Supplemental Security Income: Administrative and Program Savings Possible
by Directly Accessing State Data (Letter Report, 08/29/96,
GAO/HEHS-96-163).

Pursuant to a congressional request, GAO reviewed the Social Security
Administration's (SSA) use of online access to state databases on income
information to determine whether it could: (1) improve the
administration of the Supplemental Security Income (SSI) Program; (2)
reduce overpayments; and (3) easily implement online access nationwide
in SSA field offices.

GAO found that the use of online access to state-maintained income data
could: (1) improve the administration of the SSI program by cutting the
time needed to verify client information; (2) prevent or detect
overpayments due to underreported and unreported benefit income; (3)
replace the current computer-matching system that relies on old data;
(4) enable necessary security measures to protect client
confidentiality; and (5) be inexpensively implemented nationwide with
minimal programming.

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

 REPORTNUM:  HEHS-96-163
     TITLE:  Supplemental Security Income: Administrative and Program 
             Savings Possible by Directly Accessing State Data
      DATE:  08/29/96
   SUBJECT:  Overpayments
             Claims processing
             Income maintenance programs
             Systems compatibility
             Administrative costs
             Eligibility determinations
             Computerized information systems
             Federal/state relations
             Computer security
IDENTIFIER:  AFDC
             Aid to Families with Dependent Children Program
             Supplemental Security Income Program
             Unemployment Insurance Program
             
******************************************************************
** 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 Congressional Requesters

August 1996

SUPPLEMENTAL SECURITY INCOME -
ADMINISTRATIVE AND PROGRAM SAVINGS
POSSIBLE BY DIRECTLY ACCESSING
STATE DATA

GAO/HEHS-96-163

Accessing State Data

(105145)


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

  AFDC - Aid to Families With Dependent Children program
  FTMS - File Transfer Management System
  IBON - income based on need
  IDA - Index of Dollar Accuracy data
  SSA - Social Security Administration
  SSI - Supplemental Security Income program
  UI - Unemployment Insurance program
  WC - Workers' Compensation program

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


B-271177

August 29, 1996

The Honorable Nancy L.  Johnson
Chairman, Subcommittee on Oversight
Committee on Ways and Means
House of Representatives

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

The rapid growth in the number of individuals receiving Supplemental
Security Income (SSI) payments and the related increase in
administrative and programmatic costs continue to be a concern of the
Congress.  The SSI program provides cash benefits to aged, blind, and
disabled individuals whose income and resources are below a certain
level.  In 1995, over 6 million SSI clients received nearly $25
billion in federal benefits and $3 billion in state benefits, while
SSI administrative costs totaled over $1.7 billion. 

The Social Security Administration (SSA) administers the SSI program
and is responsible for ensuring that only those who are eligible for
SSI receive benefits and that payment amounts are correct.  To
fulfill these responsibilities, SSA needs the most accurate and
timely information possible when making these decisions.  SSA is
currently testing the use of computer access to state-collected and
state-maintained income information to determine whether this access
can reduce administrative costs and payment errors.  Such access
consists of direct--commonly referred to as on-line--connections
between SSA's computers and the databases maintained by state
agencies.\1 You asked us to assess whether this access could (1)
improve the administration of the SSI program, (2) reduce
overpayments, and (3) be easily implemented nationwide in SSA field
offices where the public applies for SSI. 

To accomplish these objectives, we visited three states where SSA
field offices were testing on-line access to state data and
interviewed officials at SSA headquarters and selected regional and
field offices.  We also analyzed SSI data on overpayments caused when
SSA did not have adequate income information on applicants or
recipients (hereafter referred to as clients).  We did our work
between July 1995 and July 1996 in accordance with generally accepted
government auditing standards.  (See app.  I for more information on
our scope and methodology.)


--------------------
\1 The state agencies are employment and welfare departments that
maintain electronic data on earnings and on benefits from the Aid to
Families With Dependent Children (AFDC), Workers' Compensation (WC),
and Unemployment Insurance (UI) programs. 


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

The SSI program could be administered more efficiently and, more
importantly, millions of dollars in overpayments could be prevented
or more quickly detected if information needed for claims processing
was available immediately on-line during initial and subsequent
assessments of eligibility. 

We estimate that direct on-line access to state computerized income
information could have prevented or more quickly detected about
$131.3 million in overpayments caused by unreported or underreported
income nationwide in one 12-month period.  However, in SSA field
offices where such direct access to computerized state information
has been implemented, SSA representatives did not use it for
overpayment detection.  These claims representatives did use it,
however, to process claims more efficiently, and SSA's preliminary
results have shown that its use has reduced administrative costs. 

According to SSA officials, they did not experience technical
difficulties establishing on-line access between SSA field offices
and state agency databases in participating states and they
considered the costs incurred to be minimal.  Although the ease and
cost of implementing on-line access will vary with the degree to
which a state has automated its data files, only minimal computer
programming would be required in most states, with some states
needing additional hardware such as computer lines. 


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

The SSI program is administered by SSA.  Claims representatives at
SSA field offices are responsible for processing all applications and
determining the amount of monthly SSI payments.  SSI is a program
based on need.  The maximum monthly SSI benefit in 1996 was $470. 
Clients receive less than the maximum or become ineligible for the
program altogether if their earned and unearned income and resources
exceed certain thresholds.  Monthly changes in the amount of non-SSI
income that clients receive increase or decrease the amount of SSI
benefits to which they are entitled.  If clients do not report these
income fluctuations promptly to SSA, an overpayment or underpayment
will accrue. 

SSI clients are required to self-report any income that they receive
to claims representatives when applying for SSI and are also required
to self-report any changes in their monthly income by the 10th day of
the following month.  SSA policy requires that claims representatives
verify this reported income, but does not require that claims
representatives check for unreported income unless they suspect that
clients are not reporting all their income during initial or
subsequent eligibility determinations. 

SSA uses both financial eligibility reviews, known as
redeterminations, and computer matching to identify and prevent
overpayments.  During redeterminations, clients report their income
on mailed questionnaires or during 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 detect some types of income
that clients have not reported.  They consist of comparing the SSI
payment records against client information contained in the payment
files of other government agencies.  In order to detect unreported
income, the earnings and unemployment insurance (UI) benefits
reported by SSI clients are compared or matched, for example, against
earnings and UI information that employers report to state
agencies.\2 However, no computer matches are done that could identify
unreported or underreported income from the Aid to Families With
Dependent Children (AFDC) and workers' compensation (WC) programs. 

In 1994, SSA began establishing on-line connections between its field
offices and state agencies that had automated databases that could
easily be linked to SSA's computer system.  SSA did this so that
information on earnings and AFDC and UI benefits contained in these
databases could be automatically obtained as soon as claims
representatives requested it.  Claims representatives use such
information for a variety of purposes, including verifying the amount
of income that clients report when applying for SSI and during
redeterminations of their continuing financial eligibility. 

On-line access began in a limited number of SSA field offices in
Nashville, Tennessee, in July 1994.  Currently, claims
representatives at all 30 Tennessee field offices are able to access
state data on earnings, UI, and AFDC for any client using the
computers on their desks.  In addition, at the time of this report,
on-line access to WC information was also being negotiated in
Tennessee.  This access takes only a minute or two to retrieve the
pertinent data.  As of mid-1996, on-line access to state wage, UI,
and AFDC data was also fully implemented in SSA field offices in
North Carolina, South Carolina, and Kentucky, and partially
implemented in SSA field offices in 2 other states.  Initial contacts
or negotiations for on-line access were also being conducted in 11
other states. 


--------------------
\2 SSA conducts two types of computer matches using earnings data. 
First, SSA conducts matches twice each year using quarterly earnings
data provided by 41 states and the District of Columbia.  These data
are 6 to 12 months old by the time the results of the computer
matching are sent to field offices for claims representatives to
investigate.  Second, SSA performs computer matches once a year using
its own annual earnings data.  These data are 9 to 21 months old by
the time the results are sent to the field offices for investigation. 


   ON-LINE ACCESS REDUCES
   ADMINISTRATIVE COSTS
------------------------------------------------------------ Letter :3

SSA estimates that in Tennessee, on-line access saves at least $6.50
in administrative costs every time a claims representative obtains
data on-line, which occurs thousands of times each month.  SSA
further estimates that having such access results in clients
receiving their first benefit check at least 1 week earlier.  These
improvements occur because in processing an initial claim and in
conducting redeterminations, claims representatives no longer have to
telephone, write, fax, or sometimes even visit state agencies to
obtain necessary documentation--they simply obtain the information on
their computers. 

Claims representatives use on-line access for a variety of purposes,
including verifying the amount of AFDC or other benefit income a
client reports receiving.  According to SSA, in one field office, SSA
staff found that on-line access saves 30 to 45 minutes of a claim
representative's time per claim.  In another field office, a claims
representative told us that she was able to obtain needed AFDC
information on-line for a co-worker in a field office in another
state who had been waiting for 6 weeks to get this information
through conventional channels. 

Another way that claims representatives use on-line access is to
follow up on computer-matching results indicating that clients may
not have reported all their earnings or UI benefits.  This entails
contacting the recipient and often an employer or government agency
to confirm the receipt of earnings or benefits.  One claims
representative told us, for example, that she uses on-line access to
obtain more up-to-date addresses on the clients she is investigating
and that without such information it has taken her up to 2 months to
find correct addresses.  Other claims representatives told us that
they routinely check earnings on-line when investigating
computer-matching results because the on-line access can provide more
current earnings information and identify additional employers who
are not listed on SSA's computer matches with their own earnings data
or state data. 

A final way that claims representatives use on-line access is to
obtain miscellaneous personal information, such as the Social
Security numbers and birthdays of members of a client's household, as
well as the types of public assistance household members receive. 
Because this type of information is frequently available on-line from
the states, claims representatives no longer have to delay claims
processing while clients search for this information. 

SSA believes that the benefits of on-line access have been
sufficiently demonstrated in field offices in the six states where it
has been fully or partially implemented to warrant expanding it to
other states.  These benefits, which we observed during our field
office visits, include reducing the amount of time and paperwork
required for income verification as well as the amount of time
required for claims processing.  At the time of our work, SSA was
developing a long-range plan to expand on-line access nationwide,
which included developing benchmarks to better measure the amount of
time and money such access will save.  SSA was also actively pursuing
such expansion in field offices in at least 11 additional states.  In
addition to expanding on-line access to other states, however, SSA
could also use it to improve the accuracy of its payments. 


   USING ON-LINE ACCESS TO STATE
   DATA MAY SAVE MILLIONS IN
   PROGRAM DOLLARS
------------------------------------------------------------ Letter :4

Our analysis of SSA overpayment data shows that an estimated $131.3
million in overpayments nationwide occurred between June 1993 and May
1994 because some SSI clients did not report or underreported income
they received.  More than $34 million of these overpayments resulted
from unreported or underreported AFDC, UI, and WC benefit income, and
more than $97 million resulted from unreported or underreported
earned income. 

If SSA field offices nationwide had used on-line access to state
databases, SSI program dollars could have been saved because the
overpayments might have been prevented or more quickly detected. 
Preventing overpayments saves program dollars for two reasons: 
overpayments would never be made and SSA would not have to spend
additional money trying to recover them.  Detecting overpayments
sooner also saves program dollars because, according to SSA
officials, the sooner overpayments are detected, the more likely it
is that they will be recovered. 


      OVERPAYMENTS CAUSED BY
      UNREPORTED AND UNDERREPORTED
      BENEFITS MAY BE PREVENTED OR
      DETECTED EARLIER
---------------------------------------------------------- Letter :4.1

Of the $34.1 million in overpayments related to AFDC, UI, and WC
income, about 89 percent ($30.5 million) occurred because both newly
eligible and ongoing clients did not report to the claims
representatives handling their cases that they were receiving
state-administered benefits.  The remaining 11 percent ($3.6 million)
resulted when newly eligible and ongoing clients who had reported
receiving benefits did not report increases in the amount of their
benefits or did so only after they had received their SSI check
covering the period during which the increase occurred.  (See
table 1.)



                                         Table 1
                         
                          Nationwide Overpayments Resulting from
                          Unreported and Underreported AFDC, UI,
                           and WC Benefits, June 1993-May 1994

                                  (Dollars in Millions)


Overpayments
related to AFDC,
UI, and WC
benefits               Amount     Percent      Amount     Percent      Amount     Percent
-----------------  ----------  ----------  ----------  ----------  ----------  ----------
Unreported              $30.5        89.4       $20.1        97.6       $10.4        77.0
Underreported             3.6        10.6         0.5         2.4         3.1        23.0
=========================================================================================
Total                   $34.1       100.0       $20.6       100.0       $13.5       100.0
-----------------------------------------------------------------------------------------
Note:  An additional $0.8 million in overpayments due to UI benefits
was also made during this period.  The reason given on the SSA
database for these overpayments was miscellaneous.  Percentage totals
may not add to 100 because of rounding. 

Source:  SSA's Index of Dollar Accuracy (IDA) data. 


         OVERPAYMENTS CAUSED BY
         UNREPORTED BENEFITS MIGHT
         HAVE BEEN PREVENTED
-------------------------------------------------------- Letter :4.1.1

The $30.5 million in overpayments resulting from unreported AFDC, UI,
and WC benefits shown in table 1 might have been prevented if claims
representatives had used on-line access to state information on these
benefits.\3 State welfare and employment departments maintain such
information and, according to officials we spoke with, generally
update it at least monthly.  Through an on-line connection to these
data, SSA could easily check to see if SSI clients were receiving
these benefits before becoming eligible for SSI or whether clients
were receiving these benefits and SSI simultaneously.  Alternatively,
programming could be put in place that would automatically compare
SSA records to relevant state earnings and benefit data and generate
lists of discrepancies that were found.  Verification would make it
less likely that such income would be undetected. 


--------------------
\3 This assumes that claims representatives in every state could have
information available to them on other states' data on-line. 


         OVERPAYMENTS CAUSED BY
         UNDERREPORTED BENEFITS
         MIGHT HAVE BEEN DETECTED
         EARLIER
-------------------------------------------------------- Letter :4.1.2

On-line access would not have prevented all of the $3.6 million in
overpayments resulting from underreported AFDC, UI, and WC from
occurring, but might have detected the overpayments sooner. 
Moreover, had they been discovered sooner, the amount overpaid would
have been less. 

These overpayments resulted because clients either did not
self-report increases in their income to claims representatives or
did so only after receiving their SSI checks.  On-line access might
not have prevented these overpayments from occurring because the
benefit income that caused these overpayments would have to have been
paid in order for this income to be reflected in the databases that
claims representatives access on-line.  On-line access might have
detected these overpayments sooner than they would otherwise have
been detected, however, because such access allows SSA to verify the
amount of these benefits as soon as the benefit amount is updated at
the relevant state agency, which occurs once a month or more
frequently. 

The primary ways SSA currently detects SSI overpayments are through
client self-reporting; miscellaneous tips from third parties; or, in
the case of the UI program only, computer matching.  All three
methods are limited:  both self-reporting and third-party tips may
not occur or may not be timely and computer matching, which is done
only for UI, cannot detect overpayments until 6 months after they
have begun. 

Many SSI overpayments were not discovered for periods ranging from
several months to more than 1 year after they began.  According to
fiscal year 1993 and 1994 SSA nationwide data, it took on average 9
months to discover overpayments made to newly eligible clients that
were caused by the receipt of AFDC, UI, WC, and other types of
unearned income.  For ongoing clients, it took SSA on average 15
months to discover such overpayments. 


      OVERPAYMENTS CAUSED BY
      UNREPORTED AND UNDERREPORTED
      EARNED INCOME MAY ALSO BE
      DETECTED EARLIER
---------------------------------------------------------- Letter :4.2

Of the $97.2 million in overpayments related to earned income, 34.2
percent ($33.2 million) resulted because clients failed to disclose
to claims representatives that they had any earned income at all. 
Table 2 refers to these overpayments as unreported earnings.  The
remaining 65.8 percent ($64.0 million) in overpayments occurred
because clients earned more in some months than they had reported to
SSA or reported it only after receiving SSI benefits covering the
same period.  Table 2 refers to these overpayments as underreported
earnings.  Newly eligible clients commonly either did not report or
underreported earnings.  Ongoing clients, however, much more commonly
underreported earnings as opposed to simply not reporting them.  (See
table 2.)



                                         Table 2
                         
                            Nationwide Overpayments Caused by
                           Unreported and Underreported Earned
                                Income, June 1993-May 1994

                                  (Dollars in Millions)


Overpayments
related to
earnings               Amount     Percent      Amount     Percent      Amount     Percent
-----------------  ----------  ----------  ----------  ----------  ----------  ----------
Unreported              $33.2        34.2       $19.3        50.3       $13.9        23.6
Underreported            64.0        65.8        19.1        49.7        44.9        76.4
=========================================================================================
Total                   $97.2       100.0       $38.4       100.0       $58.8       100.0
-----------------------------------------------------------------------------------------
Note:  An additional $140,200 in overpayments was made because of
computational errors in calculating the amount of earnings. 
Percentage totals may not add to 100 because of rounding. 

Source:  SSA's IDA data. 

On-line access has the potential to detect some overpayments caused
by unreported or underreported earnings earlier than they can be
detected using current SSA procedures.  On-line detection cannot
prevent overpayments caused by either unreported or underreported
earnings, however, because of the time lag between when these
earnings occur and when the data pertaining to them are available
on-line. 

In states where on-line access has been implemented, the earnings
data that SSA accesses from state employment departments are 4 to 6
months old and are updated four times a year.  Thus, by using these
data, SSA can detect overpayments 4 to 6 months after they begin. 
The data that SSA currently uses to detect earnings overpayments in
computer matches, on the other hand, are between 6 and 21 months old. 
Thus, under current review procedures, some overpayments may exist
for nearly 2 years before they are detected. 


      ON-LINE ACCESS COULD REPLACE
      CURRENT COMPUTER-MATCHING
      EFFORTS THAT RELY ON OLDER
      DATA
---------------------------------------------------------- Letter :4.3

Two of SSA's predominant methods of detecting unreported and
underreported earnings are (1) requiring that clients self-report all
non-SSI income that they receive and (2) conducting computer matches
to detect any unreported income once clients are on the rolls. 
Because self-reporting can result in information on earnings that is
even more up to date than what can be obtained on-line, some
overpayments could be detected sooner through self-reporting than
with on-line access.  The problem, however, is that SSA cannot rely
on clients to fully report all the income that they receive at or
near the time that they receive it.  In order to detect income that
clients do not report, SSA conducts computer matches. 

The earliest that computer matching could detect an overpayment
varies with the age of the data used in the match.  For example, the
earned-income computer matches rely on earnings data that are
anywhere from 6 to 21 months old.  Therefore, the earliest that
detection could occur would be 6 to 21 months after the overpayment
began.  The earliest that on-line access could detect an unreported
or underreported earnings overpayment, on the other hand, would be 4
to 6 months after it began, because that is the age of the data being
used. 

In field offices that have on-line access, information resulting from
SSA's computer matching of state earnings data to detect overpayments
duplicates information that claims representatives already have
on-line.  This is because on-line access is using the same data that
SSA headquarters obtains twice a year to conduct wage and UI computer
matches.  Two SSA officials we spoke with mentioned that replacing
the state computer matches with an automatic computerized interface
that notified claims representatives when earnings information needed
to be checked would result in more timely notification and would also
free up SSA headquarters resources currently used to conduct the
matches. 

SSA's state computer-matching program compares the earnings that SSI
clients report with the earnings data that employers submit to the
states.  Employees at SSA headquarters prepare computer tapes of the
SSI recipients in each state and mail them to the appropriate state
employment departments.\4 State employees compare these files against
their own earnings databases, adding the names of any SSI recipients
who received earnings over the previous 6 months, and mail the tapes
back to SSA.  At SSA headquarters, computers then compare the state
earnings data against the earnings reported by SSI recipients and
generate lists of recipients with unreported earnings.  SSA
headquarters then provides these lists to the appropriate field
offices where claims representatives investigate the unreported
earnings. 

With an automatic interface using the same telecommunication lines
now connecting SSA field offices on-line with the states, this
earnings comparison could be done without having SSA headquarters
employees prepare computer tapes and issue lists to field offices. 
Such an interface would consist of programming that would
automatically do this comparison and issue these lists over the
telecommunication lines, thereby saving administrative costs. 


--------------------
\4 Of the 41 states and the District of Columbia that provide the
data SSA uses in wage matches, all but a few continue to use tapes
for sending and receiving the files.  The few that use electronic
files send them over the same telecommunication lines as are used for
on-line access. 


      FIELD OFFICES WITH ON-LINE
      ACCESS ARE NOT USING IT TO
      DETECT OVERPAYMENTS
---------------------------------------------------------- Letter :4.4

The claims representatives who have on-line access in the three
states we visited--Tennessee, South Carolina, and Wisconsin--continue
to rely on the traditional methods of self-reporting and computer
matching to identify overpayments caused by unreported or
underreported income.  They normally do not use on-line access to
identify overpayments resulting either from state-administered
benefits or from earnings.  When asked why, they said it was because
they did not believe that on-line access would prevent or more
quickly detect a significant amount of overpayments.  This is because
claims representatives believe that they are normally able to tell
from experience when a client is hiding income and, in such
situations, they then try to uncover it.  They further stated that
SSA policy normally does not require that they check independent
information sources to determine whether clients may be receiving
various types of income that they have not reported, a process
referred to as negative verification.  However, several SSA officials
we spoke with mentioned that on-line access could make checking for
unreported income sources feasible in states where data are
sufficiently automated because it would be inexpensive and nearly
instantaneous. 

Using benefit and earnings overpayment data from Tennessee between
September 1994 and August 1995, we compared conventional overpayment
detection methods with detection using on-line access.  On-line
access can discover overpayments related to AFDC, UI, and WC within 1
month or less.\5 However, we determined that only 18 percent of such
overpayments were detected within this time frame using conventional
methods.\6

We also analyzed Tennessee overpayment cases to see how many
earned-income overpayments might have been detected sooner if claims
representatives had used on-line access as opposed to current
methods.  Because on-line access can discover overpayments related to
earnings in 4 to 6 months, we determined how many earnings
overpayments detected through conventional means were discovered more
than 6 months after they occurred.  We found that on-line access
could have detected nearly 60 percent of the overpayment dollars in
Tennessee occurring between September 1, 1994, and August 31, 1995,
more quickly than was actually done using conventional methods. 
Moreover, on-line access was about equally as effective in detecting
overpayments more quickly for both newly eligible and ongoing
clients. 


--------------------
\5 Detection within 1 month is possible for UI because these data are
at most 1 month old.  Detection of AFDC and WC overpayments could be
done even sooner because these data are at the most a few days old. 

\6 Our analysis assumes that all the benefits that related to these
overpayments were paid by state agencies in Tennessee.  On-line
access could not have prevented or more quickly detected any
overpayments relating to benefits paid in other states, because
claims representatives in Tennessee only have on-line access to
databases maintained by Tennessee state agencies. 


   EXPANDING ON-LINE ACCESS IS
   TECHNICALLY FEASIBLE, BUT
   HAMPERED BY NONTECHNICAL
   CONCERNS
------------------------------------------------------------ Letter :5

Establishing on-line connections between SSA field offices and the
appropriate agencies within a state is not technically difficult or
costly in many states.  Considerable effort, however, may be required
to identify state agencies willing to give SSA on-line access and to
negotiate the agreements necessary.  Some state agencies may not want
to grant SSA such access because they are concerned that the privacy
of individuals may be violated by sharing such personal information
as income and Social Security numbers on-line.  Others, while
agreeing to give SSA on-line access in principle, may refuse to
negotiate the necessary agreements to implement such access until SSA
agrees to grant them reciprocal on-line access to SSA data. 


      THE TECHNOLOGY AND COSTS
      ASSOCIATED WITH IMPLEMENTING
      ON-LINE ACCESS HAVE BEEN
      MINIMAL
---------------------------------------------------------- Letter :5.1

The technology needed for on-line access consists of (1)
telecommunication lines that link SSA field office computers to state
databases and (2) programming that establishes the actual on-line
connections.  Most states only need upgrades to existing
telecommunication lines and programming to implement on-line access. 
Telecommunication lines that already exist between SSA headquarters
and the states can be used in many instances to link state databases
with SSA field offices.\7 In cases where upgraded lines may be
needed, SSA officials told us that, based on their experience where
on-line access has been implemented, such lines are neither costly
nor difficult to install. 

According to an SSA systems official, software that is currently used
to route data between SSA's headquarters and state agencies can be
used, with minor programming changes, for the actual on-line
transmissions.  This same official further explained that the
programming changes that must be made at SSA headquarters take about
15 to 30 minutes for each state whose data are being accessed.  The
programming that must be done by the state agencies, however, is more
involved.  It consists of inputting the names and passwords of the
new SSA users so that the state computer systems will (1) allow these
individuals access to the state data and (2) block access to any data
elements containing personal information about individuals that these
users do not have a legal right to see. 

SSA has not tracked the costs it has incurred when making the
necessary changes to its systems to implement on-line access because
they are viewed as minimal.  Moreover, these costs are not projected
to be extensive nationwide, should on-line access be implemented in
all field offices.  One systems official estimated, for example, that
it would take one-fourth of 1 work year to implement on-line access
nationwide.\8

According to SSA's district managers in states that have implemented
on-line access, SSA has committed to pay a total of $67,000 for
hardware and software changes necessary for SSA to access state
systems.  Average state charges when a claims representative accesses
state records range from nothing to 1.5 cents.\9 Moreover, a number
of state agencies have expressed willingness to provide free access
to state records in exchange for on-line access to SSA data.  (For
more information on reciprocal access, see p.  14 under the heading
"States Have Resisted On-Line Access Until SSA Gives Them Reciprocal
Access.") Finally, some states that do not have fully automated data
are willing to explore giving SSA on-line access in exchange for SSA
sharing the costs of automating their data. 

SSA would incur more significant costs should an on-line system be
developed that would permit claims representatives in one state to
access on-line state information in any other state.\10 These
additional costs would include developing

  -- commonly formatted computer screens displaying state data for
     SSA claims representatives,

  -- single names and passwords that SSA claims representatives could
     use for all states so that the audit trails would still be
     maintained, and

  -- a common menu from which SSA claims representatives could access
     data from all states. 


--------------------
\7 These lines are known as the File Transfer Management System
(FTMS).  SSA installed FTMS so that it could transmit SSA data on its
clients to every state.  State agencies are required by the 1984
Deficit Reduction Act to use this information to better identify
those who are not eligible for public assistance or who are receiving
incorrect benefits. 

\8 A work year is the amount of time provided by one employee on a
full-time basis for 1 year.  For budget purposes, it is calculated on
the basis of 2,080 hours per employee. 

\9 SSA also has on-line access to vital statistics records.  Because
vital statistics bureaus are self-supporting, they have asked for
$0.50 to $3.00 per record. 

\10 An example of when such interstate information would be needed is
when a client lives in one state but works in another.  In order for
such interstate exchange of information to take place, however, state
objections to granting on-line access for claims representatives in
other states would have to be overcome. 


      STATES HAVE RESISTED ON-LINE
      ACCESS BECAUSE OF PRIVACY
      CONCERNS
---------------------------------------------------------- Letter :5.2

Some states have resisted allowing on-line access to their
information because of their concerns about privacy.  Privacy
concerns center around ensuring that personal information that an
individual provides to one government agency is protected from being
disclosed to other agencies that do not have a legal right to it.\11
Granting SSA on-line access to state data does not violate the
privacy of individuals who provide this information, because SSA is
simply using on-line connections to access information to which it
has a legal right.  SSA already routinely obtains this information
from government agencies during the claims-handling process. 
Moreover, state agencies can decide which parts of a record claims
representatives will view. 

Although on-line access has not changed the type of information that
claims representatives obtain from other agencies, it has made
obtaining this information faster and easier.  One state official
responsible for the security of data told us that this can cause
personal privacy concerns, because in making information easier to
obtain for official use, it also becomes easier to obtain for
unofficial or illegal use. 

SSA and state officials heavily involved with on-line access believe
that it is possible to develop on-line systems that minimize the
possibility of accessing personal data for unofficial or illegal use
and also identify the perpetrators, should such abuses occur.  They
further believe that abusing the access to personal information is
not more likely with on-line access than with other data-sharing
methods. 

SSA and the states have taken steps specified in federal security
standards that, we were told, ensure the confidentiality and security
of on-line data in states where on-line access has been implemented. 
These include (1) installing software that screens each user before
establishing an on-line connection to ensure that the connection is
being made from an SSA field office by a valid user and also screens
specific requests for items of information, thereby limiting what
queries a user can get answered; (2) instituting written agreements
between SSA and the state agencies regarding how the on-line data
will be used; (3) requiring that SSA claims representatives sign
releases stating that they will not access state data for any
unofficial reasons; (4) using computer lines dedicated solely to the
transmission of data between government agencies; and (5) issuing all
SSI claims representatives passwords that they must enter before
gaining access to the on-line data.  This last feature leaves an
audit trail each time an agency's data are accessed on-line.  If
abuse is suspected, an agency official can check this trail to see
under which SSA employee's identification code the state data were
accessed, which data screens were viewed, and when they were viewed. 


--------------------
\11 These protections are contained in federal and state privacy
laws.  These laws specify the conditions under which one government
agency can share personal information with other agencies.  One
condition, which applies to the information SSA obtains on-line, is
"routine use." That is, the personal information must be obtained for
a purpose that is compatible with the purpose for which it was
originally collected. 


      STATES HAVE RESISTED ON-LINE
      ACCESS UNTIL SSA GIVES THEM
      RECIPROCAL ACCESS
---------------------------------------------------------- Letter :5.3

Another reason that some state agencies have been reluctant to grant
SSA field offices on-line access to their data is that, until
recently, SSA has refused to give states on-line access to SSA data. 
State agencies use information on recipients of SSA benefits for a
variety of reasons.  For example, states are required by law to
determine whether welfare clients receive any SSA benefits.  This is
accomplished by comparing an electronic file of SSA recipients of
benefits against the client databases maintained by state agencies. 

Historically, SSA has had concerns about granting on-line access to
state agencies for some of the same reasons that have made states
reluctant to grant SSA on-line access.  For example, like some state
officials, some SSA officials are also concerned that on-line access
could compromise the security and confidentiality of the personal
information contained in SSA databases. 

Although we did not evaluate the effectiveness of SSA's security
procedures, SSA system and policy officials told us that these
procedures will be stringent enough to grant government agencies
on-line access to its data without compromising confidentiality.  SSA
intends, for example, to evaluate the type of telecommunication
connections a state will use to access SSA data and then institute
procedures specified in federal security standards that are
commensurate with the security risks such on-line access may pose. 
These measures will include up-front screening to ensure that the
on-line data requests are from valid users in valid states and the
creation of audit trails that can track on-line usage back to
individual users.  SSA will also employ software that automatically
checks for sudden changes in usage patterns, which might indicate
questionable use of the data. 

SSA is beginning to develop policies for granting on-line access to
its databases.  These policies detail the circumstances under which
agencies will be granted on-line access and how the security and
privacy of SSA data will be maintained.  According to one official
involved in developing these on-line policies, SSA is moving in this
direction so that state agencies will agree to grant SSA on-line
access to their data. 

As part of its initial move toward implementing reciprocal on-line
access, SSA is planning to pilot state access to SSA data in North
Carolina and Tennessee in early 1997.  The same connections that give
SSA field offices on-line access to state data in these states will
be used to give state agencies on-line access to SSA data.  Only
those agencies that are legally entitled to SSA data will be able to
access such data on-line.  Furthermore, these agencies will only be
able to access those portions of SSA data to which they are already
legally entitled. 


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

On-line access has demonstrated that programmatic and administrative
savings can be realized that reduce SSA's costs and improve service
to the public by reducing the time it takes to process a claim.  With
appropriate attention to privacy and computer security issues, the
expanded use of on-line access can enhance SSA's ability to
efficiently and effectively manage the SSI program.  We found that
on-line access could help SSA prevent or more quickly detect many
overpayments.  Preventing or detecting overpayments more quickly
would bolster the integrity of the SSI program by better ensuring
that clients are only receiving those benefits to which they are
entitled.  We estimated that about $131.3 million in overpayments
could have been avoided or more quickly detected with the use of
on-line access.  Although SSA has efforts under way to implement
on-line access nationwide, it has not (1) examined its policy
regarding how on-line access can be used in overpayment detection and
prevention in places where it has been implemented, (2) investigated
how such access can be used to replace less timely and
labor-intensive computer-matching methods for overpayment detection
such as those used for earnings and unemployment insurance, and (3)
determined how on-line access can be used to identify overpayments
that are not currently detected through computer matching with the
payment systems of the AFDC and WC programs. 


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

To prevent overpayments or detect them sooner, we recommend that the
Commissioner of SSA

  -- require claims representatives to use on-line access to
     routinely check for unreported sources of income when initial
     and subsequent assessments of eligibility are done, provided
     that it is cost-effective to do so and that the data available
     on-line pertain to the time periods covered by SSI payments and

  -- develop automatic interfaces with state databases that comply
     with laws and standards governing computer matching, privacy,
     and security that can (1) more fully automate the earnings and
     UI computer matches and (2) identify additional income sources
     that do not currently have computer matches. 


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

In commenting on a draft of this report, SSA agreed that on-line
access can be a useful tool for reducing overpayments and it also
agreed with our recommendations.  However, SSA officials noted that
although on-line access is easy and inexpensive in many states, this
may not be true for all states.  They cited, for example, that some
state agencies may not have automated data or the systems within an
agency or between agencies in the same state may not be compatible. 
SSA also noted that because on-line access will presumably be more
difficult and costly in some states than in others, a more thorough
analysis of its costs and benefits is necessary before it is used for
overpayment prevention.  Our report indicates that part of SSA's
expansion of on-line access would include developing benchmarks to
better measure the costs and benefits of such a system.  However, as
our report also indicates, there are states where on-line connections
can now access data inexpensively and easily.  Thus, there is no
reason why such access cannot be used for overpayment prevention in
these instances while SSA pursues the cost-effectiveness of on-line
access for other states.  SSA's formal response is in appendix II. 


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

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

If you have any questions about this report, please contact me on
(202) 512-7215.  Major contributors to this report are listed in
appendix III. 

Jane L.  Ross
Director, Income Security Issues


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

This report focuses on the extent to which on-line access can (1)
improve the administration of the SSI program, (2) reduce
overpayments, and (3) be easily implemented in SSA offices
nationwide.  To accomplish our objectives we (1) visited states where
SSA field offices had on-line access to state data, (2) interviewed
officials at SSA headquarters and regional and field offices, and (3)
analyzed overpayment data from several sources. 

We conducted field visits in two states where on-line access to state
earnings, UI, and AFDC data was fully implemented--Tennessee and
South Carolina.  We also visited Wisconsin, where on-line access was
being piloted in the largest SSA field offices and interviewed
officials involved with the project at SSA's headquarters in
Baltimore.  During these visits, we met with SSA officials and claims
representatives as well as with the state government officials whose
data were being accessed.  We discussed (1) the pros and cons of
on-line access, (2) how such access was being used in each state, (3)
how it could potentially be used to prevent overpayments and increase
SSA staff efficiency, (4) the start-up and continuing costs
associated with such access, and (5) the issues involved in
replicating on-line access in other states. 

In regard to overpayments, the report examines those caused by
unreported or underreported earnings, UI, WC, and AFDC.  These
sources were chosen because (1) they were common causes of
overpayments and (2) many of these overpayments could be prevented or
detected sooner by claims representatives electronically accessing
state data. 

We obtained nationwide aggregate data from SSA overpayment studies on
all overpayments resulting from earnings, UI, WC, and AFDC\12

between June 1, 1993, and May 31, 1994.  We then determined how many
of these overpayments resulted because clients did not report or
underreported the income they received from the above sources.  We
did this by analyzing information contained in the data regarding why
these overpayments occurred. 

The nationwide aggregate overpayment data that we obtained were
limited in that they only showed how long the overpayments lasted on
average.  We obtained more detailed information, however, by
examining all SSI beneficiary cases in Tennessee that had
overpayments caused by wages, AFDC, UI, and WC.  In order to
determine the point at which on-line access might have prevented or
more quickly detected these overpayments, we analyzed (1) when each
of these overpayments began and ended and (2) what percentage of them
could have been more quickly detected had on-line access been used. 
This determination was made on the basis of the age of the on-line
data in Tennessee and how frequently they were updated.  The records
we examined of Tennessee residents pertained to clients who were
assessed for initial or continuing eligibility between September 1,
1994, and August 31, 1995. 



(See figure in printed edition.)Appendix II

--------------------
\12 AFDC data are part of a larger welfare category known as income
based on need (IBON).  In regard to Tennessee data, we were usually
able to tell which IBON income was AFDC income by examining variables
within an individual's record.  However, because the national data we
received were aggregate data as opposed to records on individual
cases, we could not distinguish AFDC cases from other IBON cases. 
The majority of IBON overpayment cases are AFDC cases, however,
according to SSA. 


COMMENTS FROM THE SOCIAL SECURITY
ADMINISTRATION
=========================================================== Appendix I



(See figure in printed edition.)



(See figure in printed edition.)


Appendix III MAJOR CONTRIBUTORS TO
THIS REPORT
=========================================================== Appendix I

GAO CONTACTS

Roland Miller III, Assistant Director, (202) 512-7246
Nancy Cosentino, Evaluator-in-Charge, (415) 904-2117
Jill Yost, Staff Member

STAFF ACKNOWLEDGMENTS

In addition to those named above, the following individuals made
important contributions to this report:  Christopher C.  Crissman
assisted as an adviser, James Wright and John Smale assisted with
design support, and Nancy Crothers assisted in report design. 


*** End of document. ***