Social Security: Better Payment Controls for Benefit Reduction Provisions
Could Save Millions (Letter Report, 04/30/98, GAO/HEHS-98-76).

Pursuant to a congressional request, GAO reviewed: (1) how well the
Social Security Administration (SSA) administers the Government Pension
Offset and the Windfall Elimination Provision (WEP) benefit payment
provisions of the Social Security Act; and (2) the options to improve
any administrative deficiencies.

GAO noted that: (1) from several internal studies of SSA's
administration of the Government Pension Offset and WEP provisions, GAO
estimates that the agency made overpayments costing the social security
trust funds between $160 million and $355 million from 1978 to about
1995; (2) weaknesses in its internal controls are a primary cause; (3)
in implementing the benefit reduction provisions for retired federal
employees, SSA could make better use of available information; (4)
although SSA reviews information on pension payments to federal retirees
to ensure that it has properly applied the Government Pension Offset
provisions, it does not use that information to ensure the appropriate
application of the WEP provision; (5) in implementing the benefit
reduction provisions on retired state and local government workers, SSA
relies on the accuracy of information provided by the retirees regarding
whether they receive, or will in the future receive, a pension that
results from noncovered employment; (6) SSA has not developed any
independent source of this pension information; (7) thus, it cannot
verify the accuracy of the self-reported information, a basic and
effective internal control practice; (8) although SSA managers have long
suspected that its controls needed strengthening, they have not yet
decided on a way to improve them; (9) several courses of action could
improve SSA's internal controls; (10) for retired federal employees, SSA
could periodically use the pension data it already receives from the
Office of Personnel Management (OPM) to check whether WEP has been
properly applied; (11) for state and local government retirees, SSA
needs to obtain independently reported pension data to adequately
control its payments for the Government Pension Offset and WEP
reductions; (12) both retirement systems that pay benefits for
noncovered employment and the Internal Revenue Service (IRS), which
receives reports of each taxpayer's pension income from individual
retirement systems, are potential sources of pension data; and (13) both
sources have various merits and drawbacks.

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

 REPORTNUM:  HEHS-98-76
     TITLE:  Social Security: Better Payment Controls for Benefit 
             Reduction Provisions Could Save Millions
      DATE:  04/30/98
   SUBJECT:  Social security benefits
             Internal controls
             Overpayments
             Retirement pensions
             Government retirement benefits
             Data integrity
             Federal/state relations

             
******************************************************************
** 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 Social Security, Committee on
Ways and Means, House of Representatives

April 1998

SOCIAL SECURITY - BETTER PAYMENT
CONTROLS FOR BENEFIT REDUCTION
PROVISIONS COULD SAVE MILLIONS

GAO/HEHS-98-76

Benefit Reduction Provisions

(105156)


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

  GPO - Government Pension Offset
  IRS - Internal Revenue Service
  OPM - Office of Personnel Management
  SSA - Social Security Administration
  WEP - Windfall Elimination Provision

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


B-276209

April 30, 1998

The Honorable Jim Bunning
Chairman, Subcommittee on Social Security
Committee on Ways and Means
House of Representatives

Dear Mr.  Chairman: 

Some people who earn pensions through employer-sponsored retirement
plans are supposed to have their social security benefits reduced. 
Reductions in social security benefits occur whenever the private
pensions are earned from employment where a person did not have to
pay social security taxes (noncovered employment).  You expressed
concern that these benefit reductions are not always being applied,
resulting in some people's receiving more social security benefits
than they should. 

During the past 20 years, the Congress has twice enacted legislation
that required the Social Security Administration (SSA) to reduce
social security benefits whenever a beneficiary also receives a
pension from employment that was not covered by social security. 
Under the Government Pension Offset (GPO) provision, enacted in 1977,
SSA must reduce social security benefits to these beneficiaries when
their entitlement to social security is based on another person's
(usually their spouse's) social security coverage.  Their social
security benefits are to be reduced by two-thirds of the amount of
their government pension.  Under the Windfall Elimination Provision
(WEP), enacted in 1983, SSA must use a modified formula to calculate
the social security benefits people earn when they have had a limited
career in covered employment.  The modified formula reduces the
amount of payable benefits. 

These benefit reduction provisions exist because of concern about
unfair benefit advantages that accrued to noncovered government
workers.  With regard to GPO, spouse and survivor benefits were
intended to provide some social security protection to spouses with
limited working careers.  The GPO provision reduces spouse and
survivor benefits to persons who do not meet this limited working
career criterion because they worked long enough in noncovered
employment to earn a pension. 

With regard to the WEP provision, the Congress was concerned that the
design of the social security benefit formula provided unintended
windfall benefits to workers who spent most of their careers in
noncovered employment.  The formula replaces a higher portion of
preretirement social security covered earnings when people have low
average lifetime earnings than it does when people have higher
average lifetime earnings.  People who work exclusively, or have
lengthy careers, in noncovered employment appear on SSA's earnings
records as having no covered earnings or a low average of covered
lifetime earnings.  As a result, people with this type of earnings
history benefit from the advantage given to people with low average
social security earnings when in fact their total (covered plus
noncovered) lifetime earnings were higher than they appear to be for
purposes of calculating social security benefits. 

Over the years, social security coverage in the nation's workforce
has grown such that about 96 percent of all employment is covered. 
While several categories of noncovered employment remain, two are
predominant:  federal government workers hired before 1984 and state
and local government workers who participate in their
employer-sponsored retirement system and have elected not to
participate in social security.  Although they are a small portion of
the total U.S.  workforce, there are still millions of workers and
retirees who are potentially affected by the GPO and WEP benefit
reduction provisions.  You asked us to determine how well SSA
administers the GPO and WEP benefit payment provisions of the Social
Security Act and to identify options to improve any administrative
deficiencies. 

To address these questions, we

  -- met with SSA staff responsible for setting policy, maintaining
     records, and overseeing operations relative to GPO and WEP
     administration;

  -- reviewed SSA policies and procedures;

  -- visited or telephoned SSA field facilities to observe or discuss
     administrative actions for these benefit reduction provisions;

  -- examined various reports and data from the Bureau of the Census
     and the Bureau of Labor Statistics on the extent of noncovered
     employment;

  -- met with Internal Revenue Service (IRS) representatives to
     discuss the reporting of pension income for tax purposes; and

  -- met with representatives of 21 retirement systems that
     administer pension plans involving noncovered employment. 

We also identified and examined three recent internal SSA overpayment
studies for GPO and WEP.  In response to disparities relating to the
results of SSA's WEP study, we examined overpayment records for about
500 cases in the study.  We performed this assignment between
November 1996 and January 1998 in accordance with generally accepted
government auditing standards.  Appendix I contains more information
on the scope and methodology of our work. 


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

From several internal studies of SSA's administration of the GPO and
WEP provisions, we estimate that the agency made overpayments costing
the social security trust funds between $160 million and $355 million
from 1978 to about 1995.  Weaknesses in its internal controls are a
primary cause. 

  -- In implementing the benefit reduction provisions for retired
     federal employees, SSA could make better use of available
     information.  Although SSA reviews information on pension
     payments to federal retirees to ensure that it has properly
     applied the GPO provisions, it does not use that information to
     ensure the appropriate application of the WEP provision. 

  -- In implementing the benefit reduction provisions on retired
     state and local government workers, SSA relies on the accuracy
     of information provided by the retirees regarding whether they
     receive, or will in the future receive, a pension that results
     from noncovered employment.  SSA has not developed any
     independent source of this pension information.  Thus, it cannot
     verify the accuracy of the self-reported information, a basic
     and effective internal control practice. 

Although SSA managers have long suspected that its controls needed
strengthening, they have not yet decided on a way to improve them. 

Several courses of action could improve SSA's internal controls.  For
retired federal employees, SSA could periodically use the pension
data it already receives from the Office of Personnel Management
(OPM) to check whether WEP has been properly applied.  For state and
local government retirees, SSA needs to obtain independently reported
pension data to adequately control its payments for the GPO and WEP
reductions.  Both retirement systems that pay benefits for noncovered
employment and IRS, which receives reports of each taxpayer's pension
income from individual retirement systems, are potential sources of
pension data.  Both sources have various merits and drawbacks.  For
example,

  -- Going directly to retirement systems has the advantage that it
     could be the most timely source of pension data but has
     significant logistical problems.  For example, it potentially
     requires identifying and contacting thousands of retirement
     systems, addressing legal problems because many states have
     statutes that restrict disclosure of this type of information,
     and developing a process for these systems to routinely report
     needed pension data. 

  -- Obtaining pension income information from IRS would reduce the
     logistical problems of going to so many state and local
     government retirement systems and would be a relatively low-cost
     improvement.  Further, because SSA has the right to use tax data
     for program administration, there would be a major reduction in
     potential legal problems compared with getting pension
     information from the state retirement systems.  However, this
     approach would require a modification to the report filed with
     IRS so that people receiving pensions based on noncovered
     employment can be distinguished. 


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

Social security benefits are payable to people based on either taxes
paid on their own covered earnings or taxes paid on the covered
earnings of a spouse or parent.  When social security was established
in 1935, it covered only workers in commerce and industry (other than
the rail industry, which was allowed to establish its own retirement
plan).  Over the years, however, coverage has been made mandatory for
most types of employment. 

Today, workers excluded from coverage fall into five broad
categories:  federal civilian employees hired before 1984, railroad
workers, certain employees of state and local governments, household
and farm workers whose earnings are below certain minimum
requirements, and people with low net earnings from self-employment. 
The largest categories of noncovered employment are federal
government employees (about 1.3 million in 1995) and employees of
state governments and their political subdivisions covered under a
retirement system whose members have elected not to join social
security (an estimated 3.3 million to 7.9 million workers). 

To implement the GPO and WEP provisions, SSA needs to know which
social security applicants and beneficiaries are, or will be in the
future, receiving pensions earned in noncovered employment.  SSA
takes several steps to identify them.  When people apply for
benefits, SSA's staff are supposed to examine their earnings record
for significant time gaps in their annual covered earnings.  When the
staff find such time gaps, they are to determine whether noncovered
employment is the reason.  In addition, staff are supposed to ask
applicants whether they are receiving or will in the future receive a
pension (either periodic payments or a lump sum payment) from
noncovered employment.  For those who respond that they are receiving
such a pension, SSA makes the appropriate benefit reduction.  For
those who say that they will receive such a pension in the future,
SSA enters in its records the date when the pension is expected to
begin.  SSA staff are supposed to contact the beneficiaries in these
cases to determine whether an expected pension has begun and to take
any actions needed to adjust benefit payment amounts.  For applicants
who say that they are not receiving and will not receive such a
pension, SSA staff record and accept their answer unless the staff
believe further investigation is warranted. 

For federal retirees, SSA applies additional levels of verification. 
OPM, the agency that administers the federal government's pension
system, provides monthly data that list retired federal employees and
the amount of their civil service pension.  When an application for
benefits is filed, SSA compares the applicant's name and social
security number with similar identification data in the OPM
beneficiary database.  If the comparison provides evidence that the
applicant is receiving a federal annuity, the pension information
detected is appended to his or her earnings record as it is being
examined by SSA staff.  With this information, SSA can offset
benefits to comply with GPO and WEP requirements. 

In addition, for the GPO benefit reduction, SSA makes a monthly
"postentitlement" match in which it compares spouse and survivor
benefit payments to OPM pension payment records.  When SSA detects
information that people receiving spouse or survivor benefits may be
receiving a noncovered pension from their federal employment, it
generates an "alert." SSA refers "alert" cases to its program service
centers for investigation of whether an overpayment has been made and
should be recovered.\1 SSA does not, however, make the same
postentitlement comparison with OPM data to ensure proper application
of the WEP benefit reduction for federal retirees. 

SSA records show that postentitlement comparisons have been effective
in reducing overpayments.  From 1994 through 1996, the comparison
identified about 13,000 possible overpayment cases each year, even
though SSA checks OPM records when a claim is filed.  One SSA staff
person pointed out that overpayments often happen because a person
has an increase in benefits or begins to receive a noncovered federal
pension after filing for social security but has not notified SSA. 

Most of the investigated cases reflect instances in which SSA and OPM
records differ in the amount of the noncovered pension a person
received.  However, each year the comparison has also identified
between 2,000 and 3,000 cases in which SSA did not know that a
beneficiary was receiving a federal pension. 

Periodically, SSA examines a sample of about 500 of the identified
cases to compare the reduction in overpayments resulting from the
monthly GPO match with the costs of conducting the match.  As shown
in table 1, this postentitlement monthly match reduces overpayments
by millions of dollars each year; administrative expenses for the
match and investigation of cases are a fraction of the benefits that
are derived. 



                          Table 1
          
              Results of SSA's Postentitlement
            Matching for Spouse Benefits Paid to
                 Retired Federal Employees

                     Projected
                      18-month     Estimated     Benefits-
                       savings          cost       to-cost
Year of study       (millions)    (millions)         ratio
----------------  ------------  ------------  ------------
1991                      $7.5          $0.3          25:1
1993                      $6.1         $0.27          22:1
1996                      $9.9         $0.45          22:1
----------------------------------------------------------
For the other large group of noncovered workers, retired state and
local government employees, SSA has no third-party pension
information to identify those who receive pensions from their
noncovered employment.  The varying level of SSA's payment controls
for both major groups of noncovered employees is summarized in table
2. 



                          Table 2
          
           Summary of Internal Control System for
            Ensuring Application of GPO and WEP
                     Benefit Reductions

                                         State and local
                    Federal government      government
                         retirees            retirees
                    ------------------  ------------------
Internal control
action                   GPO       WEP       GPO       WEP
------------------  --------  --------  --------  --------
Asks questions           Yes       Yes       Yes       Yes
 based on earnings
 records
Checks pension           Yes       Yes        No        No
 data provided by
 third party
Makes                    Yes        No        No        No
 postentitlement
 match to pension
 data
----------------------------------------------------------
Timely identification of payment errors is important.  According to
an SSA official, SSA issued regulations that limit its ability to
change benefit payment amounts in order to provide for efficient
program administration and protect beneficiaries from having to repay
large overpayments caused by SSA errors.  Under SSA regulations,
reductions in the amount of benefits paid to an individual generally
must be made within 4 years of the date of a benefit determination. 
If more than 4 years pass, SSA regulations preclude changing the
benefit amount either to correct it or to recover any overpayment,
unless fraud or some other unusual circumstance exists.  However,
favorable changes in benefit determinations can be made at any time. 


--------------------
\1 Six program service centers throughout the country and two offices
in Baltimore, Maryland, house and service the records of individuals
who are receiving social security benefits. 


   INTERNAL STUDIES INDICATE THAT
   SSA'S PAYMENT CONTROLS ARE
   INADEQUATE
------------------------------------------------------------ Letter :3

Concerned about the adequacy of its payment controls, SSA conducted
three studies in the 1990s to assess the extent of GPO- and
WEP-related overpayments.  For GPO, SSA studied spouse and survivor
benefit payments being made to retired state and local government
employees in Illinois and Ohio.  For WEP, SSA studied the utility of
using the OPM pension data it receives for postentitlement matching
of benefit payments to retired federal employees.  Although these
studies have certain limitations, they indicate that SSA made

  -- between $109 million and $274 million in GPO-related
     overpayments from 1978 to about 1993 and

  -- between $52 million and $81 million in WEP-related overpayments
     from 1986 to about 1995. 

On the basis of these studies, we estimate that GPO- and WEP-related
overpayments range between $160 million and $355 million.  While SSA
has corrected and recovered overpayments in many of the cases
studied, it has not yet established any approach to improve its
payment controls. 


      SSA'S STUDIES OF THE GPO
      PROVISION FOR RETIRED STATE
      AND LOCAL GOVERNMENT WORKERS
      INDICATE THAT OVERPAYMENTS
      ARE BEING MADE
---------------------------------------------------------- Letter :3.1

For many years, SSA has been concerned with its uneven approach to
administering the GPO provision between federal and state and local
government retirees.  Verifying spouse and survivor benefit payments
for federal employees is facilitated because they can be matched to
pension benefit data available from a single source (OPM).  Further,
legislation explicitly permits such a match to enhance program
integrity.  For state and local government employees, however, such a
match is logistically more difficult.  No one retirement system makes
benefit payments.  Rather, possibly thousands of pension plans and
retirement systems, administered by both public and private entities,
make payments based on noncovered employment.  SSA has no information
that identifies these employers or the retirement systems offered to
their employees.  Further, to protect the privacy of former
employees, many states have laws that restrict the disclosure of
detailed pension information. 

Confronted with these problems, SSA relies on its field offices to
detect retiring noncovered state and local government workers when
they apply for benefits.  By questioning applicants and examining
earnings records, field staff are expected to identify cases in which
GPO applies. 

Around 1993, SSA began to examine its application of the GPO offset
to state and local government retirees residing in Illinois.\2 One of
its primary purposes was to assess whether its different levels of
internal controls were manifest in actual overpayments.  In the
absence of information on who earned pensions from noncovered
employment, SSA had to make its best "guess" to identify beneficiary
payments for study.  SSA decided to select persons living in Illinois
who were receiving spouse or survivor benefits and had other pension
income.\3

From this group of persons, SSA then searched its earnings records to
identify possible noncovered employment.  Through this analysis, SSA
identified more than 20,000 current beneficiaries living in Illinois
to whom the GPO offset might apply and where its payment records did
not indicate that the GPO reduction had been considered.  SSA
selected 440 of these cases for study. 

Among these 440 cases, SSA found 16 cases receiving overpayments. 
The overpayments exceeded $307,000, an average of about $19,200 per
case.  It also found one case for which its regulation, which places
a 4-year limit on SSA's changing of payment determinations and
subsequent recovery of overpayments, precluded action.  Because 17
overpayment cases had been found (3.8 percent of the cases), the
Illinois study showed that there were about 730 overpayments in this
state.  If the average of $19,200 per case were applied, the study
suggested about $14 million in GPO-related overpayments. 

Believing that its selection criteria needed to be refined to reduce
the number of unproductive investigations, SSA undertook a similar
study in 1995 of GPO application to state and local government
retirees from Ohio.  Making some adjustment in the manner in which
cases were selected for study, SSA chose 500 cases for examination. 
As of October 1997, the study had not yet been completed, so we have
not used the available results to estimate overpayments.  However,
SSA had identified 19 cases with more than $312,000 in overpayments,
about the same proportion of cases and a similar amount as detected
in the Illinois study. 

Using the results of the completed Illinois study to estimate the
extent of GPO overpayments nationally is problematic because data on
the number of noncovered state and local government employees are
limited and what exists is contradictory.  For example, a recent
Bureau of the Census report indicated that in 1994 there were about
4.1 million state and local government retirees and about 13 million
workers.  The report also showed that about 60 percent of the state
and local government employees were not covered by social security. 

In contrast, a recent Bureau of Labor Statistics report for 1994
indicated that 24 percent of state and local workers were not covered
by social security.\4 In addition, SSA staff have estimated that only
about 25 percent of state and local government workers in the country
are not covered by social security.\5 If we use Census data on the
number of retired state and local government workers, 25 percent and
60 percent to represent the range of noncovered employment, and the
results of the Illinois GPO study, the extent of GPO-related
overpayments ranges from $109 million to $274 million.  Table 3 shows
how this estimate is constructed. 



                          Table 3
          
             Range of Estimated Nationwide GPO-
                    Related Overpayments

                                                      High
                                Low estimate  estimate (60
                                 (25 percent       percent
                                 noncovered)   noncovered)
------------------------------  ------------  ------------
Estimated retirees with          1.0 million   2.5 million
 noncovered pensions\a
Estimated cases needing              150,000       375,000
 investigation (15 percent of
 cases in SSA's Illinois
 study)
Estimated GPO overpayment              5,700        14,250
 cases (3.8 percent of
 investigated cases in SSA's
 Illinois study)
Estimated GPO overpayment       $109 million  $274 million
 amount ($19,200 per case in
 SSA Illinois study)
----------------------------------------------------------
\a According to 1994 Bureau of the Census data, there were 4.1
million retired state and local government workers. 


--------------------
\2 The report prepared on the Illinois study does not specify the
time period covered by the study or any time-related restrictions on
the cases selected.  It appears that the study covers cases from 1978
(when GPO became effective) until the early 1990s. 

\3 Until 1990, SSA received and processed IRS form W-2P, which
reported pension income.  In 1990, this form was eliminated, and
pensions, along with many other types of income and taxable events,
were reported directly to IRS on form 1099R. 

\4 "Employee Benefits in State and Local Governments, 1994," Bureau
of Labor Statistics, Washington, D.C., May 1996. 

\5 Census staff speculated that the difference between the estimated
levels of social security coverage in its study and others could be
attributed to several factors.  The Census study included only
retirement systems that had dedicated revenue sources other than the
administering government and a separate accounting fund.  These
criteria tend to exclude smaller, pay-as-you-go systems.  The Census
study also focused on financial aspects of state and local retirement
systems.  Supplementary data on each retirement system's membership,
type of beneficiary, and social security coverage are collected to
supplement the financial data and are considerably less reliable than
the financial data. 


      SSA'S STUDY OF THE WEP
      PROVISION FOR RETIRED
      FEDERAL GOVERNMENT EMPLOYEES
      INDICATES THAT OVERPAYMENTS
      ARE BEING MADE
---------------------------------------------------------- Letter :3.2

In August 1994, SSA referred to its service centers about 36,000
cases for which it suspected WEP-related overpayments to federal
retirees.\6 SSA had never performed a postentitlement match with OPM
data to verify WEP application to federal retirees as it does for
GPO.  To consider whether it needed to initiate a postentitlement
matching program, SSA instructed its service centers to take special
steps so that the results of investigating these cases could be
determined.  The service centers were to

  -- maintain control sheets and annotate the results of each
     referred case with designated codes.  These codes were designed
     to indicate either whether WEP applied in the referred case or
     the reason WEP did not apply.\7

  -- code the underlying cause of the payment problem in their
     overpayment accounting system in a special manner.  The special
     code would allow the generation of management reports on the
     WEP-related overpayments detected from the referred cases. 

  -- initiate overpayment recovery actions where appropriate. 

Management reports on the number of overpayments from investigating
these cases indicated that 154 overpayments (about $725,000) were
detected from the 36,000 referrals (about 0.4 percent).  Overpayments
averaged about $4,700 in these cases.  Although no formal decision
has ever been made, SSA staff told us that this level of results did
not warrant a postentitlement matching program. 

Our discussions with field staff who managed the investigation of the
referred WEP cases indicated, however, that the number of
overpayments discovered during the study was much greater than the
154 identified in SSA's management report.  In response to our
questions, service center staff told us that they had apparently
often failed to follow the instructions to use a special code when
entering the WEP overpayments in SSA's accounting system.  Thus, when
SSA generated management reports based on the special code, most of
the WEP overpayment cases identified in the study were not counted. 

We were able to obtain the control sheets several of the service
centers used and we found thousands of cases were coded "WEP
reduction applies." To estimate the extent of WEP-related
overpayments in SSA's study, we asked SSA to provide overpayment
records from the cases investigated by one of the service centers
where we had the control list of cases.  We asked for overpayment
records for all cases that had been coded "WEP reduction applies."
There were 787 of the 6,674 cases investigated by this service center
coded this way. 

SSA provided overpayment records for 528 of these cases.  Our
analysis of the 528 cases indicated that 487 had WEP-related
overpayments and that overpayment amounts totaled about $2.2 million
in these cases, or an average of about $4,500 per case.\8 The data
also showed that SSA had either collected or was in the process of
recovering most of these overpaid benefits. 

With this information, the extent of overpayments among the 36,000
referred cases can be extrapolated.  Given that the 487 overpayments
represented 7.3 percent of this service center's cases, we estimate
that there are about 2,630 overpayment cases in the 36,000 cases
referred to all the service centers.  At an average overpayment of
$4,500 per case, we estimate that about $11.8 million in WEP-related
overpayments were detected in SSA's study. 

This estimate, however, excludes cases that were uncollectible
because of SSA's 4-year time limitation for recovering overpayments. 
At the service center that we examined, an additional 827 of the
6,674 cases (12.4 percent) involved WEP-related overpayments for
which recovery was precluded by regulation. 

Once a service center determines that the regulation limiting the
time period in which overpayment recovery can occur applies, it does
not calculate the amount of the overpayment in a given case. 
Applying the average overpayment amount detected for such WEP cases
allows us to estimate the overall overpayments made in the referred
cases.  Thus, we estimate about 4,500 cases among the 36,000 referred
cases for which regulations precluded overpayment recovery.  At an
average of $4,500 per case, total lost overpayments amount to $20.3
million.\9 Coupled with the previously discussed $11.8 million in
overpayments, the absence of a postentitlement comparison to OPM
records for WEP contributed to about $32 million in overpayments to
retired federal employees. 

We were unable to identify any studies of the extent to which SSA has
made WEP-related overpayments to retired state and local government
workers.  Therefore, we used SSA's WEP study results for federal
retirees to approximate the amount of WEP-related overpayments to
retired state and local government workers.  We believe that this
approximation is reasonable for several reasons. 

First, the approximation is conservative.  SSA's WEP study for
federal employees reflects the level of overpayments occurring even
though SSA checked OPM pension data at the time beneficiaries first
applied for social security benefits.  SSA does not have a similar
ability to detect social security applicants already receiving a
pension resulting from their noncovered employment by state or local
governments.  Thus, it seems reasonable to assume that SSA makes
WEP-related overpayments to a higher proportion of state or local
government retirees than it does to federal government retirees. 
Second, the amount of the WEP reduction is based on each retiree's
covered earnings history, not on the amount of their government
pension.  Thus, even if federal pensions are higher on average than
state and local government pensions, there is no reason to believe
that the covered earnings history, and thus the appropriate size of
the WEP reduction, would differ between the two groups. 

As in the GPO estimate, because of the conflicting data on the number
of noncovered state and local government retirees who receive a
pension from noncovered employment, we used an estimating range. 
Table 4 shows that the absence of third-party pension data leads to
an estimate of undetected WEP-related overpayments among state and
local workers ranging from about $20 million to about $49 million. 



                          Table 4
          
             Range of Estimated Nationwide WEP-
           Related Overpayments to Retired State
          and Local Government Workers That Remain
                         Undetected

                                                      High
                                Low estimate  estimate (60
                                 (25 percent       percent
                                 noncovered)   noncovered)
------------------------------  ------------  ------------
Estimated retirees with          1.0 million   2.5 million
 noncovered pensions\a
Estimated possible overpayment        22,000        55,000
 cases based on WEP study (2.2
 percent of cases)

Estimated overpayment cases:
----------------------------------------------------------
--Undetected WEP overpayment           1,606         4,015
 cases (7.3 percent of cases)
--Cases in which overpayment           2,728         6,820
 recovery was precluded by SSA
 regulation (12.4 percent of
 cases)
==========================================================
Total                                  4,334        10,835
Estimated average WEP                  $19.5         $48.8
 overpayment ($4,500 per case)       million       million
----------------------------------------------------------
\a According to 1994 Bureau of the Census data, there were 4.1
million retired state and local government workers. 


--------------------
\6 The WEP study covered claims filed in 1986 (when the provision
became effective) through March 1993. 

\7 There are about a half dozen conditions under which WEP does not
apply. 

\8 Most of the overpayments that were not caused by WEP were related
to individuals who continued to have wages or net earnings from
self-employment after retirement, making them susceptible to the
annual earnings test.  Persons younger than 70 whose earnings exceed
thresholds have their social security benefits reduced. 

\9 This estimate of WEP-related overpayments is conservative.  It is
likely that the average overpayment in these cases is larger than the
average overpayment in cases in which recovery was pursued because of
the 4-year limit on overpayment recovery. 


   SSA CAN IMPROVE PAYMENT
   CONTROLS
------------------------------------------------------------ Letter :4

Improving administration of the WEP benefit reduction for retired
federal government employees is relatively straightforward.  SSA can
use the OPM pension data it currently receives to periodically check
the accuracy of benefit payments made to retired federal government
employees.  Improving GPO- and WEP-related payment controls for
retired state and local government employees is more difficult,
however.  We identified two sources of information that can be used
to identify state and local government retirees who have pension
income resulting from noncovered employment:  the individual
retirement systems making pension payments to their eligible members
and IRS, which receives reports from the retirement systems
identifying pensions paid to each beneficiary every year.  The merits
and drawbacks associated with obtaining pension data from each source
are described below. 


      OBTAINING PENSION DATA
      DIRECTLY FROM RETIREMENT
      SYSTEMS
---------------------------------------------------------- Letter :4.1

The biggest advantage derived from obtaining third-party pension data
directly from retirement systems relates to the timeliness of the
data.  As periodic payers of benefits, pension plans have the most
current information about the beneficiaries, the pension amounts they
are receiving, and the form and basis of the payment (annuity, lump
sum, survivor benefit, and so on).  In addition, by going directly to
the retirement system paying benefits, there would be no need for IRS
to act as a conduit for the information, slowing its availability to
SSA. 

Direct receipt of pension data from the retirement systems, however,
creates significant logistical problems for SSA and the systems.  SSA
would have to identify the retirement systems paying benefits to
state and local government retirees.  There are possibly thousands of
plans and systems, some paying pensions for noncovered employment and
others not making these types of payments.  These systems would have
to be contacted to determine whether they pay pensions to persons who
worked in noncovered employment. 

Once these contacts are made, it is likely that legal problems will
have to be resolved.  Our contacts with 21 retirement systems for
state and local government employees in five states and the District
of Columbia indicated that states frequently had statutes restricting
the disclosure of this type of information.  Some retirement systems
believed that the statutes would allow the disclosure of beneficiary
payment data to SSA.  Others, however, did not believe disclosure of
this type of pension data to SSA was permissible under state law. 
This could mean that either special agreements would have to be
established to obtain disclosure or perhaps litigation might be
necessary. 

A final concern relates to the administrative burden this type of
approach would place on retirement systems.  Once retirement systems
have been identified and contacted and disclosure issues resolved, a
system would still have to be established to routinely report
pertinent pension information to SSA.  This would be an additional
reporting requirement for the retirement systems, because now they
report beneficiary pension payments to IRS only.  In effect, SSA
would be seeking the same basic information being reported to IRS
each year but perhaps in a different time period, or even more
frequently. 


      OBTAINING PENSION DATA FROM
      IRS
---------------------------------------------------------- Letter :4.2

Obtaining pension data from IRS, rather than directly from retirement
systems, would largely avoid the logistical problems associated with
obtaining pension data from state and local government retirement
systems.  As noted, IRS receives annual reports of the amount of
pensions each retirement system pays each of its eligible members on
form 1099R ("Distributions From Pensions, Annuities, Retirement or
Profit-Sharing Plans, IRAs, Insurance Contracts, etc.").  In
addition, SSA can obtain access to form 1099R records for program
administrative purposes, eliminating the legal issues noted above. 
Finally, this approach does not require the separate development of a
new reporting system or the duplicate reporting of pension income,
thereby mitigating concern about additional taxpayer burdens. 

The drawback of using form 1099R is that its current format does not
provide SSA with sufficient information to target cases in which
overpayments are being made.  As the title of the 1099R form
suggests, more than pension payments are reported on it.  Payments
from life insurance contracts, charitable gift annuities, various
types of tax-free exchanges, distributions from and direct rollovers
of individual retirement accounts, lump sum distributions, and
pension income are among the many types of payments reported on form
1099R.  A further complication is that with only about 4 percent of
the workers not covered by social security, most of the pension
payments reported on form 1099R would be earned in employment covered
by social security.  These pensions are not subject to the GPO and
WEP benefit reduction provisions. 

With about 50 million form 1099R reports filed with IRS in 1995, a
way is needed to distinguish between pensions from noncovered
employment and the many types of payments being reported.  Adding a
unique code for this purpose could do this.  IRS has created a
specific data field, known as the distribution code, to differentiate
the types of payments reported on form 1099R.  A code placed in this
field indicating that the payment was a pension based on noncovered
employment would make form 1099R useful to SSA.  With this
information, SSA could then compare only the form 1099R reports
having this code with its beneficiary payment records.  SSA could
examine cases for which it has not applied the GPO and WEP
reductions.  This would allow SSA to target for investigation the
cases most likely to have an overpayment. 

We discussed with IRS ways that the form 1099R could be revised to
provide pension information useful to SSA and the effect that
processing changes in the form would have on operations.  IRS
officials said that the proposed approach would minimally affect
their processing costs and that they could be implemented for
processing year 2000. 

We also discussed these options with many of the retirement systems'
managers whom we visited.  Overall, most of the managers (12 of 14)
told us that requiring that they identify on form 1099R payments
based on noncovered employment would not pose a severe administrative
problem to them.  The managers said that the modification would just
involve minor programming changes and that they often have
information that their pension payments are based on noncovered
employment.  A concern about this change was that sufficient lead
time be provided to allow them to make the necessary adjustments to
their processes for tax reporting and to obtain the needed data. 


   CONCLUSIONS
------------------------------------------------------------ Letter :5

SSA needs better payment controls to administer the GPO and WEP
benefit reduction provisions of the Social Security Act.  Since the
enactment of these provisions, SSA has not had sufficient ways to
verify, in a timely or complete manner, whether beneficiaries are
receiving pensions earned through noncovered employment.  Recent SSA
studies indicate that the absence of such verifications has resulted
in overpayments between 1978 and 1995 that we estimate to have ranged
from $160 million to $355 million.  Further, much of these
overpayments has been lost because SSA has not detected the errors
within the 4 years required by its regulations. 

SSA has been concerned for several years about the adequacy of its
payment controls as they relate to the GPO and WEP provisions, and it
has struggled to find ways to enhance them.  Certainly, SSA can use
the OPM pension data that it already has to enhance its WEP
enforcement controls for federal employees.  But it also needs a way
to obtain independent pension data for nonfederal retired workers who
did not contribute to social security.  We believe that obtaining
form 1099R data from IRS is the most efficient and least disruptive
option.  A modification to form 1099R would provide SSA with the data
it needs to improve the administration of GPO and WEP for state and
local government retirees. 


   RECOMMENDATIONS TO THE
   COMMISSIONER OF SOCIAL SECURITY
------------------------------------------------------------ Letter :6

To improve the administration of the GPO and WEP benefit reductions,
the Commissioner of Social Security should

  -- begin using pension information obtained from OPM to establish a
     postentitlement matching program for WEP so that it can verify
     the accurate payment of social security benefits to retired
     federal government employees and

  -- work with IRS to revise the reporting of pension information on
     IRS form 1099R, so that SSA would be able to identify people
     receiving a pension from noncovered employment and to improve
     its internal controls by establishing a postentitlement matching
     program. 


   AGENCY COMMENTS AND OUR
   EVALUATION
------------------------------------------------------------ Letter :7

Both SSA and IRS were asked to comment on this report.  SSA responded
favorably to our recommendations.  SSA's comments are included as
appendix II. 

SSA acknowledged that it had been studying the way it administers the
GPO and WEP provisions in order to improve its payment controls, and
it agreed with both recommendations in our report.  The agency plans
to use OPM pension data to periodically check the accuracy of benefit
payments made to retired federal government employees.  SSA also said
that it believes that revising form 1099R to indicate payments based
on pensions from noncovered employment is the best way to improve
GPO- and WEP-related payment controls for retired state and local
government employees.  The agency will work with IRS to implement the
recommended modifications to form 1099R. 

SSA also referred to the many cases in this report for which
overpayment recovery was precluded by its regulations limiting
recovery if more than 4 years have elapsed from the date of a benefit
determination.  SSA said that the additional planned matching
operations would be scheduled frequently enough to permit benefit
reductions to occur in a more timely manner and preclude the
imposition of its regulations governing the timely identification of
payment errors. 

IRS said that it will work with SSA to revise the reporting
requirements for pension information on form 1099R so SSA will be
able to identify persons receiving a pension from noncovered
employment.  IRS said that it has been in contact with SSA to offer
its support and is in the process of scheduling meetings with
internal stakeholders to begin addressing issues identified in its
preliminary analysis.  IRS believes that the additional reporting
requirement must be constructed in a way that minimizes the
perception of increased burden on the retirement systems and that any
associated disclosure issues under Internal Revenue Code Section 6103
must be addressed immediately. 

IRS revised an earlier estimate to us that these changes could be
implemented in time for processing year 1999.  It now believes that
the necessary changes to the forms and systems cannot be made before
processing tax year 2000 returns.  We have revised our report to
reflect IRS's latest estimate. 


---------------------------------------------------------- Letter :7.1

We are providing copies of this report to the Commissioners of Social
Security and Internal Revenue, the Director of the Office of
Management and Budget, and other congressional committees with an
interest in this matter.  We will also make copies available to
others upon request.  Please contact me on (202) 512-7215 if you have
any questions about this report.  Other major contributors to this
report are listed in appendix IV. 

Sincerely yours,

Barbara D.  Bovbjerg
Associate Director
Income Security Issues


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

Work on this assignment was performed between November 1996 and
January 1998.  In doing our work, we visited Social Security
Administration (SSA) offices at its headquarters in Baltimore,
Maryland, and in its San Francisco region, which services claims from
California and has a significant number of noncovered employees.  We
also discussed various issues by telephone with SSA staff in its
Southeastern, Mid-America, Great Lakes, and Northeastern program
service centers. 

A portion of our work also involved discussions with Internal Revenue
Service (IRS) and retirement systems that pay benefits to people who
worked in state and local, noncovered government employment.  We
visited 21 retirement systems in states where there are a significant
number of workers who are not covered by social security.  The states
we visited were California, Colorado, Louisiana, Maryland, and
Virginia.  We also visited a business in the District of Columbia
that administers pension plans on a contract basis for many state and
local government agencies.  During our visits to retirement systems,
we discussed state laws governing the disclosure of plan information
and the effect of options for providing pension data to SSA on their
operations. 


   EVALUATION OF GPO AND WEP
   ADMINISTRATION
--------------------------------------------------------- Appendix I:1

To determine how well SSA administers the Government Pension Offset
(GPO) and Windfall Elimination Provision (WEP) benefit reduction
provisions, we

  -- reviewed past reports by the Office of Inspector General,
     Congressional Research Service, and Congressional Budget Office
     and SSA's Office of Program Integrity Reviews concerning the GPO
     and WEP benefit reduction provisions;

  -- met with SSA headquarters staff in components responsible for
     policy, operations, and systems for GPO and WEP;

  -- reviewed SSA operating procedures for processing benefit claims
     and verifying payments made to retired government employees who
     worked in positions not covered by social security; and

  -- examined the results of monthly computer matching to pension
     data provided under agreement with OPM to identify persons
     receiving spouse benefits and federal pensions. 

During the course of our work, we identified three SSA studies
between 1993 and 1995 of GPO and WEP administration that were
directly related to the objectives of our review.  There were two
studies concerning the adequacy of controls for GPO administration
for state and local government retirees who earned pensions from
their noncovered employment.  At the conclusion of our work, only one
of the two was completed.  The other study was close to completion in
October 1997, but several cases were still being investigated and no
formal reports or summaries of the outcome had yet been prepared. 

The study of WEP administration was designed to examine whether SSA
should make periodic comparisons of its beneficiaries to OPM pension
data to detect retired federal government workers receiving a
noncovered pension.  The results of that study were never published
or formally summarized.  Management reports on the number of
overpayments detected among the referred cases indicated to SSA staff
that a postentitlement matching operation would find few payment
errors and was probably not warranted. 

Although many of the records documenting the WEP study's results
could not be located, we were able to obtain several of the master
case control lists used by the program service centers to broadly
track progress of work and case outcomes.  To get a better
understanding of the actual results of the WEP study, we asked SSA to
provide all overpayment amounts for cases examined by one service
center, Mid-America.  We selected this service center for analysis
because its case control records were the most complete of the four
sets we had.  We asked for any overpayment records established after
July 1994 (the date the cases were referred to the service centers)
for every case in which the investigation outcome was coded "WEP
reduction applies."

The Mid-America Service Center had received 6,674 cases to
investigate for possible WEP-related payment errors.  Overall, we
requested overpayment records for 787 cases.  We received overpayment
records for 528 of these cases.  After examining these records, we
eliminated 41 from our study because available data indicated that
the overpayment was most likely related to reasons other than the WEP
study.  Most of these cases were eliminated because of the annual
earnings test that restricts the amount of wages beneficiaries can
earn before SSA begins reducing their benefits. 


   OPTIONS FOR IMPROVING GPO AND
   WEP ADMINISTRATION
--------------------------------------------------------- Appendix I:2

To identify options for improving SSA's payment controls and consider
their merits and drawbacks, we met with officials from retirement
systems in California, Colorado, Louisiana, Maryland, Virginia, and
the District of Columbia that administer pension plans that pay
benefits to persons working in noncovered employment.  At the plans,
we obtained basic information on the types of plans administered and
the number of persons in them in noncovered employment.  We discussed
state laws governing access to pension records and obtained the
opinions of officials on various options for reporting pension income
to SSA and how those options would affect their operations.  We also
met with IRS staff to discuss their views on various options for
reporting noncovered pension income. 




(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.)




(See figure in printed edition.)Appendix III
COMMENTS FROM THE INTERNAL REVENUE
SERVICE
=========================================================== Appendix I



(See figure in printed edition.)


GAO CONTACTS AND ACKNOWLEDGMENTS
========================================================== Appendix IV

GAO CONTACTS

Roland H.  Miller III, Assistant Director (202) 512-7246
William J.  Staab, Evaluator-in-Charge (202) 512-6814

ACKNOWLEDGMENTS

In addition to those named above, Gerard V.  Grant, Evaluator, John
G.  Smale Jr., Senior Social Science Analyst, and James P.  Wright,
Assistant Director (Study Design and Analysis) made contributions to
the development of this report. 


*** End of document. ***