Social Security: Trust Funds Can Be More Accurately Funded (Chapter
Report, 09/02/94, GAO/HEHS-94-48).

Each year, the social security trust funds are credited with revenues
derived from income taxes paid on social security benefits. But do they
get the right amount? GAO reports that the social security trust fund's
revenues could be increased by recognizing additional taxes identified
through the Internal Revenue Service's (IRS) efforts to locate
underreported taxable income and through better detection of
underreported tax-exempt interest. Recognizing additional taxes
identified by IRS could have boosted the trust funds by more than $200
million in tax revenue and investment income for tax years 1984 to 1989.
Further, data from the Federal Reserve and the Investment Company
Institute indicate that taxpayers may have underreported an estimated
$7.2 billion in tax-exempt income on their 1989 tax returns.

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

 REPORTNUM:  HEHS-94-48
     TITLE:  Social Security: Trust Funds Can Be More Accurately Funded
      DATE:  09/02/94
   SUBJECT:  Income taxes
             Social security benefits
             Trust funds
             Noncompliance
             Tax administration
             Tax nonpayment
             Tax exempt status
             Reporting requirements
             Federal agency accounting systems
IDENTIFIER:  Hospital Insurance Trust Fund
             Old Age and Survivors Insurance Trust Fund
             Social Security Disability Insurance Trust Fund
             IRS Underreporter Program
             IRS Taxpayer Compliance Measurement 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.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


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


Report to the Chairman, Subcommittee on Oversight, Committee on Ways
and Means, House of Representatives

September 1994

SOCIAL SECURITY - TRUST FUNDS CAN
BE MORE ACCURATELY FUNDED

GAO/HEHS-94-48

Taxation of Social Security Benefits


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

  ICI - Investment Company Institute
  IRS - Internal Revenue Service
  OID - original issue discount
  OTA - Office of Tax Analysis
  SOI - Statistics of Income
  SSA - Social Security Administration
  TCMP - Taxpayer Compliance Measurement Program

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


B-252068

September 2, 1994

The Honorable J.J.  Pickle
Chairman, Subcommittee on Oversight
Committee on Ways and Means
House of Representatives

Dear Mr.  Chairman: 

This report responds to your request that we review several matters
related to how unreported tax-exempt income affects the taxation of
social security benefits.  It also discusses how the Department of
the Treasury, with the assistance of the Internal Revenue Service,
determines the amount of revenue owed to the social security trust
funds from the taxation of benefits.  The report contains
recommendations to Treasury and the Internal Revenue Service to
improve the accuracy of the funding process. 

We are sending copies of this report today to the Secretary of the
Treasury, Commissioner of Internal Revenue, and the Commissioner of
Social Security.  We are also providing copies to the Director,
Office of Management and Budget, and interested congressional
committees.  Copies will be made available to others upon request. 

Major contributors to this report are listed in appendix IV.  If you
have any questions, please call me at (202) 512-7215. 

Sincerely yours,

Joseph F.  Delfico
Director, Income Security Issues


EXECUTIVE SUMMARY
============================================================ Chapter 0

Each year, the social security trust funds are credited with revenues
derived from income taxes paid on social security benefits.  But do
they get the right amount?  GAO examined how the Department of the
Treasury and the Internal Revenue Service (IRS) ensure that the trust
funds receive the revenues due from taxing social security benefits. 
After work began, the House Ways and Means Committee's Subcommittee
on Oversight asked GAO to pursue several related questions that are
also addressed in this report. 


   BACKGROUND
---------------------------------------------------------- Chapter 0:1

Social security benefits became subject to income taxes beginning in
1984.  When the Congress made benefits taxable, it decided that the
revenues derived from the tax should be credited to the social
security trust funds to strengthen their long-term solvency.  Under
the tax provision, taxable gross income for social security
beneficiaries must include up to one-half of their social security
benefits when their income exceeds certain thresholds:  $32,000 for a
married couple filing jointly and $25,000 for single individuals and
married persons who live apart at all times during the tax year.  The
threshold is zero dollars for married persons filing separately who
live with their spouses at any time during the year.  The amount of
social security benefits to be included in a taxpayer's income is
determined by totaling their (1) adjusted gross income, (2) interest
from tax-exempt sources, and (3) one-half of their social security
benefits.  The total of these three amounts is then compared to the
appropriate threshold.  If their income exceeds the relevant
threshold, the taxpayer must include in taxable gross income the
lesser of one-half of their social security benefits or one-half of
the excess of their income over the threshold. 

From 1984 through 1991, Treasury credited about $29 billion from
taxed social security benefits to the trust funds.  The amount of
taxes transferred to the trust funds rose over the years from about
$2.1 billion in 1984 to over $5.9 billion in 1991.  For several
reasons, this revenue source will take on greater significance in the
future as several factors combine to make even more benefit dollars
taxable. 

First, the Congress did not index the income thresholds that must be
met before benefits are taxed.  As a result, a growing percentage of
social security recipients will have their benefits taxed as their
income levels rise from inflation and growth in real wages. 

Second, the number of beneficiaries has continued to grow since
benefits became taxable.  In 1984, the Social Security Administration
(SSA) was paying benefits to about 36.5 million beneficiaries.  By
the year 2005, it will pay benefits to an estimated 50 million
persons.  With the retirement of the baby boom generation beginning
around the year 2010, the beneficiary rolls will grow even more
rapidly.  This increasing number of beneficiaries will be paid
increasing amounts of benefits that will be taxable. 

Finally, new legislation will greatly increase the amount of benefits
subject to tax.  Beginning in 1994, the maximum proportion of social
security benefits subject to income tax will increase from 50 to 85
percent.  The higher maximum tax rate applies to those beneficiaries
with incomes over an additional set of thresholds:  $34,000 for
single and $44,000 for jointly filed tax returns.  The additional
revenues from this new legislation are to be credited to the Hospital
Insurance trust fund. 


   RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:2

The social security trust funds' revenues could be increased by
recognizing additional taxes identified through IRS' efforts to
locate underreported taxable income and by better detection of
underreported tax-exempt interest. 

When Treasury adjusts the quarterly advances made to the trust funds
to reflect actual tax liabilities, it does not consider the results
of IRS assessing additional taxes on benefits when it identifies
underreported income through its information-matching program. 
Recognizing these additional taxes could have increased the trust
funds by more than $200 million in tax revenue and investment income
for tax years 1984 to 1989. 

In addition, because IRS does not receive reports from payers of
tax-exempt income, it cannot routinely detect underreported amounts
that could affect taxes owed.  Tax-exempt income must be considered
when beneficiaries determine how much of their benefits are taxable. 
This type of income is often earned by social security beneficiaries. 
Comparison of tax-exempt income reported on tax returns with earnings
estimates developed using data from the Federal Reserve and the
Investment Company Institute indicates that taxpayers may have
underreported an estimated $7.2 billion in tax-exempt income on their
1989 tax returns.  However, tax-exempt income only affects taxes for
certain taxpayers, and data are not available to determine how much
additional tax may result from reporting all tax-exempt income. 


   PRINCIPAL FINDINGS
---------------------------------------------------------- Chapter 0:3


      REVENUES FROM AUDITS NOT
      CREDITED
-------------------------------------------------------- Chapter 0:3.1

The Treasury collects federal taxes and manages the investment of
revenues for the social security trust funds.  It advances revenues
to the trust funds each quarter based on its estimate of individual
income tax liabilities related to social security benefits. 
Adjustments to correct errors in estimating are made after tax
returns have been filed and processed. 

Treasury's adjustments are based on the actual social security
benefit tax liabilities reported to IRS on individual tax returns. 
Basically, IRS calculates what the income tax liability for each
taxpayer would have been if their social security benefits had not
been taxable.  The difference between the tax liabilities reported on
tax returns and the recalculated tax liabilities represents the
amount of revenue attributable to taxing benefits. 

Treasury's methodology for estimating and transferring revenues from
taxed benefits to the trust funds does not consider the results of
IRS' underreporter tax compliance program.  In the underreporter
program, IRS verifies that taxpayers properly recognized their
taxable income by comparing tax returns with information returns
(separate reports of income it has received from the payers of that
income).  One of the comparisons IRS makes is to determine whether
taxpayers recognized taxable social security benefits on their
returns. 

For tax years 1984 to 1989, IRS assessed additional taxes, interest,
and penalties of about $225 million against 340,000 taxpayers whose
social security benefits were the primary underreporting issue. 
Limitations of IRS' accounting system make it impossible to identify
how much of that amount was just for underpaid taxes on benefits. 
Under the law, Treasury shall adjust the amount transferred to the
trust funds to recognize the additional taxes assessed. 


      CONTROLS ARE WEAK
-------------------------------------------------------- Chapter 0:3.2

Social security recipients must add their tax-exempt interest income
to their adjusted gross income to determine how much of their
benefits are subject to income tax.  IRS cannot verify that taxpayers
are properly making this calculation because it does not receive an
information return from paying institutions for this type of income. 
Without that information, the effectiveness of its underreporter
program in administering the tax on benefits is diminished. 

Third party information returns are a fundamental control in tax
administration.  With them IRS can detect underreported income and
assess additional taxes.  GAO compared aggregate data from the
Federal Reserve and Investment Company Institute on household
investments in tax-exempt securities with reports of this income on
tax returns.  The comparison indicates that taxpayers may not have
reported about $7.2 billion in tax-exempt interest in 1989. 

How this affected tax revenue from social security benefits cannot be
accurately determined because tax-exempt income only affects the
taxes of some taxpayers.  The revenue implications of this control
weakness will be greater in the future because of (1) a greater
proportion of benefits being subject to tax, (2) dramatic growth in
the number of persons who will be getting benefits, and (3) the
interaction of inflationary pressures on benefit levels with the
static income thresholds that make benefits taxable. 

Establishing this additional control would entail incremental costs. 
IRS estimated that its additional processing costs would be about 20
cents for each new information return.  These costs can be mitigated
by reporting tax-exempt interest as an additional item on an existing
information return. 

Representatives from the financial services industry did not favor
reporting tax-exempt income to IRS.  They said that reporting
earnings from unregistered and original issue discount bonds (about
10 percent of the tax-exempt market) would be very difficult because
the industry lacks the information needed for reporting.  They
acknowledged that reporting tax-exempt income to IRS would improve
tax compliance.  However, they said that because tax-exempt income
affects the tax liability of only some taxpayers, it would be
burdensome to the industry to report the earnings for all investors. 

GAO believes the reporting burden can be mitigated.  The industry
already reports earnings to investors in registered securities.  It
simply does not give IRS a copy.  Also, reporting tax-exempt earnings
as a distinct item on an existing information return, rather than
requiring a separate return for tax-exempt income, would lessen the
reporting burden. 


   RECOMMENDATIONS
---------------------------------------------------------- Chapter 0:4

To more accurately credit the social security trust funds with
revenues from taxing benefits, GAO recommends that Treasury revise
its process for adjusting the advances to the trust funds to
recognize the additional tax liabilities assessed because of IRS
detection of underreported taxes due on social security benefits. 
GAO also recommends that IRS conduct a pilot test to better estimate
the benefits and costs of reporting tax-exempt income.  If
cost-beneficial, IRS should take appropriate steps to routinely
acquire this information. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 0:5

The Commissioners of Social Security and Internal Revenue and
Treasury's Office of Tax Analysis (OTA) provided written comments on
a draft of this report.  The Commissioner of Social Security fully
agreed with the recommendations.  She said that with the number of
individuals entitled to benefits growing each year, it is imperative
that amounts credited to the trust funds accurately reflect all
social security taxes.  (See pp.  17 and 28.)

OTA said that it is aware that IRS can identify certain amounts of
additional income taxes owed because of the underreporting of social
security benefits.  It concluded, however, that no additional
transfers to the social security trust funds should be made for
underreported tax liability on social security benefits until that
liability can be more precisely measured.  GAO disagrees and believes
it is important and feasible to pursue solutions to this funding
issue.  (See pp.  17 and 18.)

The Commissioner of Internal Revenue said that IRS is interested in
measuring the extent to which taxpayers are complying with the tax on
social security benefits, particularly with the major changes in the
amount of benefits taxable beginning in 1994.  She indicated that IRS
will use its 1994 Taxpayer Compliance Measurement Program as a
vehicle to study compliance levels related to the underreporting of
tax-exempt interest.  (See p.  28.)


INTRODUCTION
============================================================ Chapter 1

The Social Security Administration (SSA) paid about $290 billion in
benefits to more than 40 million individuals in 1992.  Since 1984,
when beneficiaries' income exceeds specified levels, a portion of
their benefits has been taxable.  The Department of the Treasury
collects the revenue from taxable benefits and distributes it to the
social security trust funds.\1

Between 1984 and 1991, over $29 billion from income taxes on social
security benefits has been credited to the trust funds.  We examined
how Treasury administers this tax and its method for determining the
amount of tax revenues owed to the trust funds.  Also, at the request
of the House Subcommittee on Oversight, we considered the effect of
unreported tax-exempt interest on taxes owed by social security
beneficiaries. 


--------------------
\1 Revenues are credited to the Federal Old-Age and Survivors
Insurance, Federal Disability Insurance, and Social Security
Equivalent Benefit Account (Railroad Retirement) trust funds. 
Beginning with tax year 1994, the Federal Hospital Insurance Trust
Fund will also receive certain revenues from taxable benefits. 


   HISTORY OF BENEFIT TAXATION
---------------------------------------------------------- Chapter 1:1

When the Social Security Act was enacted in 1935, it did not address
the income tax treatment of social security benefits.  In 1938 and
1941, Treasury's Internal Revenue Service (IRS) (known then as the
Bureau of Internal Revenue) issued several administrative rulings
that concluded that program benefits would not be taxable. 
Essentially, the rulings said that social security benefits were like
gifts, made to aid the general public welfare.  Thus, taxing benefits
would defeat the underlying purposes of the Social Security Act. 

In the late 1970s and early 1980s, the financial condition of the
social security trust funds, which finance program operations, had
greatly deteriorated.  In 1975, program expenditures began to exceed
annual tax revenues for the first time.  It was predicted that unless
remedial action was taken, the Social Security system would be unable
to pay benefits on time after July 1983. 

To deal with the program's financial crisis, the Congress enacted the
Social Security Amendments of 1983.  The amendments required a number
of actions that reduced expenditures and increased revenues as a way
to correct the trust funds' solvency problem.  One of the revenue
enhancement provisions of the amendments made a person's social
security benefits taxable when specified income thresholds were
exceeded.  The Congress mandated that the revenues from this tax be
returned to the social security trust funds.\2

The taxing of social security benefits was designed to be a
significant source of income to correct the financial problems of the
trust funds.  It was estimated that taxing benefits would generate 26
percent of the revenues needed to resolve Social Security's long-term
financial problems. 


--------------------
\2 For an in-depth discussion of the history and issues related to
taxation of social security benefits, see Social Security:  Issues in
Taxing Benefits Under Current Law and Under Proposals to Tax a
Greater Share of Benefits, Congressional Research Service Report for
Congress (89-40 EPW, Jan.  12, 1989). 


   HOW THE TAX IS ADMINISTERED
---------------------------------------------------------- Chapter 1:2

Treasury is responsible for the government's financial operations. 
In regard to the tax on social security benefits, Treasury is
required to (1) make quarterly estimates of individual income tax
liabilities attributable to the benefits paid and (2) transfer these
estimated amounts to the trust funds.  Subsequently, Treasury uses
actual tax return data to correct estimating errors. 

Treasury's Office of Tax Analysis (OTA) is responsible for making the
estimates and authorizing the adjustments to the trust funds.  IRS
administers the federal income tax system and assists OTA by
providing actual tax return data. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
---------------------------------------------------------- Chapter 1:3

We began this assignment with the primary objective of examining the
methods Treasury uses to estimate the transfer of tax revenue from
social security benefits to the trust funds.  In addition, we wanted
to review how Treasury ensures that taxpayers properly recognize
their tax liability for benefits and credits the trust funds with
associated tax revenues. 

Subsequently, the House Ways and Means Subcommittee on Oversight
asked us to pursue several issues related to tax-exempt income that
must be considered in determining the amount of benefits subject to
tax.  Specifically, we were asked to determine (1) to what extent
unreported tax-exempt income results in avoidance of the tax on
social security benefits; (2) whether this is an area where an
increase in the amount of tax revenues lost because of underreported
income is to be expected; and (3) what additional compliance measures
would improve IRS' ability to properly assess and collect the tax on
social security benefits. 

We met with officials of OTA and IRS at their headquarters offices in
Washington, D.C.  We also held meetings with representatives of IRS'
regional and district offices in Philadelphia and with officials at
SSA headquarters in Baltimore.  Additionally, we discussed aspects of
this issue with officials from the Federal Reserve's Office of
Statistics, the Bureau of the Census' Office of Statistical Research,
and the National Bureau of Economic Research.  We held discussions
and meetings with representatives from the private sector, including
associations such as the Public Securities Association and the
Depository Trust Corporation in New York City.  We also held a
roundtable discussion with business leaders who represented segments
of the financial services industry (banking, investment brokers,
mutual fund managers, paying and enrolled agents). 

To identify revenue related to taxable social security benefits, we
reviewed and analyzed statistical information and computer files on
tax returns from IRS.  This assignment was conducted between March
1993 and February 1994 in accordance with generally accepted
government auditing standards. 


TREASURY SHOULD CREDIT TRUST FUNDS
WITH ADDITIONAL REVENUES FROM
TAXED BENEFITS
============================================================ Chapter 2

The social security trust funds are not credited with all revenues
from income taxes on social security benefits.  In determining how
much revenue is owed to the trust funds from taxing benefits,
Treasury does not consider additional taxes that result from IRS
assessments against taxpayers underreporting their income on
individual returns.  Shortcomings in IRS' accounting system make it
difficult to precisely measure the revenue not credited to the trust
funds.  However, the best available data indicate that the trust
funds may have lost an average of $37 million per year in additional
taxes from 1984 to 1989 (more than $200 million for just these tax
years).\3 Under the law, Treasury should consider these additional
taxes on benefits when it credits the trust funds. 


--------------------
\3 Because of lags between the time returns are filed and the time
they are audited, data on assessments for later years were not
available. 


   TRUST FUNDS SHOULD BE CREDITED
   FOR ASSESSED TAXES
---------------------------------------------------------- Chapter 2:1

The 1983 Social Security Amendments require Treasury to transfer to
the trust funds "amounts equivalent to the aggregate increase in tax
liabilities" attributable to the taxation of social security
benefits.  Each quarter, Treasury estimates the income tax revenues
attributable to taxable social security benefits and credits this
estimated amount to the social security trust funds.  These estimates
are made using a tax model developed by Treasury's OTA. 

Very briefly, this model uses as its basis IRS' Statistics of Income
(SOI) data from 1985.\4 The data are extrapolated to recent years
based on several types of growth projections and the model has a "tax
calculator" to compute changes in tax liabilities over time.  These
calculations are then weighted to represent the universe of
taxpayers.  The model also employs certain other information from SSA
and the Census Bureau that allows estimates of taxable SSA income to
be distributed among appropriate groups of taxpayers.\5

With IRS' assistance, Treasury annually adjusts estimated amounts
transferred to the social security trust funds based on actual tax
returns.  First, IRS calculates the income tax liabilities from all
the individual returns that had social security benefits included in
adjusted gross income.  For these returns, IRS calculates what the
income tax liabilities would have been if social security benefits
were excluded from adjusted gross income.  The difference between tax
liabilities of the two calculations is the amount of tax revenue
attributable to taxing social security benefits. 

The amount of the adjustment is determined by comparing the
previously transferred estimated amount for the tax year to IRS'
actual calculation of revenues from the taxing of benefits. 
Treasury's method for crediting revenue to the trust funds is
permissible under the act.\6 However, it does not consider additional
tax liabilities that IRS assesses on social security benefits after
examining filed tax returns through its underreporter program. 

IRS' underreporter program is a comprehensive electronic examination
of all tax returns.  IRS matches available information
returns--independent reports of income filed by payers of that
income--with income that individuals reported on their tax returns. 
The purpose of these matches is to ensure that taxpayers recognize
all taxable income on their returns.  In cases where unreported
income is discovered, IRS attempts to resolve discrepancies. 
Ultimately IRS assesses taxes and may also assess penalties and
interest on taxpayers who have underpaid taxes. 

Among the many matches that IRS makes in its underreporter program is
a comparison of the amount of benefits that SSA reported to it
against the individual tax returns of social security beneficiaries. 
During the period from 1984 to 1989, IRS identified almost 340,000
social security beneficiaries who underreported the amount of their
social security benefits subject to income tax.  As shown in table
2.1, IRS assessed about $225 million in additional taxes, penalties,
and interest on taxpayers in cases where social security was
considered the primary issue in dispute. 



                          Table 2.1
           
            Potential Increased Trust Fund Revenue
           Based on Results of IRS Tax Assessments
                   for Underreported Income

                    (Dollars in millions)

                                                      Potent
                                                         ial
                                                      cumula
                                                        tive
                                    From cases where   trust
                                     social security    fund
                                    benefits are the  increa
Tax year                               primary issue      se
----------------------------  ----------------------  ------
1984                                           $44.9   $44.9
1985                                            33.5    78.4
1986                                            33.0   111.4
1987                                            35.0   146.4
1988                                            42.6   189.0
1989                                            35.7   224.7
------------------------------------------------------------
Source:  IRS tax assessment data from its underreporter program. 

It is important to understand that table 2.1 estimates the potential
increases in trust fund revenues for 1984 through 1989 that would
have resulted if Treasury's process recognized IRS assessments for
underreported taxable benefits.  Several factors make it impossible
to determine the precision of this estimate from available data. 

First, IRS does not precisely track additional tax assessments by the
many different elements of income and deductible expenses that a tax
return comprises.  When it detects multiple problems involving the
underreporting of income on a tax return, IRS attributes any
additional tax assessment to the primary element where a reporting
problem was detected.  Thus, a portion of the $225 million may
represent assessments not related to social security benefits. 
Likewise, other assessments for underreported taxable benefits
probably have been made, but cannot be identified because these cases
are associated with a different primary issue.  Table 2.1 assumes
that these factors cancel each other. 

Second, IRS cannot determine how much of the $225 million in
assessments is for unpaid taxes versus interest and penalties.  This
distinction is important because the trust funds are only entitled to
revenues from the assessments for unpaid taxes.  This limitation
tends to overstate the estimated trust fund losses.  However, table
2.1 reduces the overstatement to some degree because it ignores the
investment income that the trust funds lost by not having access to
these revenues.  In effect, the table assumes that the interest
charged for underpaid taxes is roughly equivalent to the lost
investment opportunity.  Thus, only the penalty portion of the
assessments remains unaccounted for. 

Third, IRS statistics on assessments may not reflect the final
outcome of the assessments.  Taxpayers receiving assessment notices
have an opportunity to explain to IRS that a mistake has been made. 

One way to deal with these shortcomings in IRS' accounting system
would be to examine a representative sample of cases that are
assessed additional taxes.  From such a sample, a ratio could be
developed to determine the amount of assessments related to just
unpaid taxes on benefits.  This ratio could be used to allocate
assessments from the underreporter program to the trust funds.  There
may be other methods that IRS can explore to estimate how much of the
assessments from underreporting taxable social security benefits are
owed to the trust funds. 

We discussed with Treasury officials why no consideration was given
to IRS assessments for underreported taxable benefits when it makes
adjustments to the estimated amounts transferred to the trust funds. 
The officials told us that Treasury was not aware that IRS had
sufficient information to identify the amount of additional taxes
assessed on social security benefits. 


--------------------
\4 The SOI file shows aggregate tax return data for each line item on
tax returns.  The data (number of returns and amount of dollars
reported) are stratified by income levels of taxpayers.  The data are
developed from analyses of randomly selected tax returns.  The sample
is used to project overall tax data on a national level. 

\5 A detailed description of Treasury's methodology for estimating
benefit taxation is in its annual report to the Congress, Report on
the Taxation of Social Security and Railroad Retirement Benefits. 

\6 The act does not specify how Treasury is to make the adjustment in
amounts credited to the trust funds.  However, in the legislative
history of the provision, the Senate Finance Committee stated:  "A
final determination of the amount required to be transferred for a
year may [underscoring supplied for emphasis] be based on an estimate
derived from the appropriately weighted sample of individual income
tax returns for that year...."


   CONCLUSION
---------------------------------------------------------- Chapter 2:2

The social security trust funds are not being credited with all
revenues related to the taxing of social security benefits.  The
trust funds are not receiving revenues derived from the assessment of
taxes resulting from the underreporting of income that affects the
amount of taxpayers' taxable benefits.  Although the current
accounting approach is not inconsistent with the legislative history,
Treasury can add assessments of additional tax liabilities identified
by IRS when it examines tax returns.  Considering these assessments
would provide a better accounting of revenues owed to the trust
funds. 

Limitations in IRS' accounting system preclude a precise
determination of additional taxes assessed for underreported taxable
benefits.  However, Treasury can do more to improve trust fund
accountability by crediting the trust funds for this source of tax
revenues.  Revising Treasury's method for crediting the trust funds
to consider assessments for unpaid taxes on benefits could have
increased the trust funds' receipts by over $200 million for just tax
years 1984 through 1989.  The impact on the trust funds will continue
to grow and compound until action is taken to recognize the
additional revenues derived from the assessment of taxes on
underreported taxable benefits. 


   RECOMMENDATION
---------------------------------------------------------- Chapter 2:3

We recommend that the Secretary of the Treasury direct IRS to
identify the amount of additional taxes that have been assessed on
social security benefits through its underreporter program.  Treasury
should use this information to revise its methodology for
transferring to the trust funds revenues derived from taxing social
security benefits. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 2:4

The Commissioner of Social Security and the Director, Office of Tax
Analysis, at Treasury commented on this matter. 

The Commissioner of Social Security stated that she concurred with
our recommendation and added that this flaw in procedures should be
corrected as soon as possible (see app.  III).  SSA also noted that
Treasury should work with IRS to determine the best estimate of
undercrediting of the trust funds for prior tax years and transfer
these amounts (including interest lost on trust fund investments). 
If a reasonable estimate of the prior-year losses can be developed,
we agree that the trust funds should be reimbursed for any revenues
they lost under the prior approach. 

OTA said that it is aware that IRS can identify certain amounts of
income taxes owed because of the underreporting of social security
benefits (see app.  II).  However, OTA suggested that no additional
transfers of revenue be made until the liability can be determined
more precisely.  OTA cited the limitations with IRS data, which we
discussed in the chapter, as the basis for its conclusion.\7

Although we believe it would be preferable to have precise data to
make this adjustment, precise data do not presently exist. 
Furthermore, IRS would probably have to make significant and costly
changes in its accounting system to develop precise data.  However,
we do not believe the current limitations in the underreporting data
suggest that the known funding problem should remain unresolved. 
Solutions can be pursued. 

As noted on page 16, one approach would be for IRS to examine a
representative sample of cases where IRS discovered underreported
social security tax liabilities.  Analysis of these cases would
identify characteristics such as

  the proportion of the assessments that involved underreporting of
     income related only to social security benefits;

  the proportion of the assessments in the sample of cases related to
     just social security tax liability versus interest and
     penalties; and

  the proportion of the tax assessments abated by IRS. 

Statistical sampling is a widely accepted analytical technique that
can give reliable estimates of the true occurrence of events in a
population.  Answers to questions such as these can be used to
estimate, within acceptable ranges of estimating error, the amount of
additional tax liability IRS discovers in its underreporter program
that affects social security tax liabilities. 

We recognize that this approach has its limitations and is not the
ideal solution.  However, in our opinion, it is a reasonable approach
to resolve a known problem that may have cost the trust funds $225
million over the period from 1984 to 1989.  It must be remembered
that the Congress intended to provide revenue from taxing social
security benefits to the trust funds, not the general fund.  Under
the current situation, this intent is not being adequately fulfilled. 

Finally, OTA stated that the report suggests overcoming the
underreporter program data limitations by assuming that the interest
charged for underpaid taxes is roughly equivalent to the lost
investment opportunity.  OTA's comment refers to a section in the
chapter where we estimated the revenue lost to the trust funds
because IRS tax assessments for underreported tax liabilities for
social security benefits were not considered.  Our intent, however,
was simply to present a reasonable estimate of the losses to the
trust fund, not to propose that the estimating method we used be
adopted for official use. 


--------------------
\7 Essentially the limitations refer to the inability to determine
how much of the assessments from the underreporter program where
social security benefits are the primary underreporting issue is for
just underreported taxes on benefits.  As discussed in the chapter,
the amount that IRS assessed also may include interest and penalties
for underreporting and taxes assessed for underreporting other types
of income. 


CONTROLS TO COLLECT TAXES ON
BENEFITS CAN BE STRENGTHENED
============================================================ Chapter 3

Legal limitations preclude IRS from maximizing the use of one of its
most valuable controls to ensure the accurate payment of taxes on
social security benefits.  IRS does not receive independent reports
of the amount of tax-exempt interest that individuals earn.  Yet,
tax-exempt income, often earned by social security beneficiaries,
must be counted when determining how much of a taxpayer's benefits
are subject to income taxes. 

Comparing various sources of aggregate data on tax-exempt income with
earnings reported on tax returns indicates that taxpayers
underreported about $7.2 billion in tax-exempt income in 1989. 
However, the lack of information on individual tax-exempt earnings,
coupled with the fact that tax-exempt income does not affect
everyone's tax liability, makes it very difficult to estimate the
amount of tax revenues lost to the trust funds. 

Without better information on tax-exempt earnings, equitable
administration of the tax on social security benefits is not ensured. 
With dramatic growth expected in the number of persons paying tax on
their benefits and the amount of benefits taxable, the significance
of the problem and annual trust fund losses will likely grow in the
future. 


   HOW TAXPAYERS DETERMINE THE
   AMOUNT OF BENEFITS TAXABLE
---------------------------------------------------------- Chapter 3:1

Taxpayers have to go through two basic steps to determine the amount
of their social security benefits subject to income tax.  First, they
must determine if their "countable income" exceeds one of three
income thresholds based on their tax-filing status.\8 Since 1984,
these thresholds have been $32,000 for a joint tax return and $25,000
for persons who file as single, head of household, or qualifying
widow(er), or who are married but living separately during the entire
year.  The threshold is zero dollars for persons who are married,
living together for any part of the year, and filing separate
returns. 

If countable income does not exceed the applicable income threshold,
no portion of a taxpayer's social security benefits is taxable. 
However, if countable income exceeds the threshold, a portion of
their social security benefits becomes taxable.  The taxable amount
is the lesser of (1) 50 percent of the taxpayer's benefits or (2) 50
percent of the countable income that exceeds the income threshold.\9

Table 3.1 demonstrates how this threshold and tax calculation operate
for three different income situations related to joint tax returns. 
In each case, the taxpayers are assumed to have $16,000 in annual
social security benefits.  We have varied other aspects of their
income to show how much of their benefits must be included in their
adjusted gross income. 



                          Table 3.1
           
           Examples of How Taxpayers Establish the
             Amount of Their Benefits Subject to
                          Income Tax


                                                Less
                                                than
                                                half    Half
                                                  of      of
                                      Benefi  benefi  benefi
                                      ts not      ts      ts
                                      taxabl  taxabl  taxabl
Tax item                                   e       e       e
------------------------------------  ------  ------  ------
Adjusted gross income (before         $20,00  $27,00  $48,00
 threshold test)                           0       0       0
Tax-exempt interest income            $1,000  $2,000  $4,000
Half of social security benefits      $8,000  $8,000  $8,000
Countable income                      $29,00  $37,00  $60,00
                                           0       0       0
Threshold for joint return            $32,00  $32,00  $32,00
                                           0       0       0
Excess over threshold                     $0  $5,000  $28,00
                                                           0
Benefits taxable (lesser of half of       $0  $2,500  $8,000
 benefits or half of excess over
 threshold)
Percent of total benefits taxable          0    15.6      50
------------------------------------------------------------

--------------------
\8 Countable income is calculated by adding tax-exempt income and
half of a taxpayer's social security benefits to adjusted gross
income. 

\9 In 1993, the Congress increased the proportion of benefits subject
to taxation under certain conditions.  The details of this change are
discussed later in this chapter (see pp.  25-26). 


   IRS LACKS A CONTROL TO
   ROUTINELY DETECT UNDERREPORTED
   TAX-EXEMPT INTEREST
---------------------------------------------------------- Chapter 3:2

Detecting unreported income is one of IRS' fundamental goals in tax
administration.  Information returns, independent reports of income
that taxpayers and IRS receive from payers of the income, are one of
IRS' most valuable controls to detect underreported income.  By
matching income reported on information returns with tax returns, IRS
identifies millions of taxpayers that may have underreported income. 
For example, in 1987 IRS identified almost 18 million potential
underreporters.  More than 6 million of these cases were sent to IRS
tax examiners and about $1.2 billion in additional taxes, interest,
and penalties were assessed. 

IRS does not require information reporting with respect to tax-exempt
interest.  Although tax statutes require information returns be filed
for many types of income such as wages, dividends, tax refunds, and
interest, IRS explained that tax-exempt interest is specifically
exempted from reporting under one tax statute.  And, by long-standing
administrative ruling, IRS has not required the reporting of
tax-exempt income under another tax statute that grants it broad
authority to require the reporting of income equal to $600 or more. 

After social security benefits became taxable, IRS revised tax
returns to provide space for taxpayers to voluntarily report the
amount of their tax-exempt income.  For tax year 1989, individuals
reported receiving about $37.6 billion in tax-exempt interest on
their tax returns.\10

Although $37.6 billion in tax-exempt income was reported on tax
returns, IRS has no way to determine if this is the total amount of
tax-exempt income earned by taxpayers.\11

We compared reported tax-exempt earnings on 1989 returns with various
other sources of data on this type of income maintained by the
Federal Reserve and the Investment Company Institute (ICI).  Although
these bodies have some data on tax-exempt earnings, the data are
built to varying degrees on surveys and estimates.  As a result,
precise data on tax-exempt investments and tax-exempt income are
unavailable. 

The Federal Reserve data identify direct investment by households in
tax-exempt securities.  But the data cannot distinguish between
institutional and household investment in tax-exempt securities
invested through mutual and money market funds. 

ICI's data are developed from voluntary reports by its membership,
which accounts for about 95 percent of the industry assets invested
in all types of mutual funds.  However, when ICI surveys its members
to gather data on investment holdings, not all of its members
respond.  For example, in 1989 ICI members holding only 63.8 and 73.4
percent of the accounts in money market and short-term municipal bond
funds, respectively, reported to ICI.  ICI extrapolated these data to
prepare its investment data. 

Using a combination of these data, an estimate of tax-exempt income
earned by households can be made.  Table 3.2 estimates that about
$7.2 billion in tax-exempt interest was underreported.  We do not
know how precise this estimate may be because of the very limited
reporting of data on tax-exempt income.  For example, given the error
associated with the IRS estimate, the amount of underreported
tax-exempt interest ranged from $6.2 billion to $8.2 billion. 
Additional variation in the estimate results from uncertainty
associated with the Federal Reserve and ICI data and the average
interest rate we applied. 



                          Table 3.2
           
              GAO Estimate of Underreported Tax-
               Exempt Interest in Tax Year 1989

                    (Dollars in billions)

----------------------------------------------------  ------
Aggregate household investment in tax-exempt securities\a
------------------------------------------------------------
Direct holdings                                       $495.8
Mutual fund & money market holdings                    135.6
============================================================
Total holdings                                        $631.4
Average interest rate\b                                 7.1%
Estimated tax-exempt household income                  $44.8
IRS estimate of reported tax-exempt income             $37.6
                                                       (+$1)
Estimated unreported tax-exempt income                  $7.2
                                                       (+$1)

------------------------------------------------------------
\a The aggregate data on investments combine information from the
Federal Reserve (direct holdings) and ICI (mutual funds and money
market holdings).  In addition, direct household investments
represents an average investment amount for the year.  It was
developed by averaging year-end values for the stated categories for
1988 and 1989.  This average understates the value of tax-exempt
securities because in 1988 the Federal Reserve data did not include
the value of certain tax-exempt bonds.  Also, some of these data were
developed through sampling of issuances at the local level.  Thus,
the estimate of aggregate household investment is subject to error. 
Because of the nature of the data, however, we do not have sufficient
information to calculate the range of error. 

\b The 7.1-percent interest rate was obtained from the Public
Securities Association.  The association represents brokers and
underwriters who deal in municipal investments.  It developed the
interest rate from data prepared by Moody's Investor Services, Inc. 

Source:  Statistics of Income Individual Income Tax Returns 1989;
Flow of Funds Accounts, Federal Reserve System, ICI (Mar.  1992); and
Moody's Investor Services, Inc. 


--------------------
\10 At the time of our review, tax year 1989 was the most recent year
for which complete SOI data on filing were available. 

\11 The IRS data for reported tax-exempt income are developed from
its SOI file.  SOI data are based on stratified probability samples
of about 100,000 income tax returns filed with IRS.  Being based on
samples, SOI data are subject to a range of error at a given
confidence level.  The $37.6 billion in tax-exempt income has a +2.8
percent error rate at the 95-percent level of confidence.  Thus, IRS
is 95-percent certain that the true value of reported tax-exempt
interest lies between $36.6 billion and $38.6 billion. 


   WHAT EFFECT DOES THE LACK OF AN
   INFORMATION RETURN HAVE? 
---------------------------------------------------------- Chapter 3:3

Institutions are not required to submit information returns to report
the tax-exempt income they pay to investors.  Consequently, important
information needed to reliably estimate the revenue the trust funds
lose because of underreported tax-exempt income is not available. 
For example, SOI data indicate that social security beneficiaries
often receive tax-exempt income.  In 1989, about 60 percent of the
reported tax-exempt income was earned by taxpayers receiving social
security benefits.  However, it is not known if the underreported
tax-exempt income goes to

  social security beneficiaries who may have an incentive to
     underreport because their tax-exempt income could place them
     over the taxing threshold,

  beneficiaries whose income is already high enough to make their
     benefits fully taxable without considering tax-exempt income, or

  persons not affected by the tax. 

Only underreported tax-exempt income earned by social security
beneficiaries whose taxes would be increased would raise trust fund
revenues.  We are unable, however, to reliably estimate the revenue
losses from underreporting of tax-exempt income by this group.  The
information is not available to determine the portion of all
underreporting attributable to this group as a whole and then
reliably distribute amounts of the underreported income among
individual taxpayers. 

Beyond the revenue implications, questions of tax equity are also
associated with this shortfall in IRS' controls.  Tax-exempt interest
was made an element in the tax calculation to ensure equitable
treatment among social security beneficiaries.  The intent was to
prevent wealthier beneficiaries from avoiding the tax on their
benefits by investing in tax-exempt securities. 

With an information return for tax-exempt interest, IRS could be more
certain that taxpayers are accurately determining how much of their
social security benefits is taxable.  Taxpayers who invest in
tax-exempt securities would be less able to escape taxes on their
benefits.  Presently, it is only through costly detailed audits,
which are limited in number, that IRS can identify some taxpayers who
are not reporting tax-exempt interest. 


   SEVERAL FACTORS WILL CAUSE
   REVENUE LOSSES TO GROW IN THE
   FUTURE
---------------------------------------------------------- Chapter 3:4

The most compelling reasons for requiring information returns for
tax-exempt income relate to the inevitable growth in the amount of
benefit dollars that will be taxable in the near future.  In the
coming years, the amount of benefits subject to tax and the
proportion of social security beneficiaries paying income taxes on
their benefits will greatly increase.  The increase is attributable
to several factors: 

  the structure of the tax on social security benefits coupled with
     the way benefit amounts are calculated,

  the anticipated significant growth in the beneficiary population,
     and

  a recent legislative change to the tax calculation that exposes a
     greater amount of benefits to income tax. 


      EFFECT OF WAGE AND INCOME
      GROWTH ON TAXING SOCIAL
      SECURITY BENEFITS
-------------------------------------------------------- Chapter 3:4.1

A growing proportion of beneficiaries is having to pay income tax on
their benefits.  This is occurring because of the interaction of
inflation and real wage growth with the method for calculating the
taxable benefits. 

Workers' annual wages and the amount of annual wages covered under
the social security programs are increasing.  Consequently, future
retirees will receive higher benefit amounts reflecting these higher
average lifetime earnings, which are the basis of their social
security benefit amount.  Likewise, current beneficiaries receive
annual cost of living increases to adjust for inflation, which result
in progressively higher benefits. 

The Congress did not, however, index the income thresholds that must
be met to make social security benefits taxable.  Consequently, an
increasing proportion of beneficiaries has modified adjusted gross
incomes that exceed the static income thresholds.  For example, in
1987 the Congressional Budget Office estimated that 14 percent of the
beneficiaries paid taxes on their benefits.  Our analysis shows that
in 1989 about 20 percent of the beneficiaries paid taxes on their
benefits. 


      EFFECT OF RETIRED POPULATION
      GROWTH
-------------------------------------------------------- Chapter 3:4.2

The aging of the nation's population also compounds the potential
revenue consequences.  The number of beneficiaries is expected to
double over the next 40 years, primarily because of the retirement of
the baby boom generation (persons born between the end of World War
II and the mid-1960s).  Around the year 2010, the baby boom
generation will begin reaching retirement age.  The number of
beneficiaries will increase by millions at this time and billions
more in benefit dollars will become taxable. 

Beyond exposing more benefit dollars to the tax, this dramatic growth
in the beneficiary population will also greatly increase the
importance of the tax as a program revenue source.  Program
operations are primarily supported by the taxes workers and their
employers pay on wages.  With the baby boom generation in retirement,
the ratio of workers to beneficiaries is expected to drop.  This
means that revenue derived from the tax on benefits will play a
larger role in meeting existing program obligations.  Table 3.3 shows
the expected changes in number of beneficiaries, benefits paid, and
worker-to-beneficiary ratios over the next 75 years. 



                         Table 3.3
          
            Growth in Beneficiaries and Benefits
                      Paid, 1993-2070

                                         Worker-     Total
Calend           Covered  Beneficiar         to-  benefits
ar               workers         ies  beneficiar  (billion
year          (millions)  (millions)     y ratio      s)\a
------  ----------------  ----------  ----------  --------
1993               135.0        41.9         3.2   $ 309.1
1995               139.0        43.4         3.2     343.2
2000               145.8        46.7         3.1     452.3
2005               151.8        49.9         3.0     603.5
2010               156.9        53.8         2.9     814.9
2015               159.8        59.6         2.7   1,147.8
2020               161.1        66.7         2.4   1,645.6
2025               161.9        73.8         2.2   2,320.1
2030               163.0        79.2         2.1   3,174.3
2035               164.6        82.5         2.0   4,213.6
2040               166.1        83.7         2.0        \b
2045               167.1        84.9         2.0        \b
2050               167.8        86.4         1.9        \b
2055               168.2        88.8         1.9        \b
2060               168.6        91.1         1.9        \b
2065               169.2        92.9         1.8        \b
2070               169.7        94.4         1.8        \b
----------------------------------------------------------
\a Benefit amounts are in current dollars--not adjusted for
inflation. 

\b Benefits payable for years after 2035 were not available. 

Source:  1993 Annual Report of the Federal Old-Age and Survivors
Insurance and Disability Insurance Trust Fund (Washington, D.C.: 
GPO, Apr.  7, 1993). 


      NEW LEGISLATION MAKES MORE
      BENEFIT DOLLARS TAXABLE
-------------------------------------------------------- Chapter 3:4.3

As a way to reduce the annual budget deficit, the Congress passed the
Omnibus Budget Reconciliation Act of 1993.  The act increases the
maximum proportion of a taxpayer's social security benefits subject
to income tax from 50 to 85 percent beginning in 1994 for taxpayers
over new income thresholds.  In the future, taxpayers will have to
compare their countable income to two sets of thresholds when
determining how much of their benefits are taxable.  The additional
revenues from this new legislation are to be credited to the Hospital
Insurance trust fund. 

For example, if a married taxpayer filing a joint return has
countable income of between $32,000 and $44,000, no more than 50
percent of their benefits will be taxable.  However, if their
countable income is over $44,000, up to 85 percent of their benefits
will be taxable.  Similar changes exist for a single taxpayer.  The
changes in taxing of benefits take effect in 1994 and will affect the
taxes of an estimated 3 million higher-income recipients.  Table 3.4
depicts the new tax thresholds and tax rule. 



                          Table 3.4
           
               New Social Security Benefit Tax
            Provisions Effective in Tax Year 1994

                                                      Maximu
                                                           m
                                                      propor
                                                        tion
                                                          of
                                                      benefi
                                                          ts
                                                      taxabl
                                                           e
                                                      (perce
Tax filing status          Countable income              nt)
-------------------------  -------------------------  ------
Single                     $25,000-34,000                 50
                           over $34,000                   85
Married joint              $32,000-44,000                 50
                           over $44,000                   85
------------------------------------------------------------

   PRACTICALITY OF REQUIRING
   INFORMATION RETURNS
---------------------------------------------------------- Chapter 3:5

Before implementing any additional internal control, the costs of
requiring information returns for tax-exempt income need to be
considered relative to the benefits.  We spoke to both IRS and
representatives of various sectors of the financial services industry
about anticipated costs associated with reporting and processing
additional income information. 

IRS officials said there would be incremental costs for receiving and
processing information returns.  For example, the Office of the
Assistant Commissioner for Examination estimated that the incremental
cost of receiving and entering data from these returns was about 20
cents per return.  These costs could be mitigated, however, if the
tax-exempt income was reported on an existing information return
rather than requiring a new, separate information return. 

We held a roundtable discussion with members of the financial
services industry (banking, investment brokers, mutual funds, paying
agents, and enrolled agents).  Industry representatives told us that
reporting on unregistered bearer bonds and original issue discount
(OID) bonds would be difficult if not impossible.  Basically, the
financial services industry has no information on investors who hold
bearer bonds and little information on OIDs.  However, they noted
that these types of securities constitute an unknown, but assumed to
be small and declining, percentage of the tax-exempt market--probably
less than 10 percent. 

The industry representatives said that reporting tax-exempt income
would improve IRS' tax compliance capabilities.  However, they
believed reporting for all investors would impose a processing burden
on the industry.  They reasoned that many unnecessary information
returns would be required for recipients whose tax liability would
not be changed by this type of income. 

We believe that this burden is tempered by the fact that earnings
information is already generated and provided to most investors in
tax-exempt securities.  However, a copy of it is not given to IRS. 
We recognize that tax-exempt income does not affect the taxable
income of all persons investing in these securities.  However, there
are presently over 40 million social security recipients, many of
whom invest in tax-exempt securities, and a growing proportion will
be subject to the tax as inflation and real wage growth drive up
benefits and their incomes relative to the taxable income thresholds
that remain constant. 


   CONCLUSION
---------------------------------------------------------- Chapter 3:6

Controls are vital to effective financial administration, especially
when they directly involve the receipt and expenditure of revenues. 
IRS' controls over tax revenues owed on social security benefits are
weak because it lacks information returns on tax-exempt interest. 
Without these returns, IRS cannot verify that more than 40 million
beneficiaries accurately calculate the taxes owed on their benefits. 
Consequently, equitable administration of the tax on social security
benefits cannot be ensured and an unknown amount of program revenue
is being lost.  This shortcoming in IRS' controls for administering
this tax may be especially serious in the future because of the
anticipated dramatic growth in both (1) the amount of benefits
subject to the tax and (2) the government's reliance on it to meet
future program obligations. 


   RECOMMENDATION
---------------------------------------------------------- Chapter 3:7

Given the uncertainty of the tax revenue losses from underreporting
of tax-exempt income and the financial industry's potential
processing burden, we recommend that IRS conduct a detailed study of
tax returns to better identify the benefits of having payers report
tax-exempt income.  In addition, IRS should obtain data on the costs
of reporting and processing such information.  The study could
involve IRS' 1994 Taxpayer Compliance Measurement Program (TCMP) or a
pilot study that solicits the cooperation of several payers so that
the benefits and costs of reporting tax-exempt income can be
estimated.  If a favorable cost-benefit ratio is identified, IRS
should take appropriate steps to make it possible to routinely
acquire this information. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 3:8

In its written comments on a draft of the report, IRS generally
agreed that further study is needed of how tax-exempt interest
affects the tax on social security benefits.  Specifically, IRS
stated it is interested in measuring the compliance and noncompliance
levels with respect to taxable social security benefits.  IRS said
the major changes effective for 1994 in the threshold and amount of
benefits subject to income tax make this the appropriate year to
study. 

Before it commits to conducting a pilot test, IRS said it will use
its 1994 TCMP to measure the benefits of reporting.  In the 1994
TCMP, IRS plans to examine 92,000 Form 1040 returns.  A segment of
these returns will be used to measure compliance with the tax on
social security benefits.  Under TCMP all income, deduction, and
exemption items on the returns will be examined, allowing an analysis
of the levels of compliance related to underreporting of tax-exempt
interest.  Results of this survey will be published and appropriate
program changes and legislative proposals will be recommended. 

While this approach appears reasonable, we caution that assessing
underreporting of tax-exempt interest by using TCMP will be
difficult.  TCMP uses information returns to identify noncompliance
in the reporting of income and allowable deductions, but examiners
are to probe beyond such reports in determining tax liabilities. 
Because information returns do not exist for tax-exempt income, IRS
examiners will have to perform specific audit work to detect such
income.  If TCMP requires this type of work, then IRS may be able to
reliably assess compliance with the reporting of tax-exempt income
and how noncompliance affects the tax on social security benefits. 

TCMP will not, however, provide any information on the costs of
reporting and processing information on tax-exempt income.  Thus,
some additional study of this matter will be needed.  Given IRS'
commitment to study this issue in the 1994 TCMP, we have revised our
recommendation to reflect its plans. 

In a letter dated June 20, 1994, the Commissioner of Social Security
said that she concurred with our recommendation for a pilot test. 
She noted that the recommendation has the potential to increase trust
fund revenues if the test shows that collection of tax-exempt income
data is cost-beneficial. 




(See figure in printed edition.)Appendix I
COMMENTS FROM THE INTERNAL REVENUE
SERVICE
============================================================ Chapter 3



(See figure in printed edition.)




(See figure in printed edition.)Appendix II
COMMENTS FROM THE DEPARTMENT OF
THE TREASURY, OFFICE OF TAX
ANALYSIS
============================================================ Chapter 3



(See figure in printed edition.)




(See figure in printed edition.)Appendix III
COMMENTS FROM THE SOCIAL SECURITY
ADMINISTRATION
============================================================ Chapter 3



(See figure in printed edition.)



(See figure in printed edition.)


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix IV

Roland H.  Miller III, Assistant Director, (410) 965-8925
James A.  Slaterbeck, Evaluator-in-Charge
William J.  Staab, Senior Evaluator
Thomas Bloom, Technical Assistance
William F.  Schmanke, Technical Assistance

