246 1,097 752 220 Illinois 18 883 170 539 409 144 Kansas 2 38
106 3 23 22 5 Maryland 778 246 170 498 477 235 Massachusetts 9 91 93 44 616
90 67 Missouri 16 138 985 26 62 72 42 Nebraska 5 34 146 8 13 10 5 New York
88 1,097 539 498 1,268 651 North Carolina 179 752 409 477 1,268 218
Pennsylvania 37 220 144 235 651 218 Rhode Island 2 40 43 25 312 30 22
Tennessee 2 67 198 13 46 71 10 Virginia 160 158 115 336 482 583 95

Total 1,330 3,842 3,891 2,881 6,064 4,278 1,770

Note 1: This table represents the initial match hits for selected states
that were returned by DMDC for follow up. The totals for the table do not
add up to 33,000 because not all states that participated in the match are
included here. In addition, not all the match hits identify actual
incidences of duplicate benefits, and therefore, not all match hits will
yield savings in the three programs.

Note 2: Shaded blocks indicate matches between bordering states. Source: GAO
analysis of DMDC records.

About 80 percent of the match hits listed in table 2 are between states that
do not border one another. For example, North Carolina has more match hits
with Florida and New York than it does with neighboring Virginia. In
addition, in New York and Pennsylvania, match hits with nonbordering states
represented 73 percent and 50 percent of their total match hits,
respectively. Although both of these states have matched recipient data with
bordering states for years, the PARIS match identified numerous instances of
potential duplicate benefits in nonbordering states that might not otherwise
have been detected.

14 The numbers reflect the total number of matches reported by DMDC, not the
proportion of those matches that actually produced program savings.

Page 11 GAO- 01- 935 PARIS Project Can Help States

While most states do not track the savings they have achieved or the costs
they incurred because of the PARIS match, a small number of states were able
to document the results of participating in the project. Four states and the
District of Columbia provided us with their estimated savings from
participating in PARIS. Three of the states also performed cost- benefit
analyses, demonstrating that they found PARIS to be cost- beneficial. 15

 Pennsylvania estimated that two quarterly matches in 2000 produced more
than $2.8 million dollars in annual savings in the TANF, Medicaid, and Food
Stamp programs and achieved a savings- to- cost ratio of almost 12 to 1.
About $2.5 million (87 percent) of the total estimated savings are
attributed to the Medicaid program.

 Maryland estimated that its first PARIS match in 1997 produced savings of
$7.3 million in the Medicaid program alone. The match identified numerous
individuals who were originally enrolled in Medicaid due to their SSI
eligibility, but at the time of the match no longer lived in the state.
Subsequent matches conducted between November 1999 and August 2000 have
produced savings of about $144,000 in the TANF, Medicaid, and Food Stamp
programs, with a savings- to- cost ratio of about 6 to 1.

 Kansas estimated that two PARIS matches in 1999 and 2000 resulted in
savings of about $51,000 in the TANF, Medicaid and Food Stamp programs, with
a savings- to- cost ratio of about 27 to 1.

 New York reported that improper payments identified in four matches
conducted in 1999 and 2000 produced estimated savings of $5.6 million;
however, the state did not collect data on the costs associated with
investigating these matches.

 The District of Columbia estimated that one PARIS match conducted in 1997
resulted in savings of about $311,000 in the TANF and Food Stamp programs;
however, officials did not collect savings data for the Medicaid program,
nor did they collect cost data.

Our discussions with numerous state and federal officials have led us to
conclude that the substantial variation in the estimated program savings and
savings- to- cost ratios across these states is attributable to a number of
factors. These factors, which could also apply to any participating state,
include differences in

 the extent to which state and local officials follow up on (or fail to
pursue) match hits and take action to cut off benefits where appropriate;

 the methods and assumptions states use to estimate their savings; 15 We
did not independently verify the validity or reliability of the states?
data. PARIS Helped Several

States Avoid Millions in Improper Payments

Page 12 GAO- 01- 935 PARIS Project Can Help States

 the proportion of match hits that are valid in that they are found to
reflect actual improper benefits being paid in more than one state (a higher
proportion of valid match hits will generally yield more program savings
than a lower rate, and is more likely to be cost- beneficial); 16

 the estimated number of months of avoided benefit payments;

 the size of the recipient population and the monthly benefits provided in
each state under the TANF, Medicaid, and Food Stamp programs;

 how long it takes local office staff or fraud investigators to follow up
on match hits;

 the salary costs of state and local staff involved with PARIS; and

 the cost to create an automated list of recipients at the state level to
be sent to DMDC.

Because so few states had analyzed their savings and costs from
participating in PARIS, we performed an independent analysis to assess how
certain factors might influence the extent to which participating in PARIS
could achieve program savings. We studied how certain key variables, such as
the number of programs included, the proportion of match hits that are
valid, and the estimated number of months of avoided benefit payments could
affect the overall savings a state might achieve by participating in PARIS.
We used national data where available (such as average benefits paid to
recipients for each program). When national data were not available, we used
the experiences of five states for our analysis. 17 We used professional
judgment to determine the values for several key assumptions in our
analysis. Specifically, using a hypothetical example in which 100 match hits
are sent to local benefit offices for staff to investigate, we assumed that

 each match hit requires 2 hours to determine whether benefits are
improperly being paid in more than one state and costs $68.97 on average,
resulting in a total of $6,897 in salaries and related expenses to follow up
on all 100 match hits;

16 The proportion of valid match hits depends, in part, on the extent to
which states filter their raw match data. For example, many states check for
and filter out cases that have been recently closed. The cases that are
filtered out are not sent to local benefit workers for follow- up, thereby
reducing the number of match hits requiring action and ultimately lowering
the costs (and potentially the savings) of participating in PARIS.

17 We used national data for the benefits paid to individuals and families
in the TANF, Medicaid, and Food Stamp programs. Other cost and savings
figures we used are based on data provided by Kansas, Maryland, New York,
Pennsylvania, and Texas. Texas was not participating in PARIS at the time
our analysis was performed. However, we included data from this state
because we determined that a current border state match it conducts with
three neighboring states is similar to the PARIS match.

Page 13 GAO- 01- 935 PARIS Project Can Help States

 the average state cost is about $440 to generate the automated list;

 20 percent of the match hits investigated are found to be valid; and

 program savings come entirely from the future benefit payments that are
avoided.

(See app. I for a more detailed description of the data and assumptions used
in our analysis.)

Our analysis suggests that PARIS, as it currently operates, could help save
both federal and state program funds. 18 In particular, our analysis
indicates that if states include the TANF, Medicaid, and Food Stamp programs
in their matching activities, the net savings could outweigh the costs of
participation. 19 Using our hypothetical example in which 100 match hits are
sent to local benefit office staff for follow- up, we illustrate in table 3
how the savings to a state from participating in one PARIS match could vary
depending on (1) the number of programs included in the match and (2)
differences in the valid hit rate. The table assumes that the savings for
each program accrue for 3 months.

If 20 percent of the match hits are valid (they accurately identify 20 out
of 100 instances in which improper benefits are being paid in more than one
state) and the individuals identified are enrolled in all three programs,
the match would produce gross savings of almost $42,000, yielding a
savingsto- cost ratio of about 5 to 1. Ultimately, the match would result in
net savings of more than $34,000, as shown in table 3, taking into account
total match costs of about $7,000. Conversely, costs exceed savings under
only one scenario in this example. A valid hit rate of 10 percent, in which
the match only includes the Food Stamp program- a rate substantially below
what participating states have reported- would result in a net cost to the
state of about $3,300.

18 Savings from improper payments that are avoided in the TANF program would
accrue to the states, savings from the Medicaid program would accrue to both
the states and the federal government, and savings from the Food Stamp
program would generally accrue to the federal government. The proportion of
the savings in the Medicaid program accruing to the states and federal
government would vary depending on the percentage of program funds that
individual states contribute toward their Medicaid programs.

19 The figures we discuss here are net savings or costs. These figures
represent the final cost or savings a state could experience once the costs
of conducting the match have been subtracted.

Page 14 GAO- 01- 935 PARIS Project Can Help States

Table 3: Estimated Net Savings Given Varying Valid Hit Rates Valid match
hits and net savings Program 10% 20% 30%

TANF only $912 $9,161 $17,409 Medicaid only 1,225 9,787 18,349 Food Stamps
only -3,266 a 804 4,875 TANF and Food Stamps 4,983 17,302 29,622 TANF and
Medicaid 9,474 26,285 43,095 Food Stamps and Medicaid 5,296 17,928 30,561

All three $13,545 $34,426 $55,308

Note: The table assumes that 3 months of benefit payments are avoided as a
result of the match. Savings in program expenditures are in nominal 1999 and
2000 dollars. a Represents net costs rather than savings.

Source: GAO analysis.

The number of months that future benefit payments are avoided can also
influence the amount of savings that result from a PARIS match. Table 4
illustrates the variation in program savings that could result depending on
the number of months of future benefits that are avoided and the number of
programs matched, given a 20- percent valid hit rate. 20

Table 4: Estimated Net Savings Given a Varying Number of Future Benefit
Payment Months Avoided

Benefit months avoided and net savings Program 1 month 3 months 6 months

TANF only -1,838 a 9,161 25,658 Medicaid only -1,629 a 9,787 26,911 Food
Stamps only -4,623 a 804 8,946 TANF and Food Stamps 876 17,302 41,941 TANF
and Medicaid 3,870 26,285 59,906 Food Stamps and Medicaid 1,085 17,928
43,194

All three 6, 584 34,426 76,189

Note: The table assumes a valid match hit rate of 20 percent. Savings in
program expenditures are in nominal 1999 and 2000 dollars. a Represents net
costs rather than savings.

Source: GAO analysis.

20 The average valid hit rate reported by five states was 32. 4 percent.

Page 15 GAO- 01- 935 PARIS Project Can Help States

As the table shows, there are three scenarios under which a state in our
analysis would experience a net loss from participating in PARIS. One
month?s worth of TANF, Medicaid, or Food Stamp benefits avoided would yield
a net cost to the state of between approximately $2,000 and $5,000. However,
the match would produce savings in all other possible scenarios. For
example, it would yield over $83, 000 in gross savings if 6 months of
benefits are avoided and the match was performed for all three programs (a
savings- to- cost ratio of about 11 to 1). The net savings would be about
$76,000.

Our analysis assumes that only a small number of match hits are sent for
follow- up (100), which results in a small number of valid hits (20). A
larger number of valid hits would likely result in greater savings as well.
For example, while some states, such as Kansas, with smaller recipient
populations have reported relatively small numbers of valid hits and lower
levels of savings, other states, such as Pennsylvania and New York, with
larger recipient populations have had much higher numbers of valid hits and
much greater levels of savings.

Although the information provided by states and our analysis indicate that
participating in PARIS interstate matches can save federal and state funds,
savings are not the only benefit of participating in PARIS. Interstate
matches are an important internal control to help states meet their
responsibility for ensuring that public assistance payments are only made to
or on behalf of people who are eligible for them. In addition, PARIS
officials in eight states told us they believe the PARIS interstate matches
can help deter people from applying for duplicate public assistance
payments.

The PARIS project?s interstate matching has helped identify cases of
duplicate benefits that otherwise would likely have gone undetected;
however, PARIS has been limited by several factors. First, only one- third
of the states are participating in the matches, and a large portion of the
public assistance population is not covered by the matching. Second, the
project has some problems with coordination and communication among project
participants. Third, some states are giving inadequate attention to the
project. As a result, match hits are not being resolved, and in particular,
duplicate payments made for Medicaid beneficiaries receive low priority.
Finally, the project cannot help prevent duplicate benefits from occurring
in the first place, but can only identify and help stop them after they have
started. Limitations of PARIS

Page 16 GAO- 01- 935 PARIS Project Can Help States

Only one- third of the states are participating in the PARIS interstate
matches. At the time of our review, 16 states and the District of Columbia
were participating. As a result, the public assistance records of the other
34 states were not being shared with participating states. 21 These
nonparticipating states contain 64 percent of the population that is likely
to be eligible for public assistance. 22

We spoke to officials in seven nonparticipating states to learn their
reasons for not participating. They noted the state?s preoccupation with
more urgent matters, such as implementing new programs or systems, and the
fact that information about the project had not reached someone with the
interest and authority to get the state involved. They also cited some
concerns about the project. These include

 lack of data showing that participating would produce savings for their
state;

 nonparticipation of bordering states, which are perceived as the most
likely sources of valid match hits;

 lack of written guidance on coordinating the resolution of match hits with
other states; and

 inadequate federal sponsorship of PARIS and the resulting lack of
assurance that the project will continue.

Efforts by federal agencies to increase participation in the project have
been minimal. ACF, the lead agency on the project, has not officially
recognized PARIS and devotes very little resources to it. 23 ACF management
has not taken actions, such as sending letters to state TANF directors to
inform them about the project and encourage them to participate. Also, ACF
management has not asked other federal agencies to work with ACF on the
project and help get more states involved. CMS, the federal agency that
stands to reap the greatest savings from the project, has made no effort to
encourage state Medicaid agencies to

21 Some states run separate matches with other states not participating in
the PARIS project. However, these matches are usually with just a few
neighboring states, and many states do not do any interstate matching.

22 The nonparticipating states contain 62 percent of the total U. S.
population and 64 percent of the U. S. population in poverty, according to
1999 data from the U. S. Census. However, since we completed our work,
California participated in one interstate match on a trial basis, and
Louisiana, Nevada, Puerto Rico, Texas, and Utah were considering joining.

23 For example, ACF has not taken actions, such as allocating funds to the
project and including the project in its performance plan, that would
indicate official management recognition of the project. One- Third of the
States

Participate

Page 17 GAO- 01- 935 PARIS Project Can Help States

participate. In 1999, FNS sent a letter to state Food Stamp agencies
encouraging them to participate in PARIS interstate matches; otherwise, FNS
has had little involvement in the project. This lack of official support for
the PARIS project may contribute to the low participation rate. For example,
the TANF officials that we spoke with in one of the nonparticipating states
who were relatively new to their positions said they had never heard of
PARIS. In another nonparticipating state, a Medicaid official told us the
state would be much more likely to participate in PARIS if CMS encouraged it
to do so.

The PARIS project has had various problems with coordination and
communication that limit the project?s effectiveness. The problems include
the following.

 Difficulties contacting other states. Benefit workers in four of the five
participating states that we visited said they have had difficulties
contacting benefit workers in some other states to obtain information to
resolve match hits or to get the evidence needed to take action against
clients. Problems making contacts occur because the telephone numbers that
states provide for obtaining information on individual cases are sometimes
inaccurate or never answered or are central numbers that are just the
starting point for finding the right person.

 Submission of incomplete and incompatible data. We noted that some of the
states submit data for matching that are likely to increase the number of
invalid match hits and the amount of work other states will have to do to
determine if match hits are valid. For example, we found that some states
include closed cases among the active cases submitted for matching, cases
with improper SSNs, or cases that omit the dates clients started receiving
benefits.

 Uncertainties concerning responsibilities for collecting overpayments from
individuals. PARIS officials from three states said it is not clear which
state should assess and collect an overpayment when it is found that a
client has been receiving TANF or Food Stamp benefits from two or more
states. For example, it is not clear if the state where the client does not
reside should assess an overpayment because, as a nonresident, the client
was not eligible to receive benefits from the state or if the state where
the client does reside should assess the overpayment because it is much more
likely to be able to collect the overpayment. 24 Also, it is not

24 According to PARIS officials in two states, collecting overpayments from
nonresidents is difficult because the state cannot use one of its most
effective means of collection- recouping overpayments from current benefits.
Participating States Have

Difficulty With Coordination and Communication

Page 18 GAO- 01- 935 PARIS Project Can Help States

clear how to determine which state should assess an overpayment when the
client claims two residences very near each other but in different states,
and it is not known where the client actually lives.

Although some coordination and communication problems are likely to occur in
any project that involves multiple states and different federal agencies,
the project?s lack of formal guidance and processes makes these problems
more likely to occur. Currently, the formal guidance for the program only
includes the file format the states need to provide for the match. However,
it does not address matters such as the type of case information other
states? benefit workers should be able to get when they call the telephone
number provided for a case. Also, the guidance does not have written
definitions of some key terms, such as ?active case,? or explanations of how
the various data fields are to be used by states to investigate match hits.
Further, the project has no guidance or protocols for coordinating the
assessment and collection of overpayments. However, ACF, CMS, and FNS have
not provided the management or administrative support- such as a formal
focal point at the federal level- that would be needed to coordinate the
project more effectively and help develop such guidance and protocols. 25

In some states, management has given little or no attention to the PARIS
interstate matches and has allowed match hits to go unresolved. This problem
is more pronounced with Medicaid match hits because, in some states, they
are given a lower priority than match hits involving TANF or Food Stamps.

We found evidence that in at least three states that have participated in
the PARIS project since August 1999, match hits for the entire state or for
some densely populated areas were not being resolved. The PARIS coordinator
in one state told us that match hits in his state have never been sent out
to workers to be resolved. In a second state, a large metropolitan area had
not received any match hits from its district office until shortly before
our visit in February 2001. The PARIS coordinator in a third state told us
that a large county sometimes ignored the PARIS match hits sent to it for
resolution.

The problem of not resolving match hits appears to be most pronounced in the
Medicaid program. Information we received from DMDC indicates that

25 Currently, one individual at ACF handles most PARIS administrative
activities on a parttime, informal basis. Management Attention Is

Inadequate in Some States

Page 19 GAO- 01- 935 PARIS Project Can Help States

some states may not be focusing sufficient attention on their Medicaid match
hits. Because DMDC does not retain state data used for the PARIS matches, we
were not able to determine how many match hits involving Medicaid are not
resolved and thus recur each quarter. However, data provided by DMDC for the
February 2001 PARIS matches show that some states have a relatively large
percentage of match hits involving Medicaid. For example, if 40 percent of
the records a state submitted for matching were for clients receiving
benefits in a particular program, then one might reasonably expect to find
that about 40 percent of the match hits involved that program. Thus, finding
a disproportionately higher rate of match hits involving that program could
suggest a possible problem. Such is the case with six states that have
participated in PARIS since February 2000 or before. For each of the six
states, the February 2001 PARIS match resulted in a proportionately higher
percentage of match hits involving Medicaid than would generally be
expected. For example, in one state, 60 percent of the records submitted for
matching were cases involving only Medicaid benefits (not TANF or Food
Stamps), but 78 percent of the resulting match hits were for such cases. In
another state, 31 percent of the records submitted were for cases involving
Medicaid received due to eligibility for SSI, but 69 percent of the
resulting match hits were for such cases.

Match hits involving duplicate Medicaid benefits frequently occur, not
because of fraud or abuse, but because Medicaid beneficiaries often do not
notify the state when they move out of state. Therefore, a state will keep
beneficiaries on the rolls until it discovers that they have moved. The
state may make this discovery during a routine reverification of
eligibility, which is generally performed once a year or less often.
However, officials from several states have told us that their states never
reverify the eligibility of a certain type of Medicaid beneficiary, such as
one who is eligible based on his or her receipt of SSI. 26 Therefore, the
PARIS matches often involve this type of beneficiary. Although a state
receives notifications from SSA when SSI clients move out of the state,
states often do not remove Medicaid beneficiaries from their rolls based on
these notifications, according to an SSA official.

The PARIS coordinators for two states told us this problem came to light
after they examined their first PARIS interstate match results and found a

26 In 32 states and the District of Columbia, people receiving SSI are
automatically eligible to receive Medicaid; that is, they are considered to
have met the states? eligibility requirements because they are receiving
SSI. Officials in some states told us they do not periodically reverify the
Medicaid eligibility of people who are automatically eligible for Medicaid.

Page 20 GAO- 01- 935 PARIS Project Can Help States

startling number of match hits involving SSI recipients who were on the
state?s Medicaid rolls. One state compared the Medicaid match hits from its
first PARIS run with SSA files and found 5,000 SSI recipients on the state?s
Medicaid rolls who, according to SSA records, were not residing in the
state. This prompted the state to do a similar match with SSA records using
all the state?s Medicaid beneficiaries. The state then followed up with
letters to Medicaid enrollees who the matches indicated no longer lived in
the state. As a result of the PARIS and subsequent SSA matches, the state
identified 17,000 people on its Medicaid rolls who were no longer eligible
for Medicaid in the state. 27 We heard a similar story from another state.
Both states, we were told, had been making monthly payments to MCOs for the
Medicaid beneficiaries, who would have stayed on the states? rolls
indefinitely if the state had not participated in the PARIS matches.

Yet even after receiving large numbers of Medicaid match hits, some states
appear not to be resolving them or addressing the problems with their
Medicaid rolls. We have been told by some PARIS coordinators that the
departments administering Medicaid are focusing their efforts on getting
people on the Medicaid rolls rather than removing people who are no longer
eligible. PARIS officials in two states said that they believe the local
benefit workers or the offices responsible for Medicaid have not adjusted
their thinking to recognize the shift from a fee- for- service to a managed
care environment. In the past, when Medicaid services were provided on a
fee- for- service basis, costs were incurred only if beneficiaries sought
medical treatment and providers submitted bills for the treatment.
Therefore, if a beneficiary moved out of state but remained on the state?s
Medicaid rolls, medical expenses were not incurred for the beneficiary if he
or she did not seek treatment in the state. However, when the state makes a
fixed monthly payment to an MCO for each Medicaid beneficiary, as is done
under some managed care arrangements, the state makes payments to the MCO
regardless of whether the beneficiary ever seeks medical treatment.

The PARIS project was designed to identify duplicate benefits after they
have been provided, not to prevent the duplicate benefits from occurring in
the first place. Therefore, the PARIS matches are part of what has been
described as a ?pay and chase? process, in which states pay benefits to
clients and then try to recover overpayments when they discover the

27 The 17,000 people were no longer eligible because they no longer lived in
the state; they were deceased; or they were no longer eligible for SSI,
which was the basis for their Medicaid eligibility. PARIS Identifies
Duplicate

Benefits After They Are Provided

Page 21 GAO- 01- 935 PARIS Project Can Help States

clients were not eligible for the benefits. 28 Preventing an improper
payment in the first place is preferable to ?pay and chase? because
overpayments are often difficult to collect from low- income clients who no
longer live in the state. Also, when states make payments to MCOs for
beneficiaries who should no longer be on their Medicaid rolls, these funds
are wasted unless they can be recouped. According to a Medicaid official, it
may be difficult for states to recoup overpayments to MCOs caused by errors
in states? Medicaid rolls.

Officials from most states we spoke with said they would like a datasharing
process that could be used before benefits are provided- that is, a process
that would allow state caseworkers to check other states? data to see if an
applicant was already receiving benefits elsewhere before the state approved
an application for benefits. Such a process would have to provide prompt
responses (probably within 24 hours) to inquiries- something very different
from the quarterly PARIS matches. One option for this process includes a
national database of clients receiving public assistance in any state. Such
a database would be maintained by the federal government and would consist
of records submitted and regularly updated by the states. 29

Implementing such an option would require federal leadership and funding to
address programming and operating expenses and potentially the
standardization of data and information systems among participating states.
30 Also, while implementing this option could help prevent duplicate
payments, it must be balanced against the additional privacy concerns that
might arise. 31

28 See OMB Benefit Systems Review Team, Strategies for Efficiency: Improving
the Coordination of Government Information Resources (Washington, D. C.:
Jan. 1997). 29 An example of a national database that uses records submitted
by the states is the National Directory of New Hires, which is used to
locate parents who fail to provide child support payments.

30 For more information on options for enhanced data sharing, see OMB,
Strategies for Efficiency, and FNS, Options for a National Database to Track
Participation in Federal Means- Tested Public Assistance Programs: Report to
Congress (Nov. 1999).

31 Creation of a new database to be accessed by states could potentially
increase the risk of unauthorized access to, and use of, personal
information about public assistance recipients.

Page 22 GAO- 01- 935 PARIS Project Can Help States

The PARIS project offers states a potentially powerful tool for improving
the financial integrity of their TANF, Medicaid, and Food Stamp programs.
However, the project has fallen short of realizing its full potential, as is
most clearly evidenced by relatively low state participation. While PARIS?
success ultimately rests in the hands of the states, key federal players
have not done enough to provide a formal structure to the project that
encourages and facilitates state participation. More specifically, ACF, CMS,
and FNS have not taken the lead in establishing a focal point in the federal
government for coordinating the project. This is crucial given the
complicated relationships among the three programs and among the federal,
state, and local government entities responsible for implementing them.
Additionally, the three federal agencies have not worked together to develop
guidance and protocols that are key for helping states share information and
best practices. Finally, these agencies have not formally recognized, nor
devoted sufficient resources to, the project, despite its potential to
identify improper payments and save program funds. Importantly, this lack of
formal federal recognition might signal to some states that the project
should not be taken seriously.

To help states improve the effectiveness of PARIS and prevent duplicate
benefit payments to TANF and Medicaid recipients, we recommend that the
Secretary of HHS direct the Administrators of ACF and CMS to formally
support PARIS and provide guidance to participating states. Such support and
guidance should include the following actions:

 Create a focal point charged with helping states more effectively
coordinate and communicate with one another. An existing entity, such as the
Interagency Working Group, could provide the mechanism for such a focal
point. This entity could also serve as a clearinghouse for sharing best
practices information that all states could use to improve their procedures,
such as comparisons of match filtering systems.

 Take the lead to help the PARIS states develop a more formal set of
protocols or guidelines for coordinating their match follow- up activities
and communicating with one another.

 Develop a plan to reach out to nonparticipating states and encourage them
to become involved in PARIS. At a minimum, all states should be encouraged
to provide their TANF and Medicaid recipient data for other states to match,
even if they choose not to fully participate in PARIS. This would help to
ensure that all recipients nationally are included in PARIS matches.

 Coordinate with the USDA/ FNS Food Stamp program to encourage their
participation in PARIS at the federal level as well as their working more
Conclusions

Recommendations for Executive Action

Page 23 GAO- 01- 935 PARIS Project Can Help States

closely with individual states to improve the effectiveness of PARIS and
helping more states to participate.

Officials from the Department of Health and Human Services and the Food and
Nutrition Service provided comments on our report, the full text of which
appear in appendixes II and III, respectively. The agencies also included
some technical comments, which we have incorporated where appropriate.

In general, HHS agreed with the overall intent of our recommendations, but
consistently stressed the need for additional funding and staff resources to
increase their PARIS activities. With regard to our first recommendation,
HHS commented that it had created a PARIS work group composed of
representatives from ACF and DMDC and has encouraged other agencies, such as
CMS and FNS, to participate more actively in PARIS. HHS also stated that
additional funding and staff resources from all involved agencies could help
the work group to improve its services. We believe that while the PARIS
workgroup provides useful guidance to participating states, to date it has
been unable to resolve the problems and limitations we identified during our
review. As we note in the report, this is due in part to ACF, CMS, and FNS
not providing the management or administrative support necessary to correct
these problems. Our recommendation is intended to encourage greater
leadership by ACF, CMS, and FNS and a more coordinated proactive approach
among the agencies to working together and with the states to address the
limitations in PARIS.

With regard to our second recommendation, HHS cautioned that it is not
appropriate for a federal agency to dictate or appear to dictate the
protocol states use in their interaction with other states. HHS also argued
that states are best able to determine the necessary procedures for PARIS.
However, HHS acknowledged that with additional resources, ACF could help
states develop such procedures and disseminate them to other states as
necessary. We continue to believe that active federal leadership is needed
to solve the communication and coordination problems discussed in the
report. Consequently, we believe that ACF should act as a facilitator at the
federal level to help states overcome some of the challenges they have
reported communicating and coordinating with one another. Moreover, such
facilitation can and should occur without impinging on the states? ability
to administer the TANF, Medicaid, and Food Stamp programs in a manner that
best fits their needs. Agency Comments

and Our Evaluation

Page 24 GAO- 01- 935 PARIS Project Can Help States

With respect to our third recommendation, HHS generally agreed that ACF
could do a better job to reach out to additional states to persuade them to
participate in PARIS. However, HHS did not agree with our statement that
states could, at a minimum, provide their data for others to use, even if
they do not directly participate in PARIS themselves. We believe that while
full participation by all the states is clearly the preferred outcome, the
inclusion of nonparticipating states? public assistance data for use by
states participating in PARIS could help save additional benefit funds in
the TANF, Medicaid, and Food Stamp programs.

Finally, with regard to our fourth recommendation, HHS noted that ACF has
consistently coordinated with FNS in all PARIS activities, but agrees that a
closer working relationship with FNS would add to the effectiveness of
PARIS. We concur with HHS? assessment that ACF and FNS should work more
closely together to improve existing PARIS operations and persuade
additional states to participate.

FNS noted that the report is a balanced and fair description of the PARIS
project, but they expressed a concern that certain passages in the report
suggest that FNS should have a more formal role in PARIS, despite the fact
that PARIS is primarily an ACF project. They also identified several reasons
why PARIS is not used more by the Food Stamp program. They emphasized that
FNS is not required by statute to track interstate receipt of Food Stamp
benefits and that many states are already engaging in such activity on their
own.

Although we recognize that FNS is not the lead agency responsible for PARIS,
we do believe that FNS could take a more proactive stance to help coordinate
the program at the federal level and persuade additional states to
participate in PARIS. Moreover, we believe that although FNS is not mandated
by statute to participate in PARIS, the benefits of PARIS in terms of
potential program savings and enhanced program integrity warrant a more
active role for the agency. Our analysis suggests that federal leadership
from each of the involved federal agencies is critical to the success of
PARIS, particularly with regard to expanding the number of states that
participate in the project. In addition, while some states engage in
interstate matching as noted in the report, we believe a more structured,
far- reaching approach like that offered by PARIS is more effective.

FNS also commented that PARIS cannot prevent the initial duplicate payment
of benefits and that the matching activity may not be costeffective. We
believe that although PARIS cannot prevent duplicate benefits from being
provided when states initially determine individuals?

Page 25 GAO- 01- 935 PARIS Project Can Help States

eligibility for benefits, using PARIS is preferable to not matching at all.
Finally, the report notes that matching for the Food Stamp program alone may
not be cost- effective and emphasizes the advantage of matching for multiple
programs simultaneously.

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days after
its date. At that time, we will send copies to the Chairman, Senate
Committee on Governmental Affairs; Secretary of HHS; Administrator for CMS;
Administrator for FNS; and to other interested parties. Copies will also be
made available to those who request them.

Please contact me or Kay Brown at (202) 512- 7215 if you have any questions
concerning this report or need additional information. Jeremy Cox, Kathleen
Peyman, James Wright, and Jill Yost made key contributions to this report.

Sincerely yours, Robert E. Robertson Director, Education, Workforce,

and Income Security Issues

Appendix I: Data and Assumptions Used in Our Analysis Page 26 GAO- 01- 935
PARIS Project Can Help States

In the table below, we describe the data and assumptions used to support our
discussion on pages 12- 15.

Our analysis incorporated data from five states (Kansas, Maryland, New York,
Pennsylvania, and Texas), two federal agencies (Centers for Medicare &
Medicaid Services and Food and Nutrition Service), and selected research
studies.

Table 5: Scenario With Assumptions Used in Report Variable Value used Basis
Details Savings

New York 34. 0 Maryland 62.0 Pennsylvania 19.0 a Texas 21.0 Kansas 28.0
Proportion of valid match hits 20% The estimate contained in a cost- benefit

analysis by the Administration for Children and Families.

Average 32.8%

New York 65. 0 Pennsylvania 95.0 Texas 97.0 Closure (71%) Average for the 4
states providing data.

Kansas 28.0 New York 35. 0 Pennsylvania 5. 0 Texas 3. 0 Proportion of valid
match hits that

result in case closure or benefit reduction

Benefit reduction (29%) Average for the 4 states providing data.

Kansas 72.0 Monthly benefit for

Household Existing research and calculation. TANF $357.27 HHS Third Annual
Report to Congress, 2000 Medicaid 349.16 GAO calculation based on Urban
Institute,

1999 c Food Stamps 162.00 USDA Office of Analysis, Nutrition and

Evaluation, Statistics, FY1999 Individual

TANF 73.47 b HHS Third Annual Report to Congress, 2000. Medicaid 129.32
Urban Institute, 1999 c Food Stamps 71.27 GAO/ RCED- 99- 180 d

Kansas 6 Maryland 12 New York 3 Pennsylvania 12 Monthly benefit

payments avoided 3 States can participate in 4 quarterly PARIS

matches per year, thus, the number of benefit months a state could expect to
save in each program may be closer to 3 months than the average (7 months)
based on data reported by 5 states. Texas 2. 4

Appendix I: Data and Assumptions Used in Our Analysis

Appendix I: Data and Assumptions Used in Our Analysis Page 27 GAO- 01- 935
PARIS Project Can Help States

Variable Value used Basis Details Costs

Pennsylvania 150 minutes2.5 hours New York 60 minutes Maryland 30 minutes
Time to follow up or resolve one

match hit 120 minutes Because of the wide variation in the time 4 states
reported an average match hit requires

to resolve, we selected a figure closer to the longest estimate. Average
time based on data from 4 states is 67.5 minutes. Kansas 30 minutes

New York $27.20 Maryland 23.91 Pennsylvania 68.97 Kansas 7.00 Salary cost to
resolve one match 68.97 Pennsylvania?s figure is used because it

includes state personnel, and local office staff costs of PARIS follow- up.
Average based on data from 5 states is $31.77.

California 63.21 Maryland 757.00 Pennsylvania 375.00 State costs to generate
automated

list of recipients for matching $440 per match Average based on data from 3
states. Kansas 189.45

Other

Average case or household size 2.7 individuals Monthly Labor Review, May
2000 e Note: Figures are in nominal 1999 and 2000 dollars. a The match hit
rate reported by Pennsylvania is the hit rate based on ?raw? (unfiltered)
data. The hit

rate based on filtered data is likely to be higher. b Figure represents
average amount that a household?s monthly TANF benefit would change if one

person was removed from the case, based on data from HHS Third Annual Report
to Congress, 2000. C Medicaid Managed Care Payment Methods and Capitation
Rates: Results of a National Survey,

1999.

D Food Stamp Program: Households Collect Benefits for Persons Disqualified
for Intentional Program Violations (GAO/ RCED- 99- 180, July 8, 1999). e
Spending Patterns of Public- Assisted Families, Monthly Labor Review, May
2000.

Source: GAO analysis.

S= (A x F + B x I) x X

where

S = Savings per benefit case avoided, A = Proportion of match hits in which
entire case is closed, F = Family (case) monthly benefit, B = Proportion of
match hits where household members are removed from the case, Formulas Used
in Analysis

Savings

Appendix I: Data and Assumptions Used in Our Analysis Page 28 GAO- 01- 935
PARIS Project Can Help States

I = Monthly benefit for that number of individuals, and X = Months of future
benefit payments avoided.

This calculation was performed for each of the three programs (TANF, Food
Stamps, and Medicaid) separately to demonstrate how the costeffectiveness of
a ?good? PARIS match hit could change depending on the number of programs
that are included in a match.

C = (N x W) + L

where

C = Costs the state incurs for each case sent to field office staff for
followup, N = Number of hours required to work an average case, W = Average
hourly wage of individuals following up on match hits, and L = Average cost
per case that the state incurs to create the automated list of recipients
each time it participates in the PARIS matches.

R = S / C

where

R = Ratio of savings to costs, S = Savings, and C = Costs.

The savings that a state might experience from participating in PARIS could
differ from those we have reported in the report, depending on which
assumptions are used. Tables 6 and 7 illustrate the possible savings a state
could realize if we use the averages reported by each state instead of the
more conservative assumptions cited in the report. The assumptions used in
these tables, where they vary from the values used in the report, are as
follows:

 Number of benefit months avoided per valid match hit: 7 months;

 Valid hit rate: 30 percent;

 Each match hit requires 60 minutes (1 hour) to resolve;

 Cost to follow up on 1 match hit: $31.77; and

 Cost to follow up on all 100 match hits: $31.77 x 100 hits = $3,177 + $440
(cost of creating file of recipients for matching each time the PARIS match
is performed) = $3,617 total cost. Costs

Savings/ Cost Ratio Alternative Scenario

Appendix I: Data and Assumptions Used in Our Analysis Page 29 GAO- 01- 935
PARIS Project Can Help States

Using our hypothetical example in which 100 match hits are sent to local
benefit office staff for follow up, table 6 illustrates how the savings to a
state from participating in one PARIS match could vary depending on the
number of programs included in the match and differences in the valid hit
rate. The table assumes that the savings for each program accrue for 7
months.

Table 6: Estimated Net Savings Given Varying Valid Hit Rates Valid match
hits and net savings Program 10% 30% 50%

TANF only $15,630 $54,125 $92,619 Medicaid only 16,361 56,317 96,273 Food
Stamps only 5,881 24,878 43,875 TANF and Food Stamps 25,129 82,620 140,111
TANF and Medicaid 35,608 114,059 192,509 Food Stamps and Medicaid 25,859
84,812 143,765

All three programs $45,107 $142,554 $240,001

Notes: The table assumes that 7 months of benefit payments are avoided as a
result of the match. Figures are in nominal 1999 and 2000 dollars.

Source: GAO analysis.

If 30 percent of the match hits are valid (they accurately identify 30
instances of duplicate benefits being paid) and the individuals identified
are enrolled in all three programs, the match would produce gross savings of
more than $146,000 yielding a savings- to- cost ratio of about 40 to 1.
After factoring in total costs of $3, 617 to participate and follow up on
the match hits, the net savings are more than $142,000. A valid hit rate of
10 percent- a rate substantially below what participating states have
reported- in which the match only includes the Food Stamp program would
still result in gross savings of about $9,500 (a savings- to- cost ratio of
almost 3 to 1).

The number of months that future benefit payments are avoided can also
influence the amount of savings that result from a PARIS match. Table 7
illustrates the variation in program savings that could result depending on
the number of months of future benefits that are avoided and the number of
programs matched, given a 30 percent valid hit rate (30 match hits out of
the 100 match hits sent for follow up result in some savings).

Appendix I: Data and Assumptions Used in Our Analysis Page 30 GAO- 01- 935
PARIS Project Can Help States

Table 7: Estimated Net Savings Given a Varying Number of Future Benefit
Payment Months Avoided

Benefit months avoided and net savings Program 3 months 9 months 15 months
TANF only $21,129 $70,622 $120,115 Medicaid only 22,069 73,441 124,813 Food
Stamps only 8,595 33,019 57,444 TANF and Food Stamps 33,342 107,259 181,176
TANF and Medicaid 46,815 147,680 248,545 Food Stamps and Medicaid 34,281
110,077 185,874

All three $59,028 $184,317 $309,606

Notes: The table assumes a valid match hit rate of 30 percent. The average
hit rate reported by the five states is 32.4 percent. Figures are nominal
1999 and 2000 dollars.

Source: GAO analysis.

As the table shows, the state would experience net savings from
participating in PARIS under each scenario, although the range of potential
savings varies considerably. Three months? worth of Food Stamp benefits
avoided would yield gross savings of more than $12,000, (a savings- to- cost
ratio of about 3 to 1). The net savings would be about $8,600. However, the
match could produce gross savings of about $313,000 if 15 months of benefits
were avoided and the match was performed for all three programs (a savings-
to- cost ratio of about 87 to 1). Net savings would be about $309,600.

Appendix II: Comments From the Department of Health and Human Services

Page 31 GAO- 01- 935 PARIS Project Can Help States

Appendix II: Comments From the Department of Health and Human Services

Note: GAO comments supplementing those in the report text appear at the end
of this appendix.

Appendix II: Comments From the Department of Health and Human Services

Page 32 GAO- 01- 935 PARIS Project Can Help States

See GAO comment on p. 35.

Appendix II: Comments From the Department of Health and Human Services

Page 33 GAO- 01- 935 PARIS Project Can Help States

Appendix II: Comments From the Department of Health and Human Services

Page 34 GAO- 01- 935 PARIS Project Can Help States

Appendix II: Comments From the Department of Health and Human Services

Page 35 GAO- 01- 935 PARIS Project Can Help States

The following is GAO?s comment on the Department of Health and Human
Services? letter dated August 17, 2001.

The HHS comment concerning ?the stated need for a real- time, GAO online
system? is inaccurate. Although we discuss a national database as one option
for providing prompt responses to interstate inquiries about public
assistance applicants? eligibility for benefits, the report does not state
that we or any other agency should develop or operate such a system. GAO
Comment

Appendix III: Comments From the Food and Nutrition Service

Page 36 GAO- 01- 935 PARIS Project Can Help States

Appendix III: Comments From the Food and Nutrition Service

The USDA Food and Nutrition Service did not provide a formal comment letter.
However, as we agreed with Food Stamp Program officials, we are reprinting
the comments faxed to us on July 24, 2001.

Now on p. 17.

Appendix III: Comments From the Food and Nutrition Service

Page 37 GAO- 01- 935 PARIS Project Can Help States (207117)

The first copy of each GAO report is free. Additional copies of reports are
$2 each. A check or money order should be made out to the Superintendent of
Documents. VISA and MasterCard credit cards are also accepted.

Orders for 100 or more copies to be mailed to a single address are
discounted 25 percent.

Orders by mail:

U. S. General Accounting Office P. O. Box 37050 Washington, DC 20013

Orders by visiting:

Room 1100 700 4 th St., NW (corner of 4 th and G Sts. NW) Washington, DC
20013

Orders by phone:

(202) 512- 6000 fax: (202) 512- 6061 TDD (202) 512- 2537

Each day, GAO issues a list of newly available reports and testimony. To
receive facsimile copies of the daily list or any list from the past 30
days, please call (202) 512- 6000 using a touchtone phone. A recorded menu
will provide information on how to obtain these lists.

Orders by Internet

For information on how to access GAO reports on the Internet, send an email
message with ?info? in the body to:

Info@ www. gao. gov or visit GAO?s World Wide Web home page at: http:// www.
gao. gov

Contact one:

 Web site: http:// www. gao. gov/ fraudnet/ fraudnet. htm

 E- mail: fraudnet@ gao. gov

 1- 800- 424- 5454 (automated answering system) Ordering Information

To Report Fraud, Waste, and Abuse in Federal Programs
*** End of document. ***