Tax Administration: IRS' Partnership Compliance Activities Could Be
Improved (Letter Report, 06/16/95, GAO/GGD-95-151).

Pursuant to a congressional request, GAO reviewed the Internal Revenue
Service's (IRS) partnership compliance activities, focusing on: (1) the
extent of partnership compliance with tax laws; and (2) steps IRS is
taking or could take to improve partnership compliance.

GAO found that: (1) the most current partnership compliance data showed
that individual partners owed an additional $2.4 billion in taxes in
1982; (2) IRS will not have more current partnership compliance data
until October 1998 when it completes its audits of tax year 1994
partnership returns; (3) IRS relies exclusively on audits to detect
partnership noncompliance, which has proven to be an inadequate
strategy; (4) relatively few partnership returns have been audited in
recent years because IRS has focused its resources on corporate audits;
(5) partnership audits have resulted in audit adjustments more
frequently than other business audits; (6) IRS could not measure the
amount of net taxes assessed per partnership because it did not maintain
data on the specific results of partnership audit adjustments; (7) IRS
discontinued its delinquent partnership return program in 1989, but it
plans to reinstate the program in 1996; (8) IRS does not have a computer
document matching program for partnerships, but may develop one in its
modernization efforts that will be completed in 2001; and (9) IRS is
training revenue agents to identify partnership compliance issues, since
it takes an average of 12 months to complete each partnership audit.

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

 REPORTNUM:  GGD-95-151
     TITLE:  Tax Administration: IRS' Partnership Compliance Activities 
             Could Be Improved
      DATE:  06/16/95
   SUBJECT:  Tax administration
             Tax nonpayment
             Tax law
             Computerized information systems
             Taxpayers
             Voluntary compliance
             Tax return audits
             Data collection operations
             Associations
             Income taxes
IDENTIFIER:  IRS Taxpayer Compliance Measurement Program
             IRS Audit Information Management System
             IRS Business Master Plan
             IRS Industry Specialization Program
             IRS Partnership TCMP Survey
             
**************************************************************************
* 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 Joint Committee on Taxation, Congress of the United
States

June 1995

TAX ADMINISTRATION - IRS'
PARTNERSHIP COMPLIANCE ACTIVITIES
COULD BE IMPROVED

GAO/GGD-95-151

IRS' Partnership Compliance Activities


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

  AIMS - Audit Information Management Systems
  IRS - Internal Revenue Service
  RTF - Returns Transaction File
  SOI - Statistics of Income
  TCMP - Taxpayer Compliance Measurement Program
  TEFRA - Tax Equity and Fiscal Responsibility Act of 1982
  TRA - Tax Reform Act of 1986
  TSM - Tax Systems Modernization

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


B-260093

June 16, 1995

The Honorable Bill Archer
Chairman
The Honorable Robert Packwood
Vice Chairman
Joint Committee on Taxation
Congress of the United States

In fiscal year 1992, the number of partnerships totaled about 1.5
million--about the same as in 1981.  Yet, during this time period,
the number of partners nearly doubled from 9.1 million to 15.7
million, partnership assets increased from $1 trillion to $1.9
trillion, and partnership net income went from a net loss of $4.1
billion to a net gain of $42.9 billion.\1 Tax law changes made in the
1980s affected partnerships' reportable income as well as the
Internal Revenue Service's (IRS) approaches to administering
partnerships' compliance with the tax laws. 

This report responds to your predecessors' request that we determine
(1) the extent of partnership compliance with federal tax laws, (2)
any steps IRS is taking to improve partnership compliance, and (3)
any additional efforts that IRS could take to improve partnership
compliance. 


--------------------
\1 Dollar figures are expressed in 1992 constant dollars. 


   BACKGROUND
------------------------------------------------------------ Letter :1

A partnership is an association of two or more entities or persons
that join together to carry on a trade or business, with each partner
contributing money, property, labor, or skill and each partner
expecting to share in the profits and losses.  Partnerships are not
taxed directly on their income.  Rather, all income (or loss),
credits, and other tax-related items "flow through" to the partners,
who are to report their individual shares on their own returns.\2
Partners can be individuals, corporations, other partnerships, or
virtually any other legal entity. 

Partnerships are required to file a Form 1065, U.S.  Partnership
Return of Income, Credits and Deductions, Etc., which includes the
Schedule K that shows the partnership's income or loss, deductions,
and credits that are to be distributed to the partners.  Partnerships
also are required to file Schedule K-1, Partner's Share of Income,
Credits and Deductions, Etc., which shows each partner's separate
share of the total partnership business activity. 

Because partnership income and deductions flow through to partners,
partnerships can provide an attractive structure for sheltering
income from tax.  A tax shelter is an investment in which a
significant portion of the investor's return consists of tax benefits
that not only offset any tax liability that might otherwise arise
from the investment, but also "shelter" other income of the investor,
usually from the investor's business or professional activities. 
This tax sheltering aspect led to the development of abusive tax
shelters during the 1970s and early 1980s, which generally took
advantage of loopholes in the law in ways that were not intended by
Congress.  IRS generally defines an abusive tax shelter as one that
lacks economic reality or viability when viewed in its entirety. 

Legislation enacted by Congress over the past decade has
significantly reduced the potential to use a partnership as a tax
shelter.  The Tax Equity and Fiscal Responsibility Act of 1982
(TEFRA) introduced (1) special partnership audit mechanisms to
streamline the audit process for partnerships and (2) various
penalties to deter the promotion of and investment in abusive tax
shelters.  For example, section 6700 of the Internal Revenue Code,
added by TEFRA, provides a nontaxpayer penalty to those who promote
abusive tax shelters. 

Of various substantive provisions directed at tax shelters, the
at-risk rules and the passive activity loss rules are the most
prominent.\3 The at-risk rules of section 465 of the Code limit the
deductibility of losses from an activity to the amount the taxpayer
has at risk with respect to the activity.  The passive activity loss
rules found in section 469 of the Code, added by the Tax Reform Act
of 1986 (TRA), provide the broadest attack on tax shelters by
preventing a taxpayer from using net losses incurred in a passive
activity from offsetting income derived from nonpassive activities. 


--------------------
\2 The amount of partnership loss a partner may deduct is limited by
basis rules, at-risk limitations, passive activity limitations, and
limitations applicable to specific deductions. 

\3 For this purpose, the term "passive activity" is defined as any
activity that involves the conduct of a trade or business in which
the taxpayer does not materially participate and rental activities. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :2

The current extent of partnership tax compliance is unknown.  IRS'
most current partnership compliance data were collected under its tax
year 1982 partnership Taxpayer Compliance Measurement Program
(TCMP).\4 These data showed that partnerships underreported their net
income by $13 billion in 1982 which we estimate resulted in an
underpayment of taxes by partners approaching $3.6 billion.  Even
when partnerships reported all of their income, partners sometimes
failed to include it in their own tax returns.  Thus, IRS estimated
that individual partners owed an additional $2.4 billion in taxes in
1982.  But significant tax law changes in the intervening years make
these data unreliable indicators of the present situation.  IRS will
not have more current partnership compliance data until October 1998
when its TCMP audits of tax year 1994 partnership returns are
scheduled to be completed. 

We found that IRS' compliance strategy for addressing partnership
compliance relied almost exclusively on audits to detect
noncompliance.  The strategy did not include either a nonfiler or
computer document matching component.  IRS did have a limited
document matching program to identify partners who do not report
partnership income on their individual income tax returns. 

We made several observations concerning the audit program. 

  In recent years, relatively few partnership returns were audited
     because IRS focused its business audit resources on taxable
     entities, such as corporations.  In fiscal year 1992, IRS
     audited 0.5 percent of the partnership returns filed and changed
     63 percent of those audited, making positive income adjustments
     of $4.2 billion.  The partnerships appealed about $3.5 billion,
     or 84 percent, of the adjustments, and the results are still
     pending on $2.4 billion. 

  Partnership audits were not as productive as other types of
     business returns when measured by the percent of returns audited
     that resulted in audit adjustments, which is one measure that
     IRS uses to determine audit productivity.  About 37 percent of
     the returns audited were closed without any change to the
     partnership return, while the no-change rate on corporate
     returns was 23 percent.  The partnership no-change rate may be
     higher because the formula used to select returns for audit was
     developed from 1982 TCMP data, while the corporate formula was
     based on 1987 TCMP data.  Also, we found that the partnership
     audits may be more productive than indicated by the no-change
     rate because adjustments made to partners' returns that did not
     change the partnership return were considered no-change audits. 

  IRS' primary measure of audit productivity--the amount of net taxes
     assessed per hour of audit time--could not be used for
     partnership audits because IRS did not have data on the
     additional taxes partners were assessed or refunded as a result
     of partnership audit adjustments.  These data would help IRS to
     more fully evaluate audit productivity and could be developed as
     IRS modernizes its computer systems. 

  Until a new partnership audit selection formula is developed, IRS
     could analyze current partnership audit results for leads to the
     types of partnership returns that are more likely to be adjusted
     during audits. 

IRS did not have an active program to detect partnerships that
stopped filing required returns.  It discontinued this program in
1989 to concentrate its nonfiler efforts on taxable business returns,
such as corporation and employment tax returns.  IRS plans to
reinstate a delinquent return program for partnerships in calendar
year 1996. 

While IRS used information returns it received for individuals in a
computer document matching program, it did not have a similar program
for information returns it received for partnerships.\5 IRS already
processes information returns that can be used in a partnership
document matching program.  We could not estimate the total cost for
a matching program because IRS is in the process of modernizing its
computer and other tax administrative systems, which would affect
these costs.  IRS plans to complete its modernization projects by the
year 2001.  IRS would need to begin planning now if it decides to
have a partnership document matching program in place by that time. 

In its 1991 individual document matching program, IRS processed about
12 percent of the Schedules K-1 it received and matched them against
partners' income tax returns.  The match resulted in additional tax
assessments of $6.3 million.  We estimated that at an additional cost
of $18.6 million to IRS, about $219.5 million in additional taxes may
have been assessed had IRS matched all of the schedules.  As part of
a fiscal year 1995 compliance initiative, IRS plans to transcribe an
estimated 4.1 million Schedules K-1 and, after October 1995,
determine whether more schedules should be transcribed.  We found
that all schedules would need to be transcribed if IRS is to have the
compliance system it envisions by the year 2001.  In this regard, IRS
would need to find ways to reduce data transcription costs, such as
developing a schedule that can be electronically scanned into the
computer. 

Since it has conducted relatively few partnership audits in recent
years, IRS began a study in 1994 to collect data for training revenue
agents on how to identify partnership compliance issues.  For the
study, IRS began audits of about 220 partnerships in February 1995. 
The agents are to be trained by October 1995 at which time
partnership TCMP audits are scheduled to begin.  Our evaluation of
IRS' study plan showed that to meet its training and TCMP audit
schedule, IRS would need to supplement the study with data from
recently completed audits because it takes, on average, about 12
months to complete audits of the types of partnerships included in
the study.  IRS is now collecting these data. 


--------------------
\4 Under TCMP, IRS examiners do line-by-line audits of tax returns on
a probability sample taken for the entire return population.  IRS
uses TCMP data to determine compliance levels, develop formulas for
objectively selecting returns for audit, and allocate audit
resources. 

\5 IRS has a computer matching program to identify individual
taxpayers who potentially underreported their taxable income or
failed to file returns.  Third parties, such as banks and other
businesses, are required to file annual information returns to report
various payments made to or by individuals and partnerships.  IRS
matches amounts on information returns against amounts reported on
individual tax returns, but not partnership returns. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
------------------------------------------------------------ Letter :3

Our objectives were to determine (1) the extent of partnership
compliance with federal tax laws, (2) any steps IRS is taking to
improve partnership compliance, and (3) any additional efforts that
could be taken to improve partnership compliance. 

To determine the extent of current partnership compliance, we
reviewed data from the last partnership TCMP, which IRS conducted on
returns filed in 1982.  Using these data, we estimated the tax effect
of partnerships underreporting their tax year 1982 income.  We also
obtained data from IRS' tax gap estimates on the estimated amount of
taxes taxpayers owed on income received from partnerships, but did
not report on their individual income tax returns. 

To determine what steps IRS is taking to correct partnership
compliance, we analyzed IRS' Audit Information Management System
(AIMS) database on cases closed in fiscal year 1992 to determine
audit results by various characteristics.  We also reviewed the
examination workpapers from 177 partnership audits to determine the
types of compliance issues identified by IRS.  These audits were
randomly selected from the 8,229 audits completed in fiscal year
1992.  The results of our analyses are not projectable to the
universe.  We excluded tax shelters because TRA removed the major
incentive for using partnerships as tax shelters.  We analyzed tax
year 1990 Statistics of Income (SOI) data on partnerships and IRS'
1990 and 1991 Returns Transaction File (RTF) data for partnership
audits closed in fiscal year 1992 to identify and compare
characteristics of the universe of partnerships with characteristics
of audited partnerships.  Appendix I discusses this analysis and its
limitations.  We also met with IRS officials to discuss their plans
for addressing partnership compliance issues and how these plans fit
into IRS' new compliance strategies. 

To determine what additional efforts IRS could undertake to improve
partnership noncompliance, we analyzed the examination workpapers
from 177 partnership audits to determine whether the audit issues
could have been identified and resolved through computer matching. 

We did our work in IRS' San Francisco District Office; the Fresno
Service Center; and Washington, D.C., headquarters in accordance with
generally accepted government auditing standards between June 1993
and December 1994.  We requested comments on a draft of this report
from the Commissioner of Internal Revenue.  On April 26, 1995, IRS
officials provided us IRS' comments.  IRS representatives at that
meeting included the Assistant Commissioner for Examination and the
National Director, Service Center Compliance.  Their comments are
summarized on page 19 and incorporated in this letter where
appropriate. 


   THE CURRENT EXTENT OF
   PARTNERSHIP TAX COMPLIANCE IS
   UNKNOWN
------------------------------------------------------------ Letter :4

The latest compliance data available on partnerships are from IRS'
1982 Partnership TCMP Survey.  These data showed that partnerships
understated their net income and investment income by $13 billion in
1982, which was about 26 percent of these income types that should
have been reported.  IRS did not estimate the amount of additional
taxes associated with this unreported income because it did not
collect tax assessment data on the partners who were liable for the
taxes.  To develop such an estimate, IRS officials agreed that it
would be reasonable to assume that the $13 billion was taxable at the
28-percent marginal tax rate for individuals for 1982.  Using this
rate, we estimated the partners would have potentially owed about
$3.6 billion in taxes had partnerships reported this income to
them.\6

In its tax gap study, IRS did have estimates on the taxes owed by
individuals on income reported to IRS by partnerships, but not
reported by partners on their tax returns.\7 IRS estimated that in
1982 this amount was about $2.4 billion in taxes.  Therefore, the
total potential tax loss associated with partnership noncompliance
may have been about $6.0 billion in 1982. 

The unreported income estimated in the 1982 partnership TCMP is not a
reliable indicator of unreported partnership income for later years. 
Over 10 years have elapsed since these data were collected--a time
period in which significant tax law changes have affected
partnerships.  TEFRA and TRA made major changes to partnership law. 
Under TEFRA, the tax treatment of partnership items is to be
determined by IRS in one consolidated audit and the results applied
automatically to the partners.  TRA removed most incentives for
individual partners to use partnerships for tax reduction purposes. 
IRS plans to include partnerships in its 1994 TCMP survey, but does
not plan to collect data on audit adjustments made to partners' tax
returns.  Without this information, IRS will not be able to estimate
the tax effect of partnership noncompliance.\8


--------------------
\6 While we believe this estimate provides a reasonable indication of
the underpaid taxes attributable to partnership noncompliance, it is
far from being precise.  We do not know the tax position of the
individual partners and whether additional partnership income would
have increased their tax liability. 

\7 IRS computes a partnership tax gap as a part of its tax gap
estimates for individuals.  The tax gap is the difference between
income taxes owed and income taxes voluntarily paid.  The partnership
tax gap does not include taxes on income that partnerships fail to
report, but reflects the tax effect of individual partners' not
reporting income reported to them by partnerships on Schedule K-1. 

\8 In our report Tax Compliance:  Status of the Tax Year 1994
Compliance Measurement Program (GAO/GGD-95-39, Dec.  30, 1994), we
proposed that IRS collect such data, but we were told that it would
be too costly to collect.  We suggested that as an alternative IRS
could apply the marginal tax rate for individuals to the unreported
partnership income found in the 1994 TCMP. 


   IRS HAD A LIMITED PARTNERSHIP
   COMPLIANCE PROGRAM
------------------------------------------------------------ Letter :5

Generally, three enforcement tools can be used in a comprehensive
taxpayer compliance program.  The first is an audit program that
selects for audit those taxpayers most likely to be noncompliant. 
The second is a program that identifies taxpayers who fail to file
required returns.  And third is a computer matching program that
matches information returns against tax returns to identify taxpayers
who may have underreported their income.  Data and resource
limitations prevented IRS from making full use of these tools in its
partnership compliance program.  As a result, its partnership
compliance program was limited. 

Almost all of IRS' partnership compliance activities consisted of
audits.  While less than 0.5 percent of partnership returns were
audited by revenue agents in fiscal year 1992, about 37 percent of
the returns audited did not result in adjustments to partnership
returns.  This no-change rate was higher than that for other business
return audits.  One reason for the higher partnership no-change rate
may be that IRS neither had current compliance data nor gathered data
on audit results to help it target noncompliant partnerships.  Also,
IRS did not use its limited resources for an active partnership
nonfiler program and a computer document matching program for
determining whether income reported on information returns was
reported on partnership returns.  IRS did use about 12 percent of the
Schedules K-1 it received in its tax year 1991 matching program for
individual taxpayers to determine whether individual partners
reported their partnership income. 


      PARTNERSHIP AUDIT PROGRAM
      WAS LESS ACTIVE AND
      PRODUCTIVE THAN OTHER
      BUSINESS AUDIT PROGRAMS
---------------------------------------------------------- Letter :5.1

From fiscal year 1988 to fiscal year 1992, audit coverage of
partnerships decreased by almost one-half, while it more than doubled
for corporations and remained relatively unchanged for nonfarm sole
proprietors.  IRS has tended to focus its business audit resources on
taxable entities, such as corporations.  As a result, the number of
partnership audits has steadily declined.  Table 1 shows the audit
rates for partnerships, corporations, and nonfarm sole proprietors
for fiscal years 1988 to 1992. 



                           Table 1
           
            Revenue Agent Audit Rates for Business
              Returns, Fiscal Years 1988 to 1992

                     (Numbers in percent)

                                                     Nonfarm
                                                        sole
                          Partnershi  Corporatio  proprietor
Fiscal year                        p           n           s
------------------------  ----------  ----------  ----------
1988                            0.93        1.33        2.35
1989                            0.72        2.02        2.13
1990                            0.80        2.59        2.04
1991                            0.63        2.36        2.14
1992                            0.50        2.89        2.06
------------------------------------------------------------
Source:  IRS Annual Reports. 

In fiscal year 1992, IRS audited 8,229 partnership returns as
compared to 75,797 corporate returns and 135,099 nonfarm sole
proprietor returns.\9 Overall, IRS made positive income adjustments
of $4.2 billion to 4,653, or 57 percent, of the partnership returns
it audited in fiscal year 1992.  However, $3.5 billion, or 84
percent, of the adjustments were appealed.  At the time of our
review, about $1 billion of the adjustments had been settled in
appeals, and the remaining $2.5 billion were still open.  We could
not determine the final settlement amounts because IRS tracked these
appealed cases by adjustments made to partners' returns, but not to
the partnerships' returns. 

While adjustments were made to 63 percent of the returns audited,
partnership audits were less productive than audits of other business
entities when measured by the no-change rate.  The no-change rate is
the percentage of audits conducted where no changes were made to the
return as a result of the audit.  Our analysis of IRS data showed
that 3,019, or 37 percent, of the 8,229 partnership returns audited
in fiscal year 1992 were no-change audits.  In contrast, the
no-change rate was 23 percent for corporations and 12 percent for
nonfarm sole proprietors. 

One reason for the relatively high partnership no-change rate may be
that the formula IRS used to select returns for audit was based on
1982 partnership TCMP results.  On the other hand, the formula used
to select most corporation returns for audit was based on the 1987
corporate TCMP, and the nonfarm sole proprietor selection formula was
developed from the 1988 individual TCMP.  According to IRS officials,
the older the audit selection formula becomes, the less reliable it
is for predicting returns most likely to be adjusted.  IRS data
showed that since the current formula was first used for 1986
partnership returns the no-change rate remained constant at about 18
percent for 1986 through 1988.  In 1989, the no-change rate increased
to 21 percent and then reached the 30-percent range in 1990, 1991,
and 1992. 

We found that partnership audit productivity could be reported as
being higher than reflected by the current no-change rate because IRS
reports the results of a partnership audit as a no-change when the
adjustments do not change the partnership return, but do change the
partner's(s') return.  For the 85 no-change audits reviewed, we found
7 examples where this practice occurred.  For example, in one case a
partnership with two partners claimed a loss of over $60,000 on its
return.  Partner A, who had taxable income, was allocated 100 percent
of the loss.  Partner B had no taxable income for the year.  The
revenue agent determined that the allocation was improper because it
was done solely to shift losses to the partner with taxable
income.\10 The revenue agent reallocated the loss to both partners. 
Partner A was assessed over $10,000 in additional tax.  The
reallocation of loss did not change Partner B's tax position. 

IRS' primary measure of audit productivity is the amount of net
additional tax assessed per hour of audit time.  For partnerships,
this measure of productivity is not readily available because they
are not taxable entities.  Further, IRS' Audit Information Management
System (AIMS), which contains audit result data, did not have data on
the additional taxes partners were assessed or refunded as a result
of audit adjustments made to partnership returns. 

To have better measures of partnership audit productivity in terms of
the no-change rate and tax assessments per hour of examination time,
IRS would have to collect and analyze data that is not readily
available on AIMS.  Under the Tax Systems Modernization (TSM)
initiative, IRS plans to have the computer capability to collect
these data on AIMS so that it can better measure partnership audit
productivity. 


--------------------
\9 The actual number of partnerships audited was lower than the
number of returns audited because IRS audited several years of
returns filed by a partnership.  For fiscal year 1992, there were
5,565 partnerships associated with the 8,229 partnership returns that
were audited. 

\10 Internal Revenue Code section 704(b) requires there be a
"substantial economic effect" to the distribution of partnership
gain, loss, deduction, or credit.  A partnership may not allocate
partnership gains or losses to partners merely to reduce the tax
burden of one of the partners. 


      ANALYSIS OF PARTNERSHIP
      COMPUTER FILES COULD HELP
      IRS TARGET ITS PARTNERSHIP
      AUDIT RESOURCES
---------------------------------------------------------- Letter :5.2

IRS does not plan to develop new partnership audit selection formulas
until late 1998 when its tax year 1994 TCMP audits are scheduled to
be completed.  In the meantime, IRS could consider using other
partnership data to help target its audit resources toward those
partnerships with the greatest audit potential.  IRS has at least
three files that contain useful partnership data that could be used
for this purpose--AIMS, which contains data on audit results; RTF,
which contains all of the data that are transcribed from partnership
returns; and SOI files, which contain more data than found on RTF,
but are from a representative sample of all partnership returns filed
in a year. 

We analyzed these files to determine whether the characteristics of
partnerships with audit adjustments were different from those of
partnerships where IRS made no adjustments.  Our analysis was limited
because we did not have comparable tax year data for all three files. 
We had fiscal year 1992 AIMS data, tax years 1990 and 1991 RTF data,
and tax year 1990 SOI data.  Appendix I discusses the methodology we
used for our analysis and its limitations. 

Using these files, we compared characteristics of partnerships where
IRS proposed audit adjustments with characteristics of partnerships
where no audit adjustments were proposed.  Our analysis of the SOI
data showed that partnership income generally fell into three
categories:  (1) ordinary income from trade or business, (2) income
from rental real-estate activities, and (3) portfolio or investment
income. 

Our analysis of AIMS and RTF data showed that for audits closed in
fiscal year 1992, the most productive audits in terms of total income
adjustment per hour of examination were those where one source of
income was 90 percent or more of total reported income and the
partnerships were not tax shelters but were subject to TEFRA audit
procedures.\11 For the 3,547 cases for which we had data, about 18
percent of the audited partnerships fell into this category.  These
cases accounted for about 42 percent of total dollars adjusted and 23
percent of the partnership examination hours.  This analysis suggests
that more productive audits may have occurred if resources had been
directed to those partnerships subject to TEFRA audit procedures that
had 90 percent or more of their income from any one source. 

We recognize the limitations of our analysis and do not expect IRS to
change its selection criteria on the basis of the results.  Instead,
we made our analysis to demonstrate how IRS' computerized files may
be used to better target its partnership audit resources.  In this
regard, IRS has established 31 District Office Research and Analysis
sites throughout the United States to do compliance research at the
district and regional office levels, which were formed to do these
types of analyses. 


--------------------
\11 Generally, these procedures cover partnerships with 10 or more
partners and partnerships with 1 or more nonindividual partners
(i.e., a corporation or another partnership).  TEFRA requires that
IRS follow certain notification procedures when conducting these
consolidated audits. 


      IRS DID NOT PURSUE
      PARTNERSHIPS THAT STOPPED
      FILING RETURNS
---------------------------------------------------------- Letter :5.3

One of IRS' strategic objectives for fiscal year 2001, as described
in its Business Master Plan, is to (1) encourage all taxpayers to
voluntarily file timely and accurate tax returns and (2) take
appropriate enforcement action when taxpayers fail to follow this
practice.  However, in 1989, IRS discontinued pursuing partnerships
under its computerized delinquency check program, which identifies
businesses that stop filing required returns.\12 According to IRS
officials, the program was discontinued because IRS wanted to
concentrate its business nonfiler resources on returns that could
produce additional tax revenues directly, such as corporation and
employment tax returns.  The decision to discontinue the delinquency
check program was also based on IRS' policy not to process and match
delinquent Schedule K-1 forms to partners' tax returns.  IRS
officials said that a delinquency check program for partnerships that
did not file tax year 1994 returns will be reinstituted in calendar
year 1996. 

IRS did not have readily available data on the number of partnerships
that stopped filing required returns.  However, our analysis of the
5,565 partnerships whose audits were closed in fiscal year 1992
showed that 556, or 10 percent, had not filed required partnership
returns for tax year 1991.  Of these partnerships, IRS had made over
$655 million in audit adjustments for prior years for 380, or 68
percent, of the partnerships.  Without tax returns for these
partnerships, IRS has no way of knowing whether they are operating in
compliance with the tax laws. 


--------------------
\12 Under the delinquency check program, IRS matches business tax
returns filed to that business' filing requirements recorded on IRS'
business master file.  If the program identifies a business that has
a filing requirement, but no corresponding return, the business is to
be classified as a nonfiler and sent a series of notices requesting
the return or an explanation as to why the return does not have to be
filed.  If no response to the notices is received, IRS is to attempt
to contact the business by telephone, and the case may later be given
to revenue officers for further investigation. 


      IRS DID NOT FULLY USE
      INFORMATION RETURNS TO
      VERIFY PARTNERSHIP
      COMPLIANCE
---------------------------------------------------------- Letter :5.4

Our analysis of 181 partnership audits completed in fiscal year 1992
showed that information returns were useful for identifying various
types of unreported partnership income.  We found that 25 of 116
partnerships for which income was reported on an information return
did not report over $2 million of the income reported on the returns. 
In one case, information returns issued to a partnership showed gross
stocks or bonds sales of over $1 million.  However, the partnership
did not report a capital gain or loss on its partnership return.  In
a second case, a partnership failed to report a gain or loss from the
gross sale of over $600,000 in stocks or bonds.  According to IRS
officials who reviewed these cases for us, in neither case did the
revenue agent identify the unreported income. 

IRS officials said that auditors routinely receive transcripts of all
information returns associated with partnership returns they are
auditing.  The audit files we reviewed did not contain these
transcripts.  However, we did verify with IRS San Francisco District
Office officials that information return transcripts were provided to
revenue agents for use in partnership audits. 

While IRS' current policy is to use information return data for the
partnerships it audits, a matching program could help identify
underreporting of income by many partnerships not selected for audit. 
For tax year 1991, IRS received 5.9 million information returns
showing payments of $487 billion made to 829,000 partnerships.  About
$277 billion, or 57 percent, of the $487 billion was gross stocks and
bond sales.  The remaining amount of about $211 billion of
information return income represented about 37 percent of the $572
billion in business receipts and investment income reported by the
1.5 million partnerships in 1991.  Yet, IRS has no plans to develop a
computer document matching program to verify partnership income. 

In 1987, IRS took the position that consolidated audit procedures of
TEFRA would effectively preclude an information returns matching
program for partnerships subject to the TEFRA procedures.  These
procedures generally cover partnerships with 10 or more partners and
partnerships with 1 or more nonindividual partners (i.e., a
corporation or another partnership).  IRS procedures for its
individual matching program include assessing additional tax where
the taxpayer agrees with IRS and issuing a statutory notice of
deficiency where the taxpayer does not agree or does not respond. 
The TEFRA procedures do not preclude merely matching information
returns to the partnership return.  However, an IRS contact with a
partnership about a potential underreported income issue could
constitute an audit requiring the TEFRA notice procedures to be
followed before IRS could make any adjustment to the partnership's
reported income.  According to IRS officials, the TEFRA procedures
are extremely complex and have produced procedural and administrative
problems for IRS. 

We found that most partnerships are not subject to the TEFRA audit
rules.  Our estimate is based on tax year 1990 SOI data, which showed
that about 358,000 (23 percent) of 1.6 million partnerships were
TEFRA partnerships.  The remaining 1.2 million (77 percent)
partnerships could have been included in an information returns
matching program under current assessment and statutory notice
procedures.  TEFRA partnerships could be included in a matching
program without adversely affecting most of IRS' partnership audits. 
Most partnerships would be selected for audit before the start of the
matching program.  IRS' individual matching program starts about 16
months after tax returns are required to be filed.  The same time
frames would probably apply to partnerships, and by this time, IRS
would have already selected most of the partnership returns it
planned to audit. 

We did not estimate the potential cost to develop and operate a
partnership document matching program because data were not available
to make reliable estimates.  However, IRS already transcribes Forms
1065 and Schedule K information and receives and processes
partnership information returns so little or no additional data
transcription cost should be incurred.  The unknown costs are those
associated with matching the documents and investigating the
potential underreporter cases.  According to IRS officials, these
costs would be higher in a partnership document matching program than
in the individual document matching program.  They said additional
costs would be incurred to make flow-through adjustments in addition
to costs related to the added complexity of reconciling information
documents to fiscal year accrual basis taxpayers. 

Costs to match and investigate potential underreporter cases are
difficult to estimate because IRS is in the process of developing
various TSM systems that would affect program costs.  These systems
are planned to improve IRS' ability to efficiently match documents
and resolve various compliance issues including potential
underreporter issues.  To take advantage of its modernization
efforts, IRS could begin planning now to have a partnership document
matching program by 2001. 


      IRS DOES NOT TRANSCRIBE ALL
      SCHEDULES K-1 FOR USE IN ITS
      INDIVIDUAL DOCUMENT MATCHING
      PROGRAM
---------------------------------------------------------- Letter :5.5

While IRS did not computer match information returns to verify
partnership income, it did use a limited number of Schedules K-1,
submitted by partnerships in its individual computer document
matching program, to verify that partners correctly reported their
partnership income.  For tax year 1991, 2.5 million (12 percent) of
the 20.5 million Schedules K-1 partnerships filed were processed for
use in this program.  About 52 percent, or 1.3 million of the 2.5
million schedules, were filed electronically, while the remaining 1.2
million were manually transcribed onto IRS computers.\13

For tax year 1991, IRS assessed partners $6.3 million in additional
taxes, penalties, and interest attributable to unreported partnership
income by matching the Schedules K-1 to individual tax returns.\14
IRS estimated that it assessed about $11.80 in additional taxes for
each $1.00 spent on underreporter cases attributable to Schedules
K-1.  On the basis of these data, we estimated that at a total
additional cost of $18.6 million, about $219.5 million in additional
taxes may have been assessed if IRS had transcribed and used in its
matching program for all of the tax year 1991 Schedules K-1 it
received.  Appendix II shows how we derived these estimates. 

The $11.80 to $1.00 revenue-to-cost ratio for matching tax year 1991
Schedules K-1 data to individual tax returns was lower than the tax
year 1990 revenue-to-cost ratio, which was $19.30 to $1.00.  IRS
officials believe the difference in the two revenue-to-cost ratios
may reflect the large differences in the volume of cases worked
during those two periods.  They said IRS worked 855 cases for tax
year 1990, all of which had potential for a high audit adjustment. 
However, they said for tax year 1991, IRS worked 37,789 cases which,
overall, had potential for lower audit adjustments. 

IRS' Business Master Plan for fiscal years 1995 to 1999 contains a
compliance initiative to increase the use of Schedule K-1
information.  Also, as a part of its fiscal year 1995 budget, IRS
received an additional $2.7 million to transcribe and match an
estimated 4.1 million schedules.\15 By October 1995, IRS plans to
establish a workload selection system to increase its use of these
schedules.  After this time, IRS plans to determine how many
additional Schedules K-1 it should transcribe and match. 

To reduce the cost of transcribing additional Schedule K-1 data, IRS
could encourage more partnerships to file the schedules
electronically.  Partnerships currently have the option of filing
Schedules K-1 electronically.  Of the total 17.3 million Schedules
K-1 filed for tax year 1992, 344 partnerships filed 1.7 million (9.8
percent) electronically, which was over 40 times the 41,000
electronically filed tax year 1986 schedules.  Another alternative to
manually transcribing paper documents would be to develop a Schedule
K-1 that can be electronically scanned into the computer.  IRS
already does this for certain tax forms, such as the Form 1040EZ,
Income Tax Return for Single and Joint Filers With No Dependents. 

If more Schedule K-1 data could be computerized, IRS could use these
data to identify potential compliance issues other than unreported
income.  In addition to the partner's share of income, deductions,
and credits, the schedule shows the partner's percent of interest in
gains, losses, and ownership; and, it shows an analysis of the
partner's capital account.  Analysis of this information, together
with information from the Form 1065, could identify potential
inappropriate allocations of partners' gains and losses.  This
analysis could be done by comparing the information on Schedule K-1,
showing the percent of income and loss to which each partner is
entitled, with (1) the amount actually allocated to the partner and
(2) the aggregated amount of income or loss shown on Schedule K. 

It is especially important that all Schedules K-1 be transcribed as
IRS moves forward with its modernization efforts.  IRS envisions that
under TSM it will have an integrated case- processing system by the
year 2001 that would allow the computer to make various compliance
checks (e.g., underreporter and nonfiler).  This system, as planned,
would also allow all compliance issues for a taxpayer to be handled
at the same time.  To be efficient, however, IRS would have to build
cross-reference files to do these checks.  Otherwise, IRS would be
unable to create a comprehensive compliance case on the taxpayer,
making separate audits of a taxpayer's individual, corporate, and
partnership returns more likely.  An integrated system would enable
IRS to audit the taxpayer rather than the return and could result in
more effective and timely audits. 


--------------------
\13 To determine which Schedules K-1 to match, IRS randomly sampled
taxpayers' social security numbers ending in certain digits. 

\14 For tax year 1992, approximately 2 million, or 12 percent, of the
17.3 million Schedules K-1 filed were used in the matching program. 
In contrast, IRS used 100 percent of Forms 1099 information returns. 
The results of matching these documents to individual returns were
not available when we completed our work. 

\15 IRS did not have data on the exact number of partnership
Schedules K-1 that will be transcribed.  IRS officials said that we
could extrapolate from tax year 1991 data which indicated that 65.7
percent of all paper Schedules K-1 received were from partnerships. 
The remaining 34.3 percent were received from S-corporations, trusts,
and estates. 


   IRS' EFFORTS TO CHANGE ITS
   PARTNERSHIP AUDIT STRATEGY
------------------------------------------------------------ Letter :6

In 1994, IRS began changing its partnership audit strategy after IRS
compliance officials determined that some partnerships were being
established solely for the purpose of avoiding taxes.  The Department
of the Treasury issued regulations to help combat this problem. 
Also, IRS began making plans to increase the number of partnership
audits and to train auditors to identify partnership issues when they
conduct operational audits and audits for the 1994 TCMP survey. 

IRS did not have data on the number of partnerships that were
established to avoid taxes.  On the basis of their knowledge of
partnership issues and experience in auditing partnerships, IRS
compliance officials said that some partnerships, such as those that
invest in certain types of financial instruments, were being formed
for this purpose.  Further, to ensure that taxpayers understood that
IRS had the authority to recast transactions that exploit and misuse
the partnership provisions of the Internal Revenue Code to avoid
paying taxes, anti-abuse regulations were issued in December 1994. 
These regulations were designed to make it clear that the literal
application of partnership rules to create losses was inconsistent
with tax laws and regulations and would be questioned by IRS. 

In September 1994, IRS began a partnership study to collect data for
determining how it will revamp its partnership audit program.  In
February 1995, IRS began auditing about 220 partnerships to (1)
identify potential partnership compliance issues, (2) develop new
techniques for auditing partnership returns, and (3) develop a
partnership training program for revenue agents.  These partnerships
were identified from SOI's tax year 1992 partnership data.  About 180
of the returns were for partnerships engaged in the finance,
insurance, and real-estate industries.  Revenue agents having
extensive backgrounds in partnership issues are conducting the
audits.  Information collected during the audits is to be used to
complete a questionnaire and transcribed into a computer database for
analysis.  IRS plans to incorporate the study results into a training
program for revenue agents who are to audit about 12,500 partnerships
as part of the 1994 TCMP survey.  These revenue agents are scheduled
to be trained by September 1995, and the TCMP audits are scheduled to
begin in October 1995. 

To coordinate the study and the partnership audit program, IRS
established a new team under its Industry Specialization Program for
partnership audits.\16 This new partnership team is also to
coordinate audit issues raised under the proposed partnership
anti-abuse regulations. 

It is very important that revenue agents be trained to identify
partnership compliance issues when conducting TCMP audits because it
has been over 10 years since the last partnership TCMP survey. 
However, IRS' schedule for collecting and analyzing data to develop
the training program and train the revenue agents may be optimistic. 
Of IRS' partnership audits completed in fiscal year 1992 that fell
within the description or category of the study's industry types,
more than one-half took over 12 months to complete.  Less than
one-half of them had adjustments.  If this year's audits require
similar time and achieve similar results, IRS may neither have enough
data to develop an effective training program nor have the time to
train the agents. 

To better ensure that it has enough data to meet its TCMP training
and audit schedule, we suggested to IRS officials that they
supplement the study with other audit data.  IRS could evaluate those
recently completed audits that have audit adjustments and are in the
industries being studied.  IRS officials agreed with our suggestion
and on March 30, 1995, informed field staff to gather data on closed
audit cases. 


--------------------
\16 The Industry Specialization Program is an existing national
program to provide oversight and coordination of all issues in a
designated area. 


   CONCLUSIONS
------------------------------------------------------------ Letter :7

In recent years, resource constraints have not allowed IRS to place
much emphasis on partnership compliance activities.  As a result,
partnership audit coverage declined, and IRS discontinued its
partnership delinquency check program.  At the same time, IRS did not
have current compliance data to help it target its partnership audit
resources to returns with compliance issues. 

IRS is taking some steps to address partnership compliance issues. 
For example, it is planning to conduct partnership TCMP audits to
determine the level of partnership compliance and to develop audit
selection formulas.  However, the results of these audits will not be
available until late 1998.  IRS is also in the process of modernizing
the tax system with plans such as developing an integrated
case-processing system that would allow IRS to more effectively and
efficiently identify noncompliant taxpayers.  This system is
scheduled to be in place by 2001. 

In concert with its TCMP and modernization efforts, IRS could begin
now to improve its partnership compliance activities.  IRS could
begin developing plans to modify its audit information management
system so that tax changes made to partners' returns that result from
audits of partnership returns, including allocation issues, could be
entered into the system.  These data would allow IRS to better
evaluate the effectiveness of its audit program.  IRS could also use
its various computer files that contain partnership data to help
develop audit leads until new partnership audit selection formulas
are developed. 

IRS may start a delinquency check for partnerships that did not file
tax year 1994 returns.  This program would be consistent with IRS'
strategic objective of improving voluntary compliance by encouraging
taxpayers to file timely and accurate returns and taking appropriate
enforcement actions when they do not. 

Also, to take full advantage of modernization efforts, IRS could plan
for the orderly development of a partnership computer document
matching program.  Similarly, to meet planned TSM integrated
case-processing needs, IRS could begin to develop ways, such as forms
that can be scanned onto the computer, to enter all Schedules K-1
data onto the computer without having to manually transcribe them. 

IRS is doing a study to gather data so that agents can be trained on
current partnership compliance issues.  However, the study may not
produce enough data in time to meet IRS' training and TCMP audit
schedule.  As a backup, IRS could use data from recently closed
audits so that it can improve its chances of meeting these schedules. 
IRS is now collecting these data. 


   RECOMMENDATIONS
------------------------------------------------------------ Letter :8

We recommend that as IRS moves forward with its modernization
efforts, the Commissioner of Internal Revenue

  develop plans to modify audit management information systems to
     more fully reflect the results of partnership audits by
     including information on the (1) tax assessment on partners'
     income tax returns and (2) changes in allocations of profits and
     losses among partners,

  analyze computer partnership files to develop audit leads and
     select returns for audit,

  reinstitute the delinquency check program for partnerships to
     identify partnerships that do not file required tax returns,

  develop plans for a document matching program using information
     returns to verify partnership income, and

  devise ways to enter all Schedules K-l onto the computer so they
     can be used in the individual computer document matching program
     and for other compliance purposes. 


   AGENCY COMMENTS AND OUR
   EVALUATION
------------------------------------------------------------ Letter :9

We requested comments on a draft of this report from the Commissioner
of Internal Revenue.  Responsible IRS officials, including the
Assistant Commissioner for Examination and the National Director,
Service Center Compliance, provided IRS' comments in a meeting on
April 26, 1995.  The officials generally agreed with our
recommendations and said that: 

  IRS' current information systems are unable to accommodate data on
     partners' income tax assessments and changes in allocating
     partners' profits and losses.  IRS will address the need for
     expanded data on partnerships and partners in its plans to
     modernize information systems. 

  IRS is beginning to use partnership computer files to develop leads
     and select returns to audit through its newly established
     District Office Research and Analysis sites.  In addition, IRS
     will use results from its partnership study started in October
     1994 to select returns for audit.  Ultimately, the data gathered
     from the 1995 TCMP will be used to select potentially more
     productive partnership audits. 

  IRS will reinstate the partnership delinquency check program for
     tax year 1994 in calendar year 1996. 

  IRS will test the feasibility (cost/benefit) of a document matching
     program for non-TEFRA partnerships.  If the test proves
     successful, IRS will develop plans for a non-TEFRA partnership
     matching program as TSM progresses.  IRS did not believe that
     TEFRA partnerships could be included in a matching program
     without adversely affecting IRS' audit program. 

  IRS will consider additional steps that could be taken to more
     fully utilize Schedule K-1 data, including encouraging more
     partnerships to file electronically and determining the
     feasibility of devising a scannable Schedule K-1.  IRS cautioned
     that because of limited resources and its inability to require
     partnerships to file electronically, it may not be able to
     capture data and use all Schedule K-1 information. 

We believe the actions that IRS proposes are responsive to our
recommendations. 


---------------------------------------------------------- Letter :9.1

As agreed with the Committee, we will send copies of this report to
the Commissioner of Internal Revenue and other interested parties. 
We also will make copies available to others upon request. 

This report was prepared under the supervision of Natwar M.  Gandhi,
Associate Director.  Major contributors to the report are listed in
appendix III.  If you have any questions, please contact me at (202)
512-5407. 

Jennie S.  Stathis
Director, Tax Policy
 and Administration Issues


METHODOLOGY FOR ANALYZING
PARTNERSHIP AUDITS CLOSED IN
FISCAL YEAR 1992
=========================================================== Appendix I

This appendix describes our methodology for analyzing partnership
audits closed in fiscal year 1992.  We analyzed these audits to
identify characteristics of audited cases that could help the
Internal Revenue Service (IRS) to better target its partnership audit
resources. 

We obtained IRS' Audit Information Management System (AIMS) data for
all partnership audits closed in fiscal year 1992.  Because return
data for the actual years audited were not available, we obtained
Returns Transaction File (RTF) data for tax years 1990 and 1991 for
nontax shelter and tax shelter partnerships, respectively, closed in
fiscal year 1992.  We used the AIMS and RTF data to identify
characteristics of audited returns.  We used Statistics of Income
(SOI) partnership data for tax year 1990 to identify characteristics
of the nationwide universe of partnerships. 


   AIMS DATA
--------------------------------------------------------- Appendix I:1

The AIMS data contained information on 8,229 audited returns filed by
5,565 partnerships.  The tax years audited ranged from 1970 to 1991
with most of the audited years being 1983 through 1990.  Table I.1
shows the distribution of audited returns by tax year. 



                          Table I.1
           
              Population of Partnership Tax Year
              Audits Closed in Fiscal Year 1992

                                 Tax       Nontax
                             shelter      shelter      Total
                           partnersh  partnership    audited
Tax year                         ips            s    returns
-------------------------  ---------  -----------  ---------
1970 -1980                        21           33         54
1981                              19           32         51
1982                              46           68        114
1983                             530           88        618
1984                             184          108        292
1985                             202          147        349
1986                             238          350        588
1987                             313          715      1,028
1988                             226        1,992      2,218
1989                             143        1,690      1,833
1990                              59          982      1,041
1991                               2           41         43
============================================================
Total                          1,983        6,246    8,229\a
------------------------------------------------------------
\a Number represents audited returns of 5,565 partnerships. 

Source:  IRS' AIMS data for fiscal year 1992. 

To determine which partnerships were tax shelters, we identified all
returns that had a tax shelter source code and then identified the
taxpayer identification number associated with those returns.  If a
partnership had more than one return examined and any one of the
returns had a tax shelter source code, we considered the partnership
to be a tax shelter.  Using this approach, we identified 1,209 tax
shelter partnerships associated with 1,983 returns and 4,356 nontax
shelter partnerships associated with 6,246 tax returns. 

Our computations of audit adjustments and examination hours are
aggregated figures for all returns associated with one partnership. 
The net adjustment amount for each entity determined whether an audit
result was treated as a decrease in income, no change in income, or
an increase in income. 


   RTF DATA
--------------------------------------------------------- Appendix I:2

IRS' AIMS file contains little information about an audited
partnership.  To increase the number of partnership characteristics
we could analyze, we obtained tax year 1990 and 1991 RTF information. 
Return data for the actual years audited were not available.  For
nontax shelters, we obtained tax year 1990 data; and, for tax
shelters, we obtained tax year 1991 data.\1 Because of the
differences in tax years, we limited our analyses to those
characteristics that we believe remain fairly stable:  (1) major
industry, (2) major income source, (3) Tax Equity and Fiscal
Responsibility Act of 1982 (TEFRA) status, and (4) tax shelter
status.  We were able to obtain data for 3,547 (64 percent) of the
5,565 partnerships.  The remaining 2,018 partnerships did not file
tax years 1990 or 1991 returns. 


--------------------
\1 IRS was unable to provide us with tax year 1991 data for all
partnerships in time for use in this report. 


   SOI DATA
--------------------------------------------------------- Appendix I:3

We analyzed the 1990 SOI partnership data to develop estimates of
characteristics of the universe of partnerships.  In analyzing these
data, we used the return elements in the SOI database that matched
directly with elements on RTF. 


   STUDIES OF AUDITED PARTNERSHIPS
--------------------------------------------------------- Appendix I:4

We selected a simple random sample of 200 partnership audits for a
detailed analysis of the examination workpapers.  IRS provided
workpapers for 92 partnership audits where the revenue agent had
recommended adjustments and 85 partnership audits where no audit
adjustments were proposed.  We analyzed these 177 audited partnership
returns to determine why the return had been selected for audit, the
nature of the audit changes, and whether the noncompliance identified
by the revenue agent could be identified through computer matching. 


   ANALYSIS OF PARTNERSHIP FILES
--------------------------------------------------------- Appendix I:5

We did a number of computer analyses of the AIMS, RTF, and SOI files
to determine whether certain characteristics of audited returns with
audit adjustments could be used to identify partnerships that may be
more likely to be adjusted if audited.  Because the data from the
three files are not comparable, our analyses are not intended to
conclusively show where IRS could direct its audit resources. 
Instead, we made our analyses to demonstrate how IRS' computerized
files may be used to better target its partnership audit resources. 
IRS is better able to access its computer files than we are and can
do more detailed analyses. 

We found that analyzing these data by major industry group and
predominant revenue source showed there were differences.  We
analyzed the audited partnerships from the AIMS file with partnership
RTF and SOI data to determine the predominant source of revenue
reported by partnerships on Forms 1065 and Schedules K.  We used IRS'
Principal Business Activity codes to determine in which major
industry groups the partnerships were shown.  We determined how many
partnerships received 90 percent or more of their total revenue from
one of the following categories: 

  Ordinary trade or business activities. 

  Rental real-estate activities. 

  Other rental activities. 

  Investment activities. 

  Other activities. 

We also analyzed the data that showed no positive revenue reported in
any of the above categories. 

As one example, our analysis by major industry group showed that the
distribution of audited partnerships among industry groups generally
paralleled the distribution of partnerships in the universe.  The
only group where there appeared to be a large difference was real
estate.  As shown in table I.2, 44 percent of the partnerships in the
universe were in this industry group compared to 26 percent of the
audited partnerships for which we obtained return information. 



                          Table I.2
           
            Comparison of Universe of Partnerships
            and Partnerships With Audits Closed in
           Fiscal Year 1992 by Major Industry Group


                 Number of               Number of
                partnershi  Percent of  partnershi   Percent
Industry group          ps     total\a          ps  of total
--------------  ----------  ----------  ----------  --------
Agriculture,       124,164           8         431        12
 forestry, and
 fishing
Mining              40,594           3          81         2
Construction        55,806           4         197         6
Manufacturing       28,972           2         184         5
Transportation      24,107           2         172         5
 ,
 communication
 , etc.
Wholesale           16,725           1         101         3
 trade
Retail trade       155,109          10         355        10
Finance,           115,015           7         159         4
 insurance
Real estate        682,776          44         903        26
Services           279,831          18         794        22
Unknown             30,431           2         170         5
============================================================
Total            1,553,530         100       3,547       100
------------------------------------------------------------
\a Total may not add due to rounding. 

Sources:  The estimated universe of partnerships' data was derived
from IRS' SOI data, and the estimates were associated with sampling
errors that were ignored for this comparison.  The audited
partnerships' data were derived from IRS' RTF and AIMS data, and only
3,547 of 5,565 audited partnerships are included in these data. 

Our analysis by predominant source of revenue showed more
differences.  As shown in table I.3, we found that the most
productive audits were of nontax shelter partnerships, subject to the
TEFRA audit rules, which received 90 percent or more of their total
revenue from one source.  About 18 percent of the partnerships for
which we had data met these criteria.  These cases accounted for
about 42 percent of the total dollars adjusted and 23 percent of the
partnership examination hours. 



                          Table I.3
           
           Partnerships With 90 Percent or More of
            Revenue From a Single Revenue Category

                       Percent
                            of
                      estimate
                             d   Percent   Percent   Percent
                      universe        of  of total  of total
                            of   audited  examinat     audit
                      partners  partners       ion  adjustme
Type of partnership       hips      hips     hours        nt
--------------------  --------  --------  --------  --------
TEFRA nonshelter          17.3      18.0      22.9      41.8
Non-TEFRA nonshelter      64.7      50.7      42.0      19.4
TEFRA tax shelter          1.4       6.9       8.7      13.3
Non-TEFRA tax              0.3       0.7       0.4        \a
 shelter
------------------------------------------------------------
\a Less than 0.1 percent. 

Sources:The estimated universe of partnerships' data was derived from
IRS' SOI data, and estimates were associated with sampling errors
that were ignored for this comparison.  All other data were derived
from IRS' RTF and AIMS data, and percents were based on 3,546
partnerships for which we received information.  One partnership with
a total adjustment of over $800 million was excluded because it
skewed the analysis. 

Another observation we made was that the average audit adjustment per
examination hour varies depending on what percent of a partnership's
total revenue comes from a single revenue category.  Tables I.4
through I.7 show the results of our analysis by percent of total
revenue from a single category, number of partnerships, total dollar
adjustment, total examination hours, and average audit adjustment by
examination hour.  These analyses are based on information we
obtained for 3,546 of the 5,565 partnerships with audits closed in
fiscal year 1992. 



                          Table I.4
           
            Analysis of Audited TEFRA Tax Shelter
           Cases by Percent of Total Revenue From a
                   Single Revenue Category

                         Number
                             of       Total  Total   Average
                       partners      dollar   exam   change/
Percent of revenue         hips  adjustment  hours      hour
---------------------  --------  ----------  -----  --------
No positive revenue          46  $22,624,62  3,671    $6,163
                                          7
Less than 70 percent         10   5,777,236    701     8,241
70 to 79.9 percent            4    -134,201    902      -149
80 to 89.9 percent           12   4,486,390    894     5,018
90 percent or more          243  100,797,81  22,48     4,483
                                          3      5
============================================================
Total                       315  $133,551,8  28,65    $4,661
                                         65      3
------------------------------------------------------------
Source:  GAO analysis of IRS' RTF and AIMS data. 



                          Table I.5
           
              Analysis of Audited Non-TEFRA Tax
              Shelter Cases by Percent of Total
            Revenue From a Single Revenue Category

                           Number     Total
                               of    dollar  Total   Average
                         partners  adjustme   exam   change/
Percent of revenue           hips        nt  hours      hour
-----------------------  --------  --------  -----  --------
No positive revenue            22  $1,083,4  1,349      $803
                                         07
Less than 70 percent            4     2,297    152        15
70 to 89.9 percent              2     4,929    295       197
90 percent or more             25    78,505  1,143        69
============================================================
Total                          53  $1,169,1  2,939      $398
                                         38
------------------------------------------------------------
Source:  GAO analysis of IRS' RTF and AIMS data. 



                          Table I.6
           
             Analysis of Audited TEFRA Nonshelter
           Cases by Percent of Total Revenue From a
                   Single Revenue Category

                                     Total
                      Number of     dollar   Total   Average
                     partnershi  adjustmen    exam   change/
Percent of revenue           ps          t   hours      hour
-------------------  ----------  ---------  ------  --------
No positive revenue          72  $18,461,0   7,687    $2,402
                                        80
Less than 70                 48  11,647,42   5,766     2,020
 percent                                 1
70 to 79.9 percent           30  3,738,723   3,350     1,116
80 to 89.9 percent           43  1,906,176   5,756       331
90 percent or more          639  317,881,4  58,993     5,388
                                        35
============================================================
Total                       832  $353,634,  81,552    $4,336
                                       835
------------------------------------------------------------
Source:  GAO analysis of IRS' RTF and AIMS data. 



                          Table I.7
           
                Analysis of Audited Non-TEFRA,
             Nonshelter Cases by Percent of Total
            Revenue From a Single Revenue Category

                     Number of       Total   Total   Average
                    partnershi      dollar    exam   change/
Percent of revenue          ps  adjustment   hours      hour
------------------  ----------  ----------  ------  --------
No positive                232  $24,839,22  12,861    $1,931
 revenue                                 7
Less than 70               134  38,229,336  10,461     3,654
 percent
70 to 79.9 percent          89  54,270,367   7,126     7,616
80 to 89.9 percent          91   7,389,686   5,117     1,444
90 percent or more       1,800  147,491,24  108,32     1,362
                                         3       9
============================================================
Total                    2,346  $272,219,8  143,89    $1,892
                                        59       4
------------------------------------------------------------
Source:  GAO analysis of IRS' RTF and AIMS data. 

As can be seen from table I.4, the greatest dollar adjustment per
examination hour ($8,241) occurred with tax shelter partnerships,
subject to TEFRA, where the revenue sources were diversified.  That
is, the largest single source of revenue was less than 70 percent of
the partnerships' total revenues.  Table I.7 shows that the next
highest dollar return per hour of examination ($7,616) occurred with
nonshelter partnerships, not subject to TEFRA, where the revenue
sources diversified.  For these partnerships, the largest single
revenue source was between 70 percent and 79.9 percent of the total
revenues.  The third largest dollar return per hour examination
($6,163), see table I.4, was for tax shelter partnerships, subject to
TEFRA, which had no positive revenue. 


COST ESTIMATED FOR PARTNERSHIP
DOCUMENT MATCHING PROGRAM
========================================================== Appendix II

We estimated that if IRS matched 100 percent of the Schedules K-1 to
individual partner tax returns, it may have been able to assess an
additional $219.5 million in tax revenue at an additional cost of
$18.6 million.  IRS matches a limited number of Schedules K-1 that it
receives to tax returns filed by individual partners.  To arrive at
our estimates, we extrapolated IRS' estimated costs and tax revenues
for the Schedules K-1 used in its tax year 1991 individual document
matching program to all of the schedules that it received that year. 

For tax year 1991, 20,504,693 partnership Schedules K-1 were received
by IRS and 2,525,825 were processed into computer format, while the
remaining 17,978,868 were not processed.\1 Of the 2,525,825 schedules
processed, IRS used 2,382,190 (94 percent) in its matching program. 
To determine which Schedules K-1 to match, IRS randomly sampled
taxpayers' social security numbers ending in certain digits.  IRS
estimated that it costs $0.75 to transcribe a Schedule K-1 and $17.61
to close an individual underreporter case.  Finally, IRS estimated
that for each $1.00 spent to match Schedules K-1, it assessed a net
$11.80 in additional tax revenue.  According to IRS officials, it is
possible that if all Schedules K-1 were matched, the net additional
tax revenue could decrease.  The officials said that as more cases
are worked the marginal yield becomes lower for all cases. 

Table II.1 shows the basis of our estimates of costs and benefits of
matching Schedules K-1 to individual partner tax returns. 



                          Table II.1
           
              Estimates of Costs and Benefits of
             Matching Schedules K-1 to Individual
                     Partner Tax Returns

                                                Estimates of
Data used to develop estimates of costs and        costs and
benefits                                            benefits
--------------------------------------------  --------------
Number of partnership Schedules K-1 used in        2,382,190
 tax year
 1991 matching program
Number of potential underreporter cases               39,948
 identified for tax
 year 1991
Ratio of Schedules K-1 to potential                     59.6
 underreporter cases (2,382,190 /39,945)
Number of partnership Schedules K-1 not used      18,122,503
 in tax
 year 1991 matching program
Number of partnership Schedules K-1               17,035,152
 transcribed that had
 valid social security numbers and could be
 matched (18,122,503 x .94)
Number of potential underreporter cases              285,825
 developed
 (17,035,152 /59.6)
Cost to transcribe 18,122,503 Schedules K-1    $13.6 million
 (18,122,503 x $0.75)
Cost to close underreporter cases ($17.61 x     $5.0 million
 285,825)
Total additional cost for matching program     $18.6 million
Potential additional tax revenues ($18.6      $219.5 million
 million x $11.80)
------------------------------------------------------------
Source:  IRS underreporter data. 


--------------------
\1 In tax year 1991, of the 2,525,825 processed partnership Schedules
K-1, 2,382,190, 94 percent, were posted to the Information Returns
Master File.  The remaining 143,635 Schedules K-1 had invalid or no
social security numbers and could not be used in a matching program. 


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

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C. 

Ralph Block, Assistant Director, Tax Policy and Administration Issues
Rodney Hobbs, Assignment Manager

SAN FRANCISCO FIELD OFFICE

Suzy Foster, Evaluator-in-Charge
Eduardo Luna, Evaluator
Samuel Scrutchins, Evaluator

