Supplemental Security Income: Opportunities Exist for Improving Payment
Accuracy (Letter Report, 03/27/98, GAO/HEHS-98-75).

Pursuant to a congressional request, GAO conducted a follow-up review on
the feasibility of the Social Security Administration (SSA) using new
data sources on earnings and financial account information to determine
applicants' eligibility for the Supplemental Security Income program,
focusing on: (1) the extent to which overpayments occur because SSI
clients fail to disclose their earnings and financial accounts; (2)
whether SSA could obtain more current and comprehensive information to
detect undisclosed earnings; and (3) whether the agency could obtain
more current and comprehensive information on undisclosed financial
accounts.

GAO noted that: (1) unreported or underreported earnings and financial
accounts continue to result in significant overpayments in the SSI
program; (2) according to SSA's overpayment data, the failure of SSI
clients to disclose earnings and financial accounts was responsible for
approximately 40 percent of the $1.6 billion in overpayments identified
for fiscal year 1996; (3) specifically, about $379.5 million in
overpayments was the result of SSI clients not fully disclosing their
earnings, and $268.1 million was the result of clients not disclosing
financial account information; (4) more current and comprehensive
information is now available to detect undisclosed earnings; (5) SSA
detects overpayments resulting from undisclosed earnings primarily by
matching information provided by SSI clients with earnings data used in
the administration of other government programs; (6) however,
computerized matches, which are not done until individuals are on SSI's
rolls, have built-in delays in detecting overpayments that range from 6
to 21 months; (7) two databases developed for use by the Office of Child
Support Enforcement (OCSE) could provide SSA with more current and
comprehensive earnings information; (8) SSA could check these databases
prior to placing applicants on the rolls and thereby prevent
overpayments caused by applicants failing to disclose earnings at the
time of application; (9) these databases would also allow SSA to detect
occurences of undisclosed earnings to ongoing recipients within 4 to 6
months and thereby reduce the number and duration of the corresponding
overpayments; (10) opportunities for improved financial account
information also exist; (11) SSA detects undisclosed financial accounts
by conducting computer matches once a client's eligibility has been
established; (12) this match, however, can only detect undisclosed
accounts that existed 9 to 21 months before; (13) SSA could obtain
up-to-date information on the financial accounts of SSI clients from
financial institutions by accessing the nationwide telecommunication
network, which links all financial institutions; (14) such information
would help ensure that applicants whose bank accounts would make them
ineligible for the program do not gain eligibility; and (15) by
eliminating ineligible individuals at the point of application, SSA
could avoid the expense of determining medical and vocation disability
and could also reduce the number and duration of overpayments to ongoing
recipients who are overpaid because of newly acquired financial accounts
or increases in existing ones.

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

 REPORTNUM:  HEHS-98-75
     TITLE:  Supplemental Security Income: Opportunities Exist for 
             Improving Payment Accuracy
      DATE:  03/27/98
   SUBJECT:  Overpayments
             Eligibility determinations
             Claims processing
             Federal social security programs
             Data bases
             Data collection
             Social security benefits
             Computer matching
             Financial records
             Beneficiaries
IDENTIFIER:  Supplemental Security Income Program
             HHS New Hire Data Base
             HHS Quarterly Wage Data Base
             
******************************************************************
** 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, Subcommittee on Human Resources, Ways and
Means Committee, House of Representatives

March 1998

SUPPLEMENTAL SECURITY INCOME -
OPPORTUNITIES EXIST FOR IMPROVING
PAYMENT ACCURACY

GAO/HEHS-98-75

Improving SSI Payment Accuracy

(105154)


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

  ATM - automated teller machine
  DOD - Department of Defense
  DOL - Department of Labor
  IRS - Internal Revenue Service
  OCSE - Office of Child Support Enforcement
  OMB - Office of Management and Budget
  OPM - Office of Personnel Management
  PRWORA - Personal Responsibility and Work Opportunity
     Reconciliation Act of 1996
  SSA - Social Security Administration
  SSI - Supplemental Security Income
  VA - Department of Veterans Affairs

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


B-278829

March 27, 1998

The Honorable E.  Clay Shaw, Jr.
Chairman, Subcommittee on Human Resources
Committee on Ways and Means
House of Representatives

Dear Mr.  Chairman,

In 1996, the Supplemental Security Income (SSI) program paid almost
$29 billion to more than 6.6 million aged, blind, and disabled
individuals.  As administrator of this program, the Social Security
Administration (SSA) is responsible for ensuring that only those who
are eligible receive benefits and that payment amounts are correct. 
However, SSA's overpayment data show that approximately $1.6 billion
of these benefits were paid in error. 

To determine SSI eligibility and payment amounts, SSA relies on
applicants and recipients to accurately report their income and
assets.  Where possible, SSA verifies these reports with independent
information once individuals begin receiving benefits.  In a previous
report,\1 we discussed how SSA could directly access state data on
earnings and government benefits to identify clients--both applicants
and recipients--who do not disclose to SSA income that they received
from these sources.\2 Since that time, new databases have been
created to assist the Department of Health and Human Services' Office
of Child Support Enforcement (OCSE) in its work.  SSA has agreed to
house and maintain these databases, which can provide SSA more
current and comprehensive information on earnings than could be
obtained by directly accessing state data.  Technological advances
have also made it possible for SSA to obtain more current and
comprehensive information on the financial accounts of SSI clients. 

In light of these developments, you asked us to conduct a follow-up
study on the feasibility of SSA using new data sources on earnings
and obtaining better financial account information.  Specifically,
you asked us to determine (1) the extent to which overpayments occur
because SSI clients fail to disclose their earnings and financial
accounts, (2) whether SSA could obtain more current and comprehensive
earnings information to detect undisclosed earnings, and (3) whether
the agency could obtain more current and comprehensive information on
undisclosed financial accounts. 

To accomplish these objectives, we interviewed officials from SSA,
the Internal Revenue Service (IRS), the Department of the Treasury,
the Office of Management and Budget (OMB), and the Federal Reserve
and obtained relevant documentation.  We also interviewed executives
from the financial institution industry and network providers
regarding the computer network that electronically links U.S. 
financial institutions together.  We analyzed SSA data on SSI
overpayments that resulted when SSA did not have adequate financial
information on clients.  We performed our work between October 1996
and January 1998 in accordance with generally accepted government
auditing standards.  (See app.  I for more information on our scope
and methodology.)


--------------------
\1 Supplemental Security Income:  Administrative and Program Savings
Possible by Directly Accessing State Data (GAO/HEHS-96-163, Aug.  29,
1996). 

\2 Nondisclosure can take two forms.  Clients may fail to disclose to
SSA that they have any income at all, or they may fail to disclose
the full amount of their income. 


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

Unreported or underreported earnings and financial accounts continue
to result in significant overpayments in the SSI program.  According
to SSA's overpayment data, the failure of SSI clients to disclose
earnings and financial accounts was responsible for approximately 40
percent of the $1.6 billion in overpayments identified for fiscal
year 1996.  Specifically, about $379.5 million in overpayments was
the result of SSI clients not fully disclosing their earnings, and
$268.1 million was the result of clients not disclosing financial
account information. 

More current and comprehensive information is now available to detect
undisclosed earnings.  Today, SSA detects overpayments resulting from
undisclosed earnings primarily by matching information provided by
SSI clients with earnings data used in the administration of other
government programs.  However, computerized matches, which are not
done until individuals are on SSI's rolls, have built-in delays in
detecting overpayments that range from 6 to 21 months.  Two databases
developed for use by OCSE could provide SSA with more current and
comprehensive earnings information:  the New Hire Data Base and the
Quarterly Wage Data Base.  The New Hire Data Base lists all new hires
within a month of their hiring; the Quarterly Wage Data Base--which
will soon be operational--will contain quarterly earnings information
on nearly all employees.\3

SSA could check these databases prior to placing applicants on the
rolls and thereby prevent overpayments caused by applicants failing
to disclose earnings at the time of application.  These databases
would also allow SSA to detect occurrences of undisclosed earnings to
ongoing recipients within 4 to 6 months and thereby reduce the number
and duration of the corresponding overpayments.  SSA officials
acknowledged that these databases would provide field staff with more
current and comprehensive earnings information.  However, at the time
of this report, the agency had put minimal effort into incorporating
the use of these databases into the claims handling process, citing
competing priorities as the reason why they had not done more. 

Opportunities for improved financial account information also exist. 
Currently, SSA detects undisclosed financial accounts by conducting
computer matches once a client's eligibility has been established. 
This match, however, can only detect undisclosed accounts that
existed 9 to 21 months before.  SSA could obtain up-to-date
information on the financial accounts of SSI clients from financial
institutions by accessing the nationwide telecommunication network,
which links all financial institutions.  Such information would help
ensure that applicants whose bank accounts would make them ineligible
for the program do not gain eligibility.  By eliminating ineligible
individuals at the point of application, SSA could avoid the expense
of determining medical and vocational disability and other processing
costs for invalid claims.  The agency could also reduce the number
and duration of overpayments to ongoing recipients who are overpaid
because of newly acquired financial accounts or increases in existing
ones. 


--------------------
\3 States vary in who is required to report quarterly earnings
information.  For example, the earnings of self-employed individuals
and those of agricultural and domestic workers are not reported in
many states.  Their earnings may be reported in the annual IRS form
W-2, however. 


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

SSI provides monthly cash benefits to qualified aged, blind, and
disabled persons.  Because it is a program based on need, monthly
changes in the amount of non-SSI income that clients receive can
increase or decrease the amount of SSI benefits to which they are
entitled or make them completely ineligible for benefits.  Resources,
including financial accounts, that exceed $2,000 for an individual or
$3,000 for a couple make that individual or couple ineligible for the
program.  To minimize occurrences of over- and underpayments, the
program requires clients to promptly report to SSA any fluctuations
in their income or assets. 

As part of the application process, SSI clients are required to
disclose all of their income and resources to SSA field staff who
process their applications.  SSA policy requires that field staff
obtain documentation to verify the amount of income and resources
that applicants report.  It does not require, however, that field
staff check for unreported income and resources unless they suspect
that applicants are not fully disclosing them.  Thus, at the time of
application, SSA normally relies on applicants to portray their
financial situations accurately. 

To ensure that newly eligible recipients have accurately portrayed
their financial condition and that ongoing recipients continue to do
so, SSA uses both financial eligibility reviews, known as
redeterminations, and computer matching to verify income and resource
levels.  During redeterminations, recipients report their income on
mail-in questionnaires or in face-to-face or telephone interviews. 
The method used to contact the client and the frequency of such
contacts depend on the likelihood that a client's financial situation
will change.  Computer matches, which compare the individual's SSI
record against data obtained from federal and state agencies, enable
SSA to detect some types of income and resources that clients have
not reported. 

The computer matching process to detect undisclosed income compares
earnings income reported by clients to the earnings information
contained on IRS form W-2s, which employers must file annually with
SSA.  SSA conducts this match annually.\4 The W-2 match is
supplemented twice a year with quarterly earnings information
provided by 45 states and the District of Columbia.\5 To do this, SSA
sends computer tapes or cartridges containing the names of SSI
recipients to each state.  The states in turn append to the bottom of
these tapes any earnings information pertaining to the SSI recipients
residing in their states and then mail the tapes back to SSA.  Once
SSA receives the tapes, it matches them against the agency's own
records to determine if recipients have disclosed all of their
earnings to the agency. 

In order to detect unreported financial accounts, information
reported by clients is compared to IRS form 1099s, which are filed
annually by financial institutions and contain the amount of interest
earned on financial accounts.  Because form 1099 data only contain
interest accrued on financial accounts, this match can detect only
interest-bearing accounts.  Each September, SSA conducts this match
using data from the previous year, which covers most SSI recipients
from that year. 


--------------------
\4 SSA also conducts one make-up match using the same year's W-2
information to detect unreported earnings of recipients who were not
in pay status at the time of the first match. 

\5 The remaining five states have chosen not to participate in the
state earnings match for various reasons, including the belief that
it would be too time-consuming to do so. 


   NONDISCLOSURE CAUSES MILLIONS
   OF DOLLARS IN OVERPAYMENTS
------------------------------------------------------------ Letter :3

A primary cause of SSI overpayments has been that clients do not
always disclose their earnings and financial accounts when they apply
for benefits or once they are receiving such payments.  For example,
SSA's fiscal year 1996 payment accuracy study shows that out of a
total of $1.6 billion in overpayments, approximately 40 percent
(nearly $647.6 million) was the result of nondisclosed earnings and
financial accounts.\6 About $379.5 million of these overpayments
occurred because SSI clients did not disclose their earnings and
$268.1 million occurred because SSI clients did not disclose their
financial accounts.  Many of these overpayments could have been
prevented or more quickly detected if more timely and comprehensive
information on the earnings and financial accounts of SSI clients had
been available. 

SSA conducts a second annual payment accuracy study, which contains
more detailed information on the amount of overpayments in the SSI
program.  However, neither study accurately estimates for the entire
SSI population the amount of overpayments made because of
nondisclosure at the time of application versus the amount made
because of nondisclosure after clients began receiving benefits. 
Regardless, both studies have consistently shown over the years that
hundreds of millions of dollars in overpayments occur at both of
these junctures.  Such findings, in turn, indicate that the agency
needs to address systemwide weaknesses in both the application and
post-entitlement procedures it uses to determine program eligibility
and payment amount. 

Of the hundreds of millions of dollars in overpayments that have been
made, SSA has gotten little of it back.  SSA statistics show that, on
average, the agency recovers only about 15 percent of all outstanding
overpayments.  The older the overpayment, the more difficult it is to
recover.  Moreover, when an individual is removed from SSI's
rolls--which can happen when an overpayment is the result of a
nondisclosed financial account--the overpayment will probably never
be recovered because the individual no longer receives a monthly SSI
benefit payment from which SSA can withhold funds.  SSA's overpayment
recovery rate is low partly because SSI recipients are poor and do
not have the funds to repay this debt. 


--------------------
\6 The other 60 percent were caused by such things as undisclosed
state or federal government benefits; resources such as non-home real
property and life insurance; and food, clothing, or shelter provided
to recipients free of charge. 


   MORE CURRENT AND COMPREHENSIVE
   INFORMATION ON EARNINGS IS
   AVAILABLE
------------------------------------------------------------ Letter :4

SSA's present data sources and procedures for detecting undisclosed
earnings do not provide up-to-date and comprehensive information on
the earnings of applicants and recipients.  Such information is
critical because earnings are a primary factor in determining both
initial program eligibility and the amount of benefits recipients
should receive each month.  SSA could obtain such information by
using new data sources on earnings and by enhancing its current
computer-matching procedures. 


      SSA'S PRESENT APPROACH TO
      DETECTING UNDISCLOSED
      EARNINGS RELIES ON OLD DATA
---------------------------------------------------------- Letter :4.1

SSA uses data that are outdated and do not reflect the current
earnings status of SSI clients.  When individuals apply for SSI
benefits, SSA field staff are required to check the agency's database
that contains IRS form W-2 information to verify that applicants have
accurately portrayed their work histories.  How current this
information is depends on when SSA enters an applicant's W-2 data and
when the applicant comes in to apply for SSI benefits.  For example,
SSA began entering 1996 W-2 earnings information into its database in
February 1997, shortly after it was reported by employers.  By April,
SSA had entered about 45 percent of the 1996 W-2s into its database,
and by September, the agency had entered 98.5 percent of the earnings
information.  Thus, in April 1997, there was about a 45-percent
chance that the 1996 earnings of an SSI applicant would be recorded
in SSA's database.  Earnings from December 1996 would be 4 months
old, and earnings from January 1996 would be 16 months old.  If an
application were made in April 1998 and the individual's 1997 W-2
information had not yet been entered, the only earnings information
available to SSA field staff would be 1996 W-2 information, which
would then be 15 to 27 months out of date. 

After clients have begun receiving SSI payments, SSA checks for
undisclosed earnings in two computer matches:  a semiannual match
that uses quarterly earnings data that employers file with the
states, and another that uses the annual W-2 information.\7 Because
of the age of the data used in the match, these matches can detect
only undisclosed earnings that were received 6 to 21 months in the
past.  For example, state quarterly data sent to SSA in March 1998
covers earnings through the quarter ending September 1997.  If a
client had earnings as of September 1997, the March 1998 match would
only detect those earnings 6 months after they were received. 
Similarly, if the client's last earnings occurred as early as July
1997, the March 1998 match would not have detected them until 9
months after they were earned.  Because state earnings information is
not provided by all states, SSA supplements this match by conducting
a computer match each September using W-2 information from the
previous year.  Thus, if a recipient had earnings in December of the
previous year, they would not be discovered for 9 months, and if a
recipient had earnings in January of the previous year, this match
would not detect them for 21 months. 

Although the state computer matches provide more current information
than the W-2 data match, state data are not always as comprehensive
as the W-2 data.  First, states provide earnings information to SSA
only for current SSI recipients, so this information cannot be used
to verify the earnings reported by new SSI applicants.  Also, only 45
states and the District of Columbia have agreed to provide SSA with
this information. 

Finally, the process of conducting the match can be unwieldy, and SSA
is often not able to complete the match for all participating states. 
To perform matches, SSA must prepare and send computer tapes or
cartridges containing the names of SSI recipients in each state to
all participating states.  The states in turn provide earnings
information, if any, on these recipients to SSA.  Often, however, the
tapes get lost or damaged in the mail, or the state prepares the data
in a format that SSA's computers cannot read.  For example, in the
first half of 1997, SSA was able to complete the match for only 37 of
the 45 states that had agreed to provide the agency with this
information.  The process of SSA and the states exchanging computer
tapes is so cumbersome that even though the states have new
information four times a year, SSA only attempts to get it twice a
year. 


--------------------
\7 SSA also conducts make-up matches using both the IRS W-2 data and
the state data to detect undisclosed earnings of recipients who were
not included in the primary match for various reasons. 


      BETTER SOURCES OF EARNINGS
      INFORMATION EXIST
---------------------------------------------------------- Letter :4.2

New data sources exist that could help improve earnings verification. 
The Personal Responsibility and Work Opportunity Reconciliation Act
of 1996 (PRWORA) requires that states report the names of newly hired
employees as well as all of the quarterly earnings information
reported for individuals working in their states to OCSE.  OCSE uses
this information to identify those parents who could make child
support payments.  SSA, which helped OCSE develop these databases, is
responsible for housing and maintaining them at its National Computer
Center and has authority under PRWORA to use the information
contained in them.  One of these data sources, the New Hire Data
Base, identifies newly hired employees and their employers within a
month of their hiring.  The second database, the Quarterly Wage Data
Base, will offer quarterly earnings information that is between 4 and
6 months old.  The quarterly information that the states will submit
for OCSE's use is the same data that the majority of states now
submit to SSA via computer tapes or cartridges.  However, the
national OCSE database will be a more comprehensive and current
information source because all states are required to participate; it
will also contain all employees--not just those on the SSI rolls at
the time of the match.  It will allow SSA to check for earnings
anywhere in the country for both applicants and recipients, and SSA
will receive data quarterly instead of only biannually. 

The New Hire Data Base has been operating since October 1997, and as
of mid-February 1998, it has received 13.7 million new-hire records,
with all states except one transmitting this information
electronically over SSA's dedicated, secure network.\8 State
employees are already submitting queries to their state directories
of new hires and are finding that this has helped them to accurately
calculate eligibility and payment amounts for state programs.  If SSA
sets up its own queries to the New Hire Data Base, the agency could
use the improved information to reduce the number of overpayments
resulting from the nondisclosure of earnings.  States began
submitting earnings to the Quarterly Wage Data Base on February 1,
1998. 

It may be possible to set up both of these databases to receive and
respond to requests for information so that SSA field staff could
check for undisclosed earnings at the time of application.  This
would prevent many overpayments that are the result of nondisclosure
at the application stage, since these databases will contain earnings
information that is between 1 and 6 months old.  Such data are often
current enough to contain earnings information for the same time
period in which benefits are received by many newly eligible
recipients.  This is because it takes, on average, more than 3-1/2
months for a decision on an SSI disability claim, and newly eligible
recipients receive benefits retroactively back to the date when they
first applied. 

SSA could also reduce the number and duration of earnings
overpayments to ongoing recipients by using the new earnings data in
a more fully automated computer matching process.  According to
officials from SSA and OCSE, the agency could develop an electronic
interface with the New Hire Data Base that retains a continually
updated list of SSI recipients and notifies SSA automatically
whenever a new hire record is reported for one of those recipients. 
SSA field staff could then contact the employer listed in this
database to verify the applicant's employment and the amount of his
or her earnings.  In addition, quarterly matching of the SSI rolls
against the much larger Quarterly Wage Data Base could be used to
detect undisclosed earnings.  An automated match could be done at
SSA's National Computer Center, eliminating the need for SSA and the
states to exchange computer tapes. 

Under both the current computer matching procedures and these new
procedures, field offices would still investigate undisclosed
earnings before reducing a recipient's benefits, declaring an
overpayment, or both.  Recipients, therefore, would still have an
opportunity to contest earnings that may not belong to them or that
fall under program rules permitting the exclusion of certain income. 


--------------------
\8 New Jersey is not required to submit new hire information until
April 1, 1998.  At that time this state will also transmit new hire
information to OCSE electronically over SSA's dedicated, secure
network. 


      SSA IS EXPLORING OTHER DATA
      SOURCES FOR EARNINGS
---------------------------------------------------------- Letter :4.3

SSA policy officials acknowledge that more current and comprehensive
sources of earnings data exist.  According to these officials, SSA is
focusing its efforts on developing a comprehensive policy that
details which new data sources are the best to use in all of its
programs and how they can be used most effectively.  Even though the
New Hire Data Base is available for immediate use and the Quarterly
Wage Data Base will be available in April 1998, SSA is putting
minimal effort into incorporating these two databases into its claims
handling processes because it will take 1 year for the OCSE databases
to contain enough earnings information to be useful for title II
programs.\9 However, OCSE databases would be immediately useful to
reduce SSI overpayments.  In fiscal year 1996, more than twice as
many overpayments were made in the SSI program as in the social
security retirement program, even though SSI payments were only about
one-tenth the size of the retirement program payments.  By not using
the OCSE databases at this time, serious problems with SSI payment
accuracy may continue. 

In the interim, SSA is focusing on developing access to two
alternative data sources:  a Department of Labor (DOL) network of
states' earnings databases and state-agency maintained databases. 
The DOL network allows government employees in one state to check
on-line for earnings in the earnings databases maintained by any or
all of the other states in the network.  At the time of our review,
33 states' earnings databases were linked to the network, and there
are plans to add the earnings databases of 7 other states to this
network in the near future.  SSA is also pursuing direct access to
state agency databases on earnings; government benefits; and vital
statistics information on births, deaths, and marriages.  In an
earlier report, we recommended that SSA pursue direct access to state
data to improve SSI payment accuracy and program administration.\10
While we continue to recommend direct access to state data as the
best approach for obtaining information such as state welfare
payments and vital statistics information, especially when national
databases do not exist, we consider the OCSE databases better sources
for earnings information. 

SSA's approach to obtaining earnings information from the DOL network
and states has several shortcomings.  First, the agency must
negotiate and thereafter renegotiate separate data-sharing agreements
with each state.\11 According to SSA officials, these tasks are both
difficult and time-consuming.  In the last several years, SSA has
been actively seeking data-sharing agreements with states, but as of
November 1997, only nine had agreed to provide direct access to their
earnings data. 

Second, these alternative sources will not necessarily provide SSA
with nationwide earnings information, which is essential for
detecting the undisclosed earnings of clients who work in one or more
states and apply for benefits in yet another.  States are not
required to participate in the DOL network or to grant SSA employees
direct access to their data.  Therefore, neither of these two
alternative data sources ensures nationwide coverage.  Further, of
the nine states that have granted direct access to SSA field staff,
none have done so for SSA staff located in other states.  In
contrast, all states are required by law to provide data to the New
Hire Data Base and the Quarterly Wage Data Base.  Moreover, because
PRWORA gives SSA the authority to use these two databases in the
administration of its programs, negotiation and renegotiation of
data-sharing agreements with the states will not be necessary. 

Third, and perhaps most significant, neither the DOL network nor
direct access to state earnings data will provide employment
information as current as that in the New Hire Data Base.  The New
Hire Data Base will allow SSA to determine within a period of weeks
that an SSI recipient has taken a job, instead of waiting a minimum
of 4 months, which generally is the delay for data obtained through
on-line access or the DOL network. 


--------------------
\9 In addition to the SSI program, the agency also administers the
old age survivors and disability insurance programs under title II of
the Social Security Act. 

\10 Supplemental Security Income:  Administrative and Program Savings
Possible by Directly Accessing State Data (GAO/HEHS-96-163, Aug.  29,
1996). 

\11 The Privacy Act states that a data-sharing agreement may not
exceed 30 months:  18 months for the original negotiated agreement
and a subsequent negotiated 12-month extension.  After this time, an
entirely new agreement must be negotiated.  Moreover, if the
extension deadline is not met, the agreement lapses and a new one
must be negotiated. 


   MORE CURRENT AND COMPREHENSIVE
   INFORMATION ON FINANCIAL
   ACCOUNTS MAY BE AVAILABLE
------------------------------------------------------------ Letter :5

As is the case with earnings, SSA's present data sources and
procedures for detecting undisclosed financial accounts do not
provide up-to-date and comprehensive information on the accounts of
applicants and recipients.  Because undisclosed financial accounts
are a major source of overpayments, obtaining such information is
critical to ensuring program integrity.  Detection of such accounts,
both at the application stage and once recipients are on SSI's rolls,
would prevent many of these overpayments and reduce the number and
duration of others.  Such detection may be possible because it is now
technologically feasible for SSA to electronically obtain account
information directly from the financial industry. 


      SSA'S PRESENT APPROACH TO
      DETECTING UNDISCLOSED
      FINANCIAL ACCOUNTS RELIES ON
      OLD DATA
---------------------------------------------------------- Letter :5.1

SSA's current approach to identifying financial accounts can result
in ineligible individuals getting on SSI's rolls and remaining there
for long periods of time.  During the application process, SSA policy
requires that field staff contact banks to verify the amount of money
in the accounts of applicants who state that their accounts exceed
$1,250.  However, when applicants state that they either do not have
accounts or that their accounts are below the $1,250 threshold, such
verification is generally not required.  Once applicants are placed
on SSI's rolls, SSA checks for both unreported and underreported
financial accounts through a computer match using IRS form 1099 data. 

Computer matches using IRS form 1099 data can take months or even
years to detect unreported bank accounts.  These matches compare the
financial account information reported to the IRS by financial
institutions with the information concerning financial accounts
reported by SSI recipients.  SSA conducts this computer match every
September.\12 However, at the time of this match, the IRS 1099 data
are between 9 and 21 months old.  For example, tax year 1997 data
will not be available for use in these matches until September 1998. 
Thus, if an SSI recipient acquired an account in December 1997 that
caused the recipient's assets to exceed SSI's resource limit, SSA
would not be able to detect it for at least 9 months.  If the account
was acquired in January 1997, the recipient could have received
monthly SSI payments to which he or she was not entitled for 21
months before the IRS 1099 match could detect the overpayment.  In
addition, if the account did not earn interest, this match would not
detect it at all, since 1099 data only pertain to interest-bearing
accounts. 

SSA field staff are required to verify the amount of financial
accounts over $1,250 that applicants and recipients disclose as well
as those that are detected through the IRS 1099 computer match.  This
is done by submitting to the designated financial institution a paper
request for verification of the account balance for all months during
which the individual was receiving SSI payments.  SSA submits about 1
million requests to financial institutions each year.  Financial
institution staff, in turn, manually search their records and mail a
response, along with an invoice for this service, back to the
requesting SSA field staff.  Because this is very time-consuming for
the financial institutions, they may charge up to $25, and some may
not respond at all. 


--------------------
\12 SSA also uses the same tax year's 1099 data to conduct two
additional matches to detect any undisclosed accounts of current or
previous SSI recipients who were excluded from the primary match in
September for various reasons. 


      BETTER SOURCES OF
      INFORMATION EXIST ON
      FINANCIAL ACCOUNTS
---------------------------------------------------------- Letter :5.2

The telecommunication network linking the financial industry together
nationwide allows financial institutions to transfer funds among
themselves and provide customer services such as automated bill
payment and automated teller machines (ATM).  SSA first began to use
these networks to deposit benefit payments directly into the accounts
of SSI recipients.  Over the past few years, the agency has expanded
its use of these networks to more fully automate direct deposits. 
For example, these networks are now used to notify a specific
financial institution of the death of a customer who had a
direct-deposit account for SSA program benefits.  These networks are
also used to set up direct-deposit accounts automatically for newly
eligible recipients of SSA program benefits, eliminating the need for
the recipient to contact the financial institution to set up an
account. 

According to various experts--such as officials from SSA and the
Department of the Treasury, financial industry executives, and
network providers--it may be possible to further expand the use of
these networks to enable SSA to contact financial institutions
electronically and determine whether SSI clients have accounts that
they have not disclosed as well as verify the amount in accounts that
clients have disclosed.  Detecting undisclosed accounts when
individuals apply for SSI benefits would prevent ineligible
applicants from being placed on SSI's rolls.  In addition to
preventing significant overpayments, the agency would save the costs
associated with processing invalid claims and determining medical and
vocational disability for ineligible applicants.  According to
financial institution officials we spoke with, handling requests
electronically would also be less costly and easier for them than the
current paper-based system and would provide the information to SSA
more quickly than the current system allows. 

Many of these experts also pointed out that financial industry
networks could be used to verify account information for both SSI
applicants and recipients.  To identify accounts undisclosed by an
applicant, SSA field staff could submit a query to financial
institutions with the name, social security number, and other
identifying information of the applicant over one or more of the
networks.  These institutions could then electronically provide
applicant account information, including balances, if any.  To
identify current SSI recipients who have failed to disclose accounts,
SSA could use the networks to periodically transmit a file of current
SSI recipients to financial institutions selected according to
criteria specified in computer profiles.  Most financial
institutions' computer systems have the capability to automatically
check their files on account holders to see if there are any matches
with the SSI recipient list.  If matches are found, the system would
send an electronic response to SSA containing the recipients' names
and account balances. 

Financial industry data would be much more current than the data used
in the IRS form 1099 match because financial institutions maintain
up-to-date records of their customers' accounts.  This does not mean
that an overpayment could be detected as soon as an undisclosed
account came into existence because the earliest point of detection
would depend on how frequently SSA conducted matches using these
data.  It does mean, however, that SSA, working with the financial
institution industry, could design a system that optimally balanced
how frequently undisclosed accounts were detected with the
cost-effectiveness of such a procedure.  It also means that SSA could
identify undisclosed accounts much earlier than it currently does and
thereby prevent many overpayments made as a result of nondisclosure. 

For SSA to detect undisclosed accounts most effectively, every
applicant and recipient would have to be checked against the records
of every financial institution in the country.  The extent to which
complete coverage could be achieved would depend upon technological
capabilities.  According to executives who manage the financial
industry networks, current technology is sufficient to permit very
broad-based checks for applicants, with minimal cost and effort. 

Financial institutions and human services departments in some states
are already exploring ways in which technology can more efficiently
provide the required information on the financial accounts of welfare
clients.  This is occurring in part because PRWORA requires financial
institutions to report this information to the states for child
support enforcement purposes.  In its 1997 business plan, SSA
acknowledged that it intends to look into expanding its use of these
networks to check for undisclosed accounts, but the agency has yet to
put together a proposal detailing when and how it will undertake such
a study. 

SSA already has a telecommunication link to the financial industry
network and routinely uses that network to transmit information to
financial institutions.  However, programming would be needed for SSA
to transmit requests for information and for financial institutions
to notify the agency that it has an account holder who is an SSI
applicant or recipient.  SSA officials and executives from the
financial industry with whom we spoke agreed that using the financial
institution network to verify financial accounts is technically
feasible but would require effort to implement.  Because financial
institutions use various types of computer operating systems and
software, each institution would have to create, test, and implement
programming specific to its system. 


      PRIVACY AND SECURITY
      CONCERNS MUST BE ADDRESSED
      BEFORE NEW ACCESS IS
      IMPLEMENTED
---------------------------------------------------------- Letter :5.3

States and the financial industry share concerns about privacy and
security.  Privacy concerns center around ensuring that personal
information provided by an individual to a government agency or
private institution is protected from being disclosed to those who do
not have a legal right to it.  Concerns about security center around
having adequate computer security controls to ensure physical
security and prevent inappropriate access. 

SSA is required by law to take certain steps to ensure the privacy
and security of data, whether that information is internal to SSA or
is shared with other entities.  These steps include developing a
security plan, audit trails, automated alerts to prevent
inappropriate requests for personal information, personal
identification numbers and passwords, training, disaster recovery
plans, and periodic internal and external evaluations of all privacy
and security measures.  An assessment of whether to institute
additional measures may also be needed. 


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

The two OCSE earnings databases, as well as data from the financial
institution industry, would provide SSA with information needed to
prevent or reduce overpayments resulting from undisclosed earnings
and financial accounts.  This information would be particularly
valuable in processing applications because, for the first time, the
agency would be able to verify with more current and comprehensive
information the financial allegations of applicants before initiating
payments to them.  Preventing overpayments or detecting them more
quickly would bolster the integrity of the SSI program by more
effectively ensuring that clients are receiving only those benefits
to which they are entitled.  We estimated that approximately $647.6
million of the overpayments that occurred in fiscal year 1996 could
have been avoided or more quickly detected if these data had been
available for SSA to use both in the application process and at
intervals after clients were on SSI's rolls. 

SSA has authority to use the OCSE databases for the administration of
its programs and is responsible for housing and maintaining them at
its National Computer Center.  The agency has not, however, directed
adequate resources to developing computerized interfaces so that
these data could be used in the SSI program.  The agency also has
authority to verify information on the financial accounts of clients
from the financial industry but has not yet investigated the
technical and economic feasibility of obtaining this information via
computer to make it an effective verification tool.  Such a system
may be economically feasible, even though it would result in SSA
verifying more financial accounts than they currently do.  According
to financial industry experts, computerized verification requests
would cost much less than the financial institutions' current charge
for such requests--which can be as much as $25 per request. 
Moreover, if SSA were able to obtain financial account information
free of charge, as is the case for most states, this system would be
even more cost-effective. 


   RECOMMENDATIONS
------------------------------------------------------------ Letter :7

We recommend that the Commissioner of SSA take the following actions: 

  -- Develop computerized interfaces necessary to access OCSE's New
     Hire Data Base and Quarterly Wage Data Base, and use them in
     accordance with applicable security and privacy laws and
     regulations to detect undisclosed earnings during initial and
     subsequent determinations of eligibility for the SSI program. 

  -- Study the feasibility of obtaining computerized information from
     financial institutions to detect financial accounts that SSI
     clients do not report during the application process and during
     subsequent determinations of eligibility.  Such a study should
     include a comparison of the cost of obtaining and using such
     information and the program savings achievable as a result of
     that use.  Security and confidentiality issues should also be
     addressed. 


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

In commenting on a draft of this report, SSA agreed that the two OCSE
databases can be useful tools in reducing SSI overpayments and stated
that they intend to begin using them by October 1, 1998.  The agency
objected, however, to our characterization that it is putting minimal
effort into incorporating these databases into the verification
process.  At the time of our review, SSA was actively developing
access to only one of these databases and only doing so to detect the
undisclosed earnings of recipients once they are placed on SSI's
rolls.  Yet, overpayment prevention is equally or more important than
overpayment detection because only a small fraction of overpayments
that are made are recovered.  Field staff could use these databases
to prevent overpayments by checking for undisclosed earnings at the
time of application.  This requires that the agency develop the
necessary computer interfaces between SSA field offices and these
databases.  At the time of our work, the agency had not begun
developing these interfaces and did not appear to have any concrete
plans to do so. 

SSA also agrees with our recommendation to study the feasibility of
using information from financial institutions to detect undisclosed
financial accounts.  The agency plans to undertake such a study and
issue its first status report no later than September 1998.  SSA's
other comments to this report were incorporated where appropriate. 
The agency's comments are contained in appendix II. 

We are sending copies of this report to relevant congressional
committees, the Commissioner of Social Security, and other interested
parties. 


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

If you have any questions about this report, please contact me on
(202) 512-7215 or Roland Miller III, Assistant Director, on (202)
512-7246.  Other major contributors to this report were Nancy
Cosentino, Senior Evaluator, and Jill Yost, Evaluator. 

Sincerely yours,

Barbara D.  Bovbjerg
Associate Director
Income Security Issues


SCOPE AND METHODOLOGY
=========================================================== Appendix I

Originally, the requester of this work, Congressman E.  Clay Shaw,
Jr., asked that GAO investigate (1) the type of data that SSA now
gets from federal agencies to identify SSI overpayments, (2) whether
federal agencies have additional computerized information on the
income of SSI clients that SSA is not receiving but would find
helpful in reducing overpayments, and (3) whether direct access to
the income data maintained by federal agencies is technically and
fiscally feasible and would reduce overpayments.  The agencies we
examined were (1) the IRS, which provides form 1099 information to
detect undisclosed financial accounts; (2) the Office of Personnel
Management (OPM), which provides information to detect undisclosed
federal pensions; (3) the Department of Veterans Affairs (VA), which
provides information to detect VA compensation and pensions; and (4)
the Department of Defense (DOD), which provides information to detect
income from military pensions, military housing, and other
incidentals. 

The report details what we discovered about the manner in which SSA
receives IRS form 1099 data and how obtaining these data
electronically from the financial industry could prevent or reduce
overpayments caused by undisclosed financial accounts.  However, we
found that the amounts of SSI overpayments that resulted from
earnings and assets that clients receive from OPM, VA, and DOD either
were not large enough to warrant a detailed study into how SSA could
obtain information from these agencies more quickly or were obtained
in such a way as to allow the detection of overpayments within 1 to 2
months after the time they were incurred. 

In the course of our work, we also discovered that a new data source
that could prevent or reduce overpayments caused by undisclosed
earnings would soon be available.  Given that the nondisclosure of
earnings and financial accounts, unlike federal benefits, are major
sources of SSI overpayments, we asked the requester whether he would
like to change the objectives of the study.  He responded in the
affirmative, stating that he would like us to examine (1) the extent
to which overpayments occur because SSI applicants and recipients
fail to disclose their earnings and financial accounts, (2) whether
SSA could obtain more current and comprehensive information than it
does now to detect the nondisclosure of earnings, and (3) whether the
agency could also obtain more current and comprehensive information
on financial accounts. 

We interviewed executives from four banks and four financial industry
network providers.  We also interviewed officials from SSA, OMB, IRS,
OCSE, the Federal Reserve, and the Department of the Treasury and
obtained relevant documentation.  From these interviews, we
ascertained the feasibility of using financial industry data to
verify bank account information supplied by SSI clients and how it
could be done.  We also interviewed government officials to determine
to what extent the earnings information from the OCSE databases would
be more current and comprehensive than the data presently used by SSA
to verify earnings information reported by SSI clients.  We examined
(1) the comparative value of these new data sources versus the data
sources currently used by SSA, (2) how SSA currently verifies
client-supplied information on earnings and financial accounts, (3)
how the new data sources could be most effectively used for
verification purposes, and (4) the issues involved in implementing
the use of the new data sources.  Finally, we obtained nationwide
aggregate data from SSA studies on the amount of overpayments that
occurred in fiscal year 1996.  We used these data to determine the
amount of overpayments attributable to the nondisclosure of earnings
and financial accounts. 




(See figure in printed edition.)Appendix II
COMMENTS FROM THE SOCIAL SECURITY
ADMINISTRATION
=========================================================== Appendix I



(See figure in printed edition.)



(See figure in printed edition.)


*** End of document. ***