Tax Administration: Alternative Strategies to Obtain Compliance Data
(Letter Report, 04/26/96, GAO/GGD-96-89).

GAO assessed the potential effects on the Internal Revenue Service's
(IRS) compliance programs of postponing the 1994 Taxpayer Compliance
Measurement Program (TCMP) survey and identified some potential short-
and long-term TCMP alternatives.

GAO found that: (1) IRS postponed the 1994 TCMP because of criticisms
and budget constraints; (2) IRS does not know how it will obtain the
taxpayer compliance data its needs; (3) the loss of 1994 TCMP data could
increase compliant taxpayers' burden over the long term because audits
may become less targeted; (4) to mitigate the data losses over the short
term, IRS could employ a number of alternatives, including doing a
smaller survey; (5) any alternative should reduce sample size to lessen
taxpayer burden and administrative costs, maintain IRS ability to update
the discriminant function scoring system, and maximize the use of
already completed work; (6) a limited survey would reduce the quantity
and quality of the data collected, but still provide national compliance
data; (7) IRS must determine how it will measure compliance over the
long term, since its workload and future revenues depend on taxpayers'
voluntary compliance; (8) long-term alternatives include conducting
small multiyear TCMP audits, using data from operational audits to
assess compliance changes, and conducting periodic national mini-TCMP
audits; (9) IRS must decide on a compliance information-gathering
alternative in the near term, since any alternative will take several
years to develop and implement; and (10) the alternatives will likely
not gather data as comprehensive as the originally planned TCMP data.

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

 REPORTNUM:  GGD-96-89
     TITLE:  Tax Administration: Alternative Strategies to Obtain 
             Compliance Data
      DATE:  04/26/96
   SUBJECT:  Tax administration systems
             Voluntary compliance
             Tax nonpayment
             Administrative costs
             Statistical data
             Data collection operations
             Surveys
             Taxpayers
             Tax return audits
IDENTIFIER:  IRS Taxpayer Compliance Measurement Program
             IRS Market Segment Specialization Program
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


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


Report to the Commissioner, Internal Revenue Service

April 1996

TAX ADMINISTRATION - ALTERNATIVE
STRATEGIES TO OBTAIN COMPLIANCE
DATA

GAO/GGD-96-89

Compliance Data

(268705)


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

  CV - coefficient of variation
  DORA - District Office of Research and Analysis
  DIF - Discriminant Function
  IRS - Internal Revenue Service
  NORA - National Office of Research and Analysis
  SOI - Statistics of Income
  TCMP - Taxpayer Compliance Measurement Program

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


B-266028

April 26, 1996

The Honorable Margaret Milner Richardson
Commissioner of Internal Revenue

Dear Ms.  Richardson: 

On October 23, 1995, the Internal Revenue Service (IRS) decided to
postpone the 1994 Taxpayer Compliance Measurement Program (TCMP)
indefinitely because of budget concerns.  Also, there was
considerable pressure from Congress, taxpayer groups, paid preparers,
and others to cancel TCMP because of its cost and burden on
taxpayers.  In light of its postponement, we initiated this
assignment to determine the potential effects on IRS' compliance
programs of postponing the 1994 TCMP and to identify some potential
short- and long-term alternatives to the planned TCMP for collecting
this data. 


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

For over 30 years, TCMP has been IRS' primary program for gathering
comprehensive and reliable taxpayer compliance data.  It has been
IRS' only program for making statistically reliable estimates of
compliance nationwide.  It has also been used to identify areas where
tax law needs to be changed to improve voluntary compliance and to
estimate the tax gap and its components.\1 TCMP data are also used
outside IRS, including by Congress to make revenue estimates for new
legislation and by the Department of Commerce's Bureau of Economic
Analysis to adjust national income accounts such as the gross
domestic product. 

The 1994 TCMP survey, which was to consist of over 150,000 income tax
returns, was to be the most comprehensive TCMP effort ever
undertaken.  By auditing the tax returns of individuals (Form 1040),
small corporations with $10 million or less in assets (Form 1120),
Partnerships (Form 1065), and S corporations\2 (Form 1120S), IRS
planned to obtain comprehensive compliance data.  Most sample results
were to be sufficiently precise to be reliable at the national level
as well as at smaller geographic areas across the country.  The 1994
TCMP was designed to fulfill the information needs for several
compliance areas expected to be important to IRS' functions over the
next decade.  The more important uses were to include development of
audit selection formulas, validation of IRS' revised approach to
categorizing returns for audit, and development of new approaches to
researching compliance across specific geographic areas.  Each of
these uses is discussed in more detail below. 

Since 1969, IRS has used TCMP data to update its Discriminant
Function (DIF) formulas, which are mathematical formulas used to
select tax returns with the greatest probability of change for audit. 
The current formulas for individuals are based on 1988 tax returns,
IRS' most recent individual TCMP audits.  Formulas for small
corporations are based on returns that were processed in 1987.  IRS
does not use DIF scores for partnerships and S corporations because
of the age of the underlying TCMP audits. 

TCMP data were also to be used to test new compliance strategies. 
IRS planned to change the way it categorized returns for audit by
adopting the market segment approach.  Market segments represent
groups of taxpayers with similar characteristics, such as those in
manufacturing.  IRS assumes that because these taxpayers have similar
external characteristics, their tax compliance behavior will exhibit
similar attributes. 

Finally, the 1994 TCMP was to provide compliance research data.  IRS
recently reorganized its compliance research function, establishing a
National Office of Research and Analysis (NORA) and 31 District
Office of Research and Analysis (DORA) sites.  The 1994 TCMP was to
be large enough to provide reliable compliance data for field and
National Research Offices.  IRS' researchers planned to use TCMP data
to identify national and geographically specific areas of
noncompliance and, by focusing on key compliance issues, develop
programs to improve voluntary compliance.  It is through these
research efforts that IRS planned to improve overall voluntary
compliance. 

Noncompliance represents a major source of lost revenue for the
nation.  IRS' most recent tax-gap estimates indicate that over $127
billion was lost to noncompliance in 1992.  In an attempt to reduce
this lost revenue, IRS established an objective of collecting at
least 90 percent of the taxes owed through voluntary compliance and
enforcement measures by the year 2001.  However, this overall
compliance rate has remained at about 87 percent since 1973.  The
1994 TCMP was intended to provide data from which other programs
could be developed to improve this rate and increase revenue. 

This nation's tax system is based on individuals and businesses
voluntarily paying the taxes they owe.  To the extent that this
system works, it improves the efficiency of tax collection. 
Measuring the extent to which the tax system works and identifying
areas in which it does not is the job of compliance measurement. 
TCMP has been IRS' only tool for measuring voluntary compliance and
determining compliance issues.  The postponed TCMP for 1994 tax
returns was to establish the voluntary compliance benchmark to carry
IRS into the next century. 


--------------------
\1 The tax gap is the difference between the amount of taxes owed and
the amount voluntarily paid in a year. 

\2 An S corporation is taxed similar to a partnership.  A qualifying
corporation may choose to be generally exempt from federal income
tax.  Its shareholders will then include in their income their share
of the corporation's separately stated items of income, deductions,
losses, and credits. 


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

According to IRS officials, criticisms of TCMP and budget
difficulties make it unlikely that the 1994 TCMP, as planned, will be
conducted.  Although IRS officials told us that data on taxpayer
compliance with tax laws are still needed, and they plan to obtain
them; they currently do not know how they will get them.  They said
the loss of 1994 TCMP compliance data could lead to an increased
burden on compliant taxpayers over the long term, if return selection
formulas become less efficient and increasing numbers of compliant
taxpayers are selected for IRS' regular audit program. 

In addition to measuring voluntary compliance, the 1994 TCMP was to
provide compliance data for several IRS initiatives, including
updating return selection formulas, validating the market segment
approach, and providing data for compliance research.  The absence of
this type of compliance data will make it more difficult to
accomplish the objectives of these initiatives. 

To mitigate the difficulties this absence of updated compliance data
may cause in the short term, IRS would have to identify and implement
an alternative method of obtaining the data.  To build on the
significant resource investment already made and to conserve time,
this effort would benefit from using data already collected for the
1994 TCMP. 

Currently, while IRS plans to mitigate the data losses resulting from
the postponement of the 1994 TCMP, it has no specific proposal on how
to accomplish this.  However, on the basis of our discussions with
IRS staff and our experiences with past TCMPs, we identified some
possible short-term sampling alternatives for IRS' consideration.  It
appears that IRS could reduce the burden and costs of TCMP by
changing the design parameters to accept a smaller sample of tax
returns to audit, thus achieving at least some of the benefits that
the postponed TCMP study was designed to achieve.  Although there are
numerous sampling strategies that could be used, some alternative
design parameters could include reducing the types of tax returns
covered, changing the precision of the sample results by reducing the
number of returns audited, and excluding the returns of taxpayers who
appear to IRS auditors to be compliant.  We estimate that these
changes could reduce the sample size by up to 119,000 of the 153,000
returns.  None of these sampling strategies would provide the
quantity or quality of data that were to be provided by the planned
TCMP, such as the ability to provide estimates of noncompliance at
the field office level.  However, they could, at a minimum, provide
national compliance information and, potentially, data to update at
least some of IRS' audit selection formulas. 

Because a significant portion of IRS' workload and future revenue
collections depends on compliance programs, IRS will need to
determine how it will measure compliance over the longer term.  Such
measurements are an ongoing need for any tax system that depends on
voluntary compliance.  The compliance measurement program's
sustainability, in view of budget limitations and taxpayer burden
concerns, is a key planning consideration for any long-term program. 
IRS officials said that potentially long-term alternatives may
include, (1) conducting multiyear TCMP type audits from smaller
samples of tax returns and combining the data from several years to
ensure the necessary precision and coverage; (2) using data from
operational audits to assess changes in compliance; and (3)
conducting a national "mini" TCMP type audit periodically to identify
emerging issues, followed by smaller audit efforts on the issues
identified in this national effort.  Each of these alternatives would
be cheaper and less burdensome to IRS and taxpayers than the proposed
TCMP sample but would also provide less comprehensive compliance
data. 

Regardless of how IRS decides to replace the information that would
have been provided by TCMP, it is important to begin soon because any
alternative is likely to require several years to put into place, and
the data will be needed to update information on IRS' compliance
programs.  For example, if data from a short-term alternative were
available by 1998, the year IRS originally estimated the completed
data would be needed, it would cause the least disruption to IRS'
compliance program. 


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

The objectives of this assignment were to (1) determine the possible
effects on IRS' compliance programs of postponing the 1994 TCMP and
(2) identify some potential short- and long-term alternatives to the
planned TCMP for collecting this data. 

To determine the possible effects of postponing the 1994 TCMP, we
talked to responsible officials in IRS' Research Division and the
Examination Division.  We obtained information on how these officials
planned to use TCMP data and what will likely be affected now that
TCMP has been postponed. 

To identify alternatives to the planned TCMP, we talked to IRS
officials responsible for planning TCMP.  We discussed alternative
sampling methodologies with officials from IRS' Statistics of Income
(SOI) Branch who were responsible for preparing the original TCMP
sample and asked them to determine sample sizes on the basis of
revised requirements.  We developed the revised requirements on the
basis of our discussions with IRS' Research Division staff as well as
officials outside IRS, including congressional staff. 

Some of the observations in this report are based on the work we have
done over the years on IRS' compliance programs as well as our
specific work on TCMP in recent years. 

We requested comments from you on a draft of this report.  On
February 23, 1996, we obtained oral comments from IRS' Director of
Research and the National Director of Compliance Specialization.  We
also obtained commented from you in a March 18, 1996, letter.  These
comments are discussed on page 13 of this report. 

We did our work in San Francisco, Dallas, and Washington, D.C.,
between August and December 1995 in accordance with generally
accepted government auditing standards. 


   LOSS OF TCMP DATA COULD DISRUPT
   IRS' COMPLIANCE STRATEGIES
------------------------------------------------------------ Letter :4

The planned TCMP for 1994 tax returns was to establish the voluntary
compliance benchmark to carry IRS into the next century.  While
agency officials said that postponing TCMP will help resolve budget
problems, our work suggests that the loss of these or comparable data
is also likely to disrupt IRS' efforts to increase the total
collection percentage to 90 percent by 2001.  For example, without
these data, IRS will have difficulty updating the formulas it uses to
select returns for audit and, thus, it would be more likely that a
higher percentage of the returns IRS selects for audit would not
result in changes to the amount of tax owed by the taxpayer. 
Additionally, without such data IRS will be unlikely to have
sufficient data to validate its market segment approach to audits or
to be used by the DORA research functions to identify programs to
improve voluntary compliance.  It is not clear whether IRS will
replace the data it had planned to obtain from TCMP.  However,
updated compliance data will be needed in the short term if IRS still
plans to update the audit selection formulas and in the long term to
validate and improve IRS' compliance efforts. 


      UPDATED COMPLIANCE DATA WERE
      NEEDED TO UPDATE RETURN
      SELECTION FORMULAS
---------------------------------------------------------- Letter :4.1

The primary system that will be disrupted by postponing TCMP is the
one used by IRS to select returns for audit.  Since 1969, IRS has
used DIF formulas to select returns for audit.  New DIF formulas are
developed periodically from TCMP data and applied to all individual
and small corporation income tax returns.  IRS then selects returns
for audit with the highest DIF scores.  In 1992, over 55 percent of
the audited returns of individuals were selected using the DIF
score.\3 The DIF selection system replaced programs that were largely
dependent on auditor's judgment.  The DIF system has not only
improved the efficiency of IRS' audit efforts but also the
consistency and objectivity of the selection process.  The use of the
DIF selection process has also resulted in fewer "no-change"
audits,\4 which not only waste IRS' resources but unnecessarily
burden compliant taxpayers.  According to IRS, use of the DIF scoring
system reduced the no-change rate from over 46 percent in 1969 to
about 15 percent in 1992. 

IRS officials believe the DIF process is dependent on periodically
updating the formulas used to score returns.  Formulas are updated so
that they will more accurately identify the returns with the greatest
probability for change.  Until 1988, data from TCMP had been used to
update formulas for individual returns every 3 years.  However, the
most recent TCMP was conducted on 1988 individual returns.  For small
corporations, partnerships, and S corporations, IRS has updated
formulas much less frequently.  TCMPs were conducted on corporate
returns filed in 1987, and partnership and S corporation returns
filed in 1982 and 1985, respectively. 

IRS is not certain how well the DIF scores will continue to perform
if not updated.  IRS officials believe that by 1998, the year IRS
planned to have TCMP data available, the DIF scores may become less
effective at identifying returns with the greatest potential for
change.  They said this decrease in effectiveness may occur because
of changes in tax laws and taxpayer behavior--resulting in an
increased no-change rate for DIF selected returns and potentially
lower revenue yields.  This would mean greater burden on compliant
taxpayers if more of them are selected for audit.  IRS officials
indicated that they plan to monitor the performance of DIF over time. 


--------------------
\3 In 1993 and 1994, a smaller proportion of the examined returns
were selected using DIF because of a 2-year special nonfiler
compliance initiative conducted by IRS' Examination Division.  IRS
officials expect the proportion of returns selected using DIF to
again increase as this nonfiler initiative is completed. 

\4 "No-change" audits are those that did not change the reported tax
liability. 


      COMPLIANCE DATA NEEDED TO
      VALIDATE MARKET SEGMENT
      APPROACH
---------------------------------------------------------- Letter :4.2

The 1994 TCMP was also intended to provide information on IRS' new
market segment approach for grouping tax returns.  IRS initiated the
market segment approach on the basis of work done in its Western
Region, which indicated that compliance rates and audit issues were
likely to be similar for taxpayers with similar characteristics, such
as businesses in the same industry (e.g., manufacturing or retail
sales).  Accordingly, IRS concluded that grouping taxpayers by market
segments might result in selecting returns for audit that have a
higher potential for change and might allow auditors to specialize in
market segments. 

The 1994 TCMP was designed to provide data to test this hypothesis as
well as to develop DIF scores by market segment rather than by audit
class, as had been done in the past.\5

Without TCMP or some alternative to provide similar information, IRS
will not have data to show whether market segments are better for
return selection purposes than traditional audit classes or be able
to determine the compliance rate or compliance issues of the market
segments.  Because of these concerns, IRS no longer plans to test a
selection of returns for audit by using the market segment approach. 
Instead, IRS plans to continue selecting returns for audit using the
DIF score within audit classes. 


--------------------
\5 IRS developed DIF formulas for each of 10 audit classes.  These
audit classes separate taxpayers on the basis of the amount of income
and type of return filed. 


      TCMP WAS DESIGNED TO SUPPORT
      IRS' RESEARCH FUNCTION
---------------------------------------------------------- Letter :4.3

Finally, the 1994 TCMP was designed to provide compliance data for
IRS' National and District Research Offices.  IRS established these
offices to research taxpayer compliance at the national and local
levels.  These researchers were to identify programs to improve
compliance not only through audits but also through larger scale
nonaudit programs, such as improved guidance and assistance to
taxpayers and tax-law changes.  TCMP also was to be used to develop
benchmark compliance data for measuring future progress and determine
how effectively managers were meeting their objectives of improving
compliance. 

Without TCMP or an alternative data source, IRS' new research
function would still be able to analyze noncompliance in filing
returns and paying taxes.  However, research on reporting compliance,
the area where most of IRS' compliance dollars are spent, would be
very limited.  Thus, researchers would have inadequate data to
identify emerging trends in reporting compliance, to develop
solutions, and to test the effectiveness of these solutions.  As a
result, IRS would likely continue its reliance on enforcement to
improve compliance.  However, enforcement has proven to be a costly
and ineffective way to increase overall voluntary compliance. 


   POSSIBLE SHORT-TERM APPROACHES
   TO COLLECT COMPLIANCE DATA
------------------------------------------------------------ Letter :5

According to IRS officials, because of criticisms of TCMP and budget
concerns, the 1994 TCMP is unlikely to be conducted.  Although IRS
officials told us they planned to use an alternative method to obtain
TCMP data, they currently have no short-term proposal on how to
obtain these data.  Regardless of how IRS plans to mitigate the loss
of 1994 TCMP data, it would have to start soon in order to minimize
the adverse effects of not updating its compliance programs. 

According to IRS officials, a number of alternative sampling
strategies could fill the short-term data gap created by postponing
TCMP indefinitely.  From these strategies, we identified several
alternative samples that met three basic objectives we considered
important:  (1) reducing the sample size to make data collection less
costly for IRS and less burdensome to taxpayers, (2) maintaining IRS'
ability to update the DIF scoring system, and (3) maximizing use of
the work already completed to identify returns and collect data for
the 1994 TCMP sample. 

One alternative sample would be for IRS to reduce the planned TCMP
sample size and still provide some of the same data, although with
less precision.  This smaller sample could also be used to update the
DIF score with little loss in accuracy.\6 On the basis of our
discussions with SOI officials, it appears IRS could reduce the
sample size in any one of several ways, including

  -- decreasing the level of acceptable statistical precision for
     individual and corporate returns;\7

  -- selecting a sample with only businesses (sole proprietors,
     corporations, partnerships, and S corporations), with reduced
     precision;

  -- classifying TCMP sample returns and eliminating returns that
     past audit experience indicates are not likely to result in an
     audit adjustment;\8 and

  -- selecting a sample that includes only sole proprietor and
     corporation returns. 

Numerous other alternatives to the sampling methodology and
characteristics may give slightly different sample sizes.  For
example, by eliminating the requirement for updating the DIF formula,
the sample size for the corporation and individual business option is
reduced by about 12 percent, to 28,275.  However, such an approach
would lessen the value of TCMP because it would limit IRS' ability to
update the DIF score, a primary purpose of TCMP audits. 

Reducing the sample size would reduce the cost of TCMP audits.  IRS'
cost estimates for the 1994 TCMP were divided into two types, (1)
staffing costs and (2) opportunity costs.  Staffing costs reflect
IRS' cost estimates for auditors to conduct the TCMP audits. 
Opportunity costs reflect IRS' estimates of the difference between
revenue generated through the regular audit program and revenue
generated by TCMP audits.  According to IRS officials, TCMP audits
generate less revenue because the returns are randomly selected
rather than identified by using the DIF score or as part of a special
project and because the returns take longer to audit. 

Table 1 shows how the variations in sampling methodology and
characteristics change the sample size and cost estimates. 



                                     Table 1
                     
                         Changes to Sample Size and Cost
                      Estimates Based on Changes to Sampling
                                 Characteristics

                              (Dollars in millions)


Sample                                                                Opportunit
characteristics         1040    1120   1120S    1065   Total  Direct           y
--------------------  ------  ------  ------  ------  ======  ------  ----------
Original sample       92,185  35,360  12,501  12,488  152,53    $558      $1,452
                                                           4
Decrease the          52,313  15,121  12,501  12,488  92,423     309         805
 precision of the                         \b      \b
 sample
Sample businesses     18,267  15,121  12,501  12,488  58,377     280         727
 only                                     \b      \b
Classify returns and  36,909   8,063   6,666   6,659  58,297     184         479
 delete no-change
 audits
Sample corporations   18,267  15,121       0       0  33,388     153         397
 and sole
 proprietors only
--------------------------------------------------------------------------------
\a The following return types were to be included in TCMP:  Form 1040
Individual Income Tax Return, Form 1120 Corporation Income Tax
Return, Form 1120S S Corporation Return, and Form 1065 Partnership
Income Tax Return. 

\b Sample size for partnership and S corporation returns remains the
same as in the original 1994 TCMP sample for all options for which
these types of returns are included, because IRS has no data on which
to determine the number of returns that would be profitable to
examine. 

Source:  Sample size data provided by IRS' SOI staff and cost
estimates calculated by GAO based on IRS' cost information for the
planned 1994 TCMP. 

Changing the sample characteristics not only reduces the size but
affects the usefulness of data from the sample.  Each of the changes
shown in table 1 has its own set of strengths and weaknesses that
relate primarily to reliability and coverage.  For example, reducing
the sample to businesses only and reducing the precision would
provide no information on nonbusiness individuals.  Also, this sample
would be of little use at the DORA level because it would not provide
statistically reliable estimates of compliance below the national
level.  This sample could, however, provide some information on
market segment compliance and be used to update the DIF formula for
businesses and the return types where voluntary compliance is the
lowest.  Also, a business-only approach could be combined with a
multiyear sample where the compliance of nonbusiness individual
returns is evaluated in a future year.  Although we did not fully
evaluate the alternatives, the table in appendix I summarizes some of
the more obvious trade-offs inherent in the alternatives discussed
above. 

Deciding how to change the sampling strategy to reduce the sample
size would require careful evaluation of the tradeoffs.  It seems
reasonable, however, to consider that any new sample should, at a
minimum, allow some updating of the DIF formulas, since this was to
be the primary purpose of the original TCMP.  To the extent that
other purposes can also be met through one of these alternative
sampling strategies, the sample would be more valuable. 


--------------------
\6 The primary criteria for updating the DIF score is the
profitable-to-audit requirement.  Each individual and small
corporation strata must have at least 500 such returns.  The
profitable-to-examine level (i.e., amount of adjustments) ranges from
$500 for certain nonbusiness individuals to $10,000 for some sole
proprietors and corporations.  The most recent TCMP is used to
estimate the number of profitable-to- audit returns within a strata. 

\7 IRS uses the coefficient of variation (CV) ratio as a measure of
the sample's precision.  The CV is the ratio between the standard
error of the mean divided by the mean.  The CV ratio measures the
precision of sample point estimates and is calculated using data from
prior TCMPs.  A sample with a CV ratio of 1 is generally twice as
reliable as a sample with a CV ratio of 2.  For our alternative
samples, we asked IRS to increase the CV ratios as follows: 
Individual nonbusiness returns to 5 percent, individual business
returns to 2 percent, corporation returns to 5 percent.  The CV
ratios were changed from 3.5 percent, 1 percent, and 2 percent,
respectively. 

\8 Included in this sample is a 2 percent subsample of the no-change
returns to validate the classification. 


   LONG-TERM COMPLIANCE
   MEASUREMENT CONSIDERATIONS
------------------------------------------------------------ Letter :6

Because a significant portion of IRS' workload and future revenue
depends on compliance programs, it is important that IRS determine
how to measure compliance.  Such measurements are an on-going need
for any tax system that depends on voluntary compliance.  It is also
important that any long-term solution to obtaining compliance
measurement information address the issue of sustainability so that
long-term consistent measurement data are available.  Sustainability
means that the program's costs, in terms of IRS' budget and perceived
burden on the taxpayer, must be clearly defensible.  Additionally, to
be efficient and effective, it would be necessary to design a program
that provides timely data and clearly identifies the objectives and
uses of these compliance data. 

We identified several alternatives to the traditional TCMP that would
meet some of the data needs that were lost when TCMP was postponed,
including (1) conducting multiyear TCMP audits on smaller samples and
combining the results; (2) using operational audit data; and (3)
conducting a mini TCMP to identify compliance issues, with a more
focused TCMP audit on the identified issues.  We discuss these three
options below. 


      USING MULTIYEAR TCMP AUDITS
---------------------------------------------------------- Letter :6.1

The multiyear TCMP alternative envisions annual TCMP-type audits on a
smaller sample of tax returns which, over the course of several
years, could be combined to obtain the required statistical
precision.  For example, IRS could disaggregate an entity type, such
as individual taxpayers, into separate market segments or audit
classes and conduct the audits of each segment on a 3-year cycle. 
Table 2 below shows an example of how such a program might operate. 



                                Table 2
                
                   Example of a Multiyear TCMP Effort

                                                                Number
                                                                    of
Ye                                                              return
ar  Type of return audited                                           s
--  ----------------------------------------------------------  ------
1   Individual, nonbusiness, total positive income less than    16,363
     $50,000
2   Individual, nonbusiness, total positive income $50,000 or   14,508
     more
3   Individual, business or farm                                18,267
======================================================================
To                                                              49,138
 t
 a
 l
----------------------------------------------------------------------
Source:  Estimates made from data provided by IRS for our alternative
sampling strategies. 

One benefit of such an approach to IRS would be that after the
initial 3-year period, new and current data would become available
for one of the segments every year, making it easier to fine-tune the
compliance system.  Such an approach, however, would require
considerable effort from IRS' statisticians to ensure that the sample
design was statistically sound.  Also, it would require a long-term
commitment from IRS managers to ensure that returns were audited
regularly. 


      USING OPERATIONAL AUDIT DATA
---------------------------------------------------------- Letter :6.2

A second option is to use data from operational audits already being
done.  Using data from operational audits would provide a large
amount of compliance data.  This option is also probably the most
sustainable of the three we discuss because it would be less
burdensome on compliant taxpayers and have no marginal staffing and
opportunity costs.  However, there are weaknesses.  IRS currently has
no system to track operational audit issues.  While such a system is
currently being developed, it is not yet operational and testing is
not planned to begin until later in 1996.  According to IRS
officials, this database is to identify audit issues as well as
provide codes to identify the causes of noncompliance.  Also, IRS
officials believe that using a database of operational audit results
could not be used for updating the DIF formulas, determining ways to
improve voluntary compliance, or systematically identifying emerging
audit issues because the audited returns would not be randomly
selected. 


      USING A MINI TCMP
---------------------------------------------------------- Letter :6.3

A third option is to periodically conduct a very small TCMP that
covers all taxpayers and follow up with mini TCMP audits on specific
issues identified as concerns.  Using this approach, IRS may be able
to reduce the sample size and focus the majority of the audits on
less compliant taxpayers, thus reducing cost and taxpayer burden. 
This approach may also provide IRS with insight into the areas of
greatest noncompliance because efforts would be more focused.  IRS
officials said that this approach, however, would probably not
provide sufficient data to update the DIF formulas and may be of
little use at DORA sites because too few randomly selected returns
would likely be examined. 


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

A significant proportion of IRS' present and future compliance
programs have been predicated on the information obtained from TCMP. 
Benchmarking current compliance, validating the market segment
approach, updating return selection formulas, researching
noncompliance issues and developing programs to address them, and
estimating the tax gap all depend on TCMP information.  Without
updated compliance data, increasing voluntary compliance, as
envisioned by IRS, is less likely to occur. 

IRS has options to replace at least some of the data that would have
been available from the 1994 TCMP audits.  Auditing a smaller sample
size by eliminating some return types and accepting a decrease in
precision, is a factor in such options.  While each of these
alternatives has limitations, they would meet some of the data needs
that were lost when TCMP was postponed.  It is important for IRS to
make a decision soon on how to replace TCMP data because it will take
some time to implement a replacement, and IRS projects that the
currently available 1988 data will be less effective by 1998. 

If IRS does not develop a sustainable compliance measurement program,
IRS' compliance programs may be disrupted as the proportion of audits
that result in no-changes increases and IRS' access to information on
emerging compliance issues decreases.  In the long term, such
disruptions are likely to result in increased burdens on compliant
taxpayers as more of them are selected for audit. 


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

To provide the data necessary to help meet the objectives of IRS'
compliance strategies, we recommend that you

  -- identify a short-term alternative strategy to minimize the
     negative effects of the compliance information that is likely to
     be lost because TCMP was postponed, and

  -- develop a cost-effective, long-term strategy to ensure the
     continued availability of reliable compliance data. 


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

We requested comments from you on a draft of this report. 
Responsible IRS officials, including the National Director,
Compliance, Research and National Director, Compliance
Specialization, provided comments in a February 23, 1996, meeting. 
These officials agreed with our recommendations and provided some
technical comments, which we have incorporated where appropriate.  In
a March 18, 1996, letter, you restated those agreement and indicated
that over the next several months IRS would devote substantial effort
to investigating all potential options for capturing reliable
compliance information as an alternative to TCMP. 

We believe the actions that IRS proposes, if properly implemented,
will be responsive to our recommendations. 


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

This report contains recommendations to you.  The head of a federal
agency is required by 31 U.S.C.  720 to submit a written statement on
actions taken on these recommendations to the Senate Committee on
Governmental Affairs and the House Committee on Government Reform and
Oversight not later than 60 days after the date of this letter.  A
written statement also must be sent to the House and Senate
Committees on Appropriations with the agency's first request for
appropriations made more than 60 days after the date of this letter. 

We are sending copies of this report to pertinent congressional
committees with responsibilities related to IRS, the Secretary of the
Treasury, and other interested parties.  Copies will be made
available to others upon request. 

The major contributors to this report are listed in appendix II.  If
you have any questions, please contact me on (202) 512-9044. 

Sincerely yours,

Natwar Gandhi
Associate Director, Tax Policy
 and Administration Issues


TRADE-OFFS WITH ALTERNATIVE
SAMPLING STRATEGIES
=========================================================== Appendix I


Sample
characteristic
s               Positive                         Losses
--------------  -------------------------------  -------------------------------
Original        Useable to update DIF scores,    Large sample size requiring
sample          provides baseline compliance     significant resource and cost
                for market segments, useable at  commitment.
                the DORA level for most market
                segments, very precise compared
                with other options.

Decrease the    Useable to update DIF scores,    Not useable at the DORA level.
precision of    provides baseline data for
the sample      national market segments,
                reduces the sample size and
                burden.

Sample          Useable to update DIF formula    Not usable to update the DIF
business        for businesses, where the most   score for individual returns,
returns only    noncompliance occurs, provides   not useable at the DORA level.
                baseline data for national
                market segments, reduces the
                sample size and burden.

Classify        Possibly useable to update DIF   Not useable at the DORA level,
returns and     formulas, would provides some    problems identifying no-change
exclude those   national market segment          returns.
that appear to  information, reduces the sample
be no-change    size and burden.
returns

Sample          Useable to update DIF formulas   Not useable to update the DIF
corporations    for selected classes of          score or identify compliance
and individual  business return, provides        issues for nonbusiness
businesses      national market segment          individuals, partnerships, and
only            compliance data, reduces the     S corporations not useable at
                burden on individual taxpayers.  the DORA level.
--------------------------------------------------------------------------------
Source:  Based on prior TCMP work and discussions with IRS staff. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix II

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C. 

Thomas D.  Short, Assistant Director, Tax Policy and Administration
Issues

SAN FRANCISCO REGIONAL OFFICE

Ralph T.  Block, Assistant Director
Louis G.  Roberts, Evaluator-in-Charge


*** End of document. ***