Tax Administration: Assessment of IRS' Report on Its Fiscal Year 1995
Compliance Initiatives (Letter Report, 08/27/97, GAO/GGD-97-158).

Pursuant to a congressional request, GAO reviewed the Internal Revenue
Service's (IRS) Fiscal Year (FY) 1995 Compliance Initiatives Report,
focusing on: (1) the methodology IRS used to allocate staff years and
revenues between the base enforcement programs and the compliance
initiatives; and (2) certain caveats to consider in interpreting IRS'
reported results.

GAO noted that: (1) IRS could not compile actual revenue results from
the FY 1995 compliance initiatives because the Enforcement Revenue
Information System (ERIS) only provides information on the total amount
of revenue collected as a result of enforcement activities and because
other systems, such as those that track IRS staffing, also do not
account separately for base enforcement activities and initiative
activities; (2) therefore, IRS developed a new methodology to allocate
FY 1995 enforcement revenues between base programs and the initiatives;
(3) IRS' new methodology: (a) accounted for the opportunity costs
associated with moving experienced staff off-line to train new staff;
(b) provided that no staff or revenue would be allocated to the
initiatives until planned staffing for base programs had been achieved;
and (c) improved the Compliance Initiatives Report's usefulness to
Congress by including total staffing and total revenue for the various
enforcement programs, allocated between base and initiatives, along with
explanations for variances between the results anticipated when the
initiatives were approved and the estimated final results; (4) although
the methodology used for the FY 1995 initiatives is an improvement over
previous methodologies, the results of that methodology are estimates
that are sensitive to assumptions embedded in the methodology about the
productivity of new staff and more experienced staff; (5) those
assumptions were based on the judgments of IRS managers rather than
empirical data; (6) GAO does not know what the correct assumptions are,
but GAO's sensitivity analyses showed that a change in productivity
rates could have a significant effect on the reported results; (7) in
considering IRS' estimates of the FY 1995 compliance initiatives, there
are two other caveats that are relevant; (8) the fact that the
initiatives generated a certain amount of revenue in FY 1995 does not
necessarily mean that IRS collected more enforcement revenue in FY 1995
than it did in FY 1994 but only that IRS collected more enforcement
revenue in FY 1995 than it had estimated it would collect without the
initiatives; (9) in fact, the amount of enforcement revenue IRS reported
collecting in FY 1995 was less than that reported for FY 1994; and (10)
because, the estimates of revenue attributable to the compliance
initiatives depended on various assumptions, including how IRS decided
to allocate staff, the results in FY 1995 are not necessarily indicative
of what other compliance initiatives would generate in their first year.

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

 REPORTNUM:  GGD-97-158
     TITLE:  Tax Administration: Assessment of IRS' Report on Its Fiscal 
             Year 1995 Compliance Initiatives
      DATE:  08/27/97
   SUBJECT:  Tax administration
             Government collections
             Productivity
             Human resources training
             Statistical methods
             Evaluation methods
             Comparative analysis
             Projections
             Agency reports
IDENTIFIER:  IRS Compliance Initiative
             IRS Enforcement Revenue Information System
             IRS Automated Financial System
             IRS Taxpayer Compliance Measurement Program
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** 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 Ranking Minority Member, Committee on Governmental
Affairs, U.S.  Senate

August 1997

TAX ADMINISTRATION - ASSESSMENT OF
IRS' REPORT ON ITS FISCAL YEAR
1995 COMPLIANCE INITIATIVES

GAO/GGD-97-158

IRS' FY 1995 Compliance Initiatives

(268739)


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

  ERIS - Enforcement Revenue Information System
  FTE - Full-time equivalent
  IRS - Internal Revenue Service

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


B-272840

August 27, 1997

The Honorable John Glenn
Ranking Minority Member
Committee on Governmental Affairs
United States Senate

Dear Senator Glenn: 

As part of its fiscal year 1995 appropriation, the Internal Revenue
Service (IRS) received $405 million to fund various compliance
initiatives.  That appropriation was intended to be the first
installment of a 5-year, $2.025 billion project, and IRS agreed to
report quarterly to Congress on the progress of the initiatives. 
However, the initiatives were terminated after the first year.\1
Thus, the report covering fiscal year 1995 (hereafter called the
"Compliance Initiatives Report") became the final report.\2

IRS originally estimated that the compliance initiatives would
generate $9.2 billion in additional enforcement revenue over 5 years,
with $331 million projected for fiscal year 1995.\3 The initiatives
were funded on the basis of this estimate, which is referred to in
the Compliance Initiatives Report as IRS' "commitment." Before the
start of fiscal year 1995, IRS revised that estimate after altering
its initiatives' staffing plans.  The revised estimate showed a
projected return on investment of $9.6 billion in 5 years, with
$728.3 million coming in fiscal year 1995.\4

In response to concerns we had raised in reports on compliance
initiatives funded in years before fiscal year 1995,\5 IRS developed
a new methodology to estimate the results of the fiscal year 1995
compliance initiatives, as reported in the Compliance Initiatives
Report.  Under the new methodology, which is discussed in more detail
later in this report, IRS starts with total enforcement revenue taken
from the Enforcement Revenue Information System (ERIS).  That system,
which was designed to track actual enforcement results in terms of
revenue collected, showed total enforcement revenues of $31,431.3
million in fiscal year 1995.  IRS then used a formula to allocate the
total revenue from ERIS between already existing (or "base")
enforcement programs and new compliance initiatives.  Using the
formula, IRS estimated that its base enforcement programs had
generated $30,628.0 million in fiscal year 1995 and that the fiscal
year 1995 initiatives had generated another $803.3 million in
additional enforcement revenue that year. 

This report responds to your request that we review the Compliance
Initiatives Report.  Specifically, we (1) assessed the methodology
IRS used to allocate staff years and revenues between the base
enforcement programs and the compliance initiatives and (2)
identified certain caveats to consider in interpreting IRS' reported
results.  As agreed with your office, this report does not discuss
the reliability of ERIS, which was the source of revenue data in the
Compliance Initiatives Report.\6 The reliability of ERIS is the
subject of another assignment currently under way. 


--------------------
\1 The administration's budget request for fiscal year 1996 included
about $4.5 billion for tax law enforcement, which was to fund, among
other things, the second year of the compliance initiatives.  When
Congress reduced that request to about $4.1 billion, IRS terminated
the initiatives. 

\2 IRS FY 1995 Compliance Initiative Final Report (Document 9389,
Jan.  1996). 

\3 IRS expected the first year's revenue to be relatively small
because of the amount of time and cost involved in hiring and
training new staff.

As used in the Compliance Initiatives Report and in this report,
"enforcement revenue" includes the direct revenue resulting from
enforcement actions, such as audits, delinquent return
investigations, or efforts to collect delinquent tax debts. 
According to an IRS official, "enforcement revenue" does not include
any revenue that might result indirectly from those enforcement
actions, such as might occur if voluntary compliance increased as a
result of an increase in IRS' enforcement presence. 

\4 Although IRS reestimated the revenues to be generated by the
compliance initiatives, it did not show the revised estimate in the
Compliance Initiatives Report.  Instead, the report compares the
results of the initiatives with IRS' commitment of $331 million. 

\5 See Tax Administration:  IRS' Implementation of the 1987 Revenue
Initiative (GAO/GGD-88-16, Dec.  2, 1987); Tax Administration: 
Difficulties in Accurately Estimating Tax Examination Yield
(GAO/GGD-88-119, Aug.  8, 1988); Tax Administration:  IRS Needs More
Reliable Information on Enforcement Revenues (GAO/GGD-90-85, June 20,
1990); Tax Administration:  IRS' Improved Estimates of Tax
Examination Yield Need to Be Refined (GAO/GGD-90-119, Sept.  5,
1990); Tax Administration:  IRS' Implementation of Certain Compliance
Initiatives (GAO/GGD-92-45FS, Jan.  30, 1992); Tax Administration: 
Congress Needs More Information on Compliance Initiative Results
(GAO/GGD-92-118, July 31, 1992). 

\6 ERIS is an automated database that tracks enforcement cases across
functional lines from initial enforcement action (such as an audit)
to resolution (such as collection or abatement). 


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

IRS could not compile actual revenue results from the fiscal year
1995 compliance initiatives because ERIS only provides information on
the total amount of revenue collected as a result of enforcement
activities and because other systems, such as those that track IRS
staffing, also do not account separately for base enforcement
activities and initiative activities.  Therefore, IRS developed a new
methodology to allocate fiscal year 1995 enforcement revenues between
base programs and the initiatives.  That methodology represented a
significant improvement over past methodologies. 

IRS' new methodology (1) accounted for the opportunity costs
associated with moving experienced staff off-line to train new staff;
(2) provided that no staff or revenue would be allocated to the
initiatives until planned staffing for base programs had been
achieved; and (3) improved the Compliance Initiatives Report's
usefulness to Congress by including total staffing and total revenue
for the various enforcement programs, allocated between base and
initiatives, along with explanations for variances between the
results anticipated when the initiatives were approved and the
estimated final results. 

Although the methodology used for the fiscal year 1995 initiatives is
an improvement over previous methodologies, the results of that
methodology are estimates that are sensitive to assumptions embedded
in the methodology about the productivity of new staff and more
experienced staff.\7 Those assumptions were based on the judgments of
IRS managers rather than empirical data.  We do not know what the
correct assumptions are, but our sensitivity analyses showed that a
change in productivity rates could have a significant effect on the
reported results. 

In considering IRS' estimates of the fiscal year 1995 compliance
initiatives, there are two other caveats that are relevant.  First,
the fact that the initiatives generated a certain amount of revenue
in fiscal year 1995 does not necessarily mean that IRS collected more
enforcement revenue in fiscal year 1995 than it did in fiscal year
1994 but only that IRS collected more enforcement revenue in fiscal
year 1995 than it had estimated it would collect without the
initiatives.  In fact, the amount of enforcement revenue IRS reported
collecting in fiscal year 1995 was less than that reported for fiscal
year 1994.  Second, because the estimates of revenue attributable to
the compliance initiatives depended on various assumptions, including
how IRS decided to allocate staff, the results in fiscal year 1995
are not necessarily indicative of what other compliance initiatives
would generate in their first year. 


--------------------
\7 Productivity is defined as the amount of dollars collected per
staff year. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :2

To assess IRS' methodology, we

  -- discussed the methodology with officials in the Financial
     Analysis Division of the Office of the Chief Financial Officer,
     which was the division responsible for preparing the Compliance
     Initiatives Report;

  -- determined the extent to which IRS' methodology addressed the
     concerns we had raised in reports on prior years' compliance
     initiatives;

  -- discussed specific assumptions in IRS' methodology with
     cognizant staff in IRS' enforcement functions;

  -- assessed the sensitivity of IRS' results to changes in certain
     key assumptions; and

  -- verified the accuracy of IRS' computations. 

IRS computed the results of the fiscal year 1995 initiatives using
data on planned and actual full-time-equivalent (FTE) staff years and
planned and actual revenues for various enforcement programs.  To
verify the accuracy of IRS' computations, we (1) traced actual FTEs
back to reports generated by IRS' Automated Financial System, (2)
traced actual revenues back to reports generated by ERIS, (3)
interviewed staff from IRS' enforcement functions about the methods
used to develop planned FTEs and revenues, (4) recomputed
calculations, and (5) resolved any inconsistencies with cognizant IRS
staff.  We did not assess the reliability of the reports generated by
ERIS or the Automated Financial System. 

We did our work from June 1996 to May 1997 in accordance with
generally accepted government auditing standards.  We requested
comments on a draft of this report from the Acting Commissioner of
Internal Revenue or his designated representative.  The Acting
Commissioner provided comments in a letter dated July 24, 1997. 
Those comments are reprinted in appendix II and are summarized and
evaluated at the end of this letter. 


   IRS USED AN IMPROVED
   METHODOLOGY TO ALLOCATE
   ENFORCEMENT REVENUES BETWEEN
   BASE PROGRAMS AND INITIATIVES,
   BUT ESTIMATES VARY DEPENDING ON
   ASSUMPTIONS
------------------------------------------------------------ Letter :3

In reviews of past compliance initiatives, we identified several
weaknesses in IRS' methodology for computing and tracking initiative
results.  Our past reviews disclosed, for example, that IRS

  -- had overstated initiative results by failing to recognize the
     opportunity costs associated with moving experienced staff
     off-line to train new staff and by failing to adequately account
     for underrealizations of planned base staffing,

  -- was unable to track actual enforcement results, and

  -- was not providing Congress with enough meaningful information on
     initiative results because it was reporting positive results
     from initiatives without recognizing negative results from
     reductions in base activities. 

Over the years, IRS revised its methodology to address those, and
other, concerns.  In preparing its fiscal year 1995 Compliance
Initiatives Report, for example, IRS recognized the impact of
opportunity costs, obtained revenue data from an automated system
(ERIS) that was designed to track actual enforcement results, and
improved the report's usefulness to Congress by including not only
the estimated results of the initiatives but also the estimated
results of the base enforcement programs and explanations for
variances between the results anticipated when the initiatives were
approved and the estimated final results.  Also, in computing the
results of the fiscal year 1995 compliance initiatives, IRS adopted a
rule that no FTEs, and thus no revenue, would be allocated to the
initiatives until planned base staffing had been achieved. 

Although the methodology used for the fiscal year 1995 initiatives is
an improvement over previous methodologies, the results of that
methodology are estimates that are sensitive to various productivity
assumptions.  Those productivity assumptions were not based on
empirical data and could, if they were erroneous, cause IRS' reported
results to be overstated or understated.  Also, in verifying IRS'
calculations, we found four errors that had a relatively minor effect
on IRS' reported results. 


      WHAT IRS REPORTED AS THE
      INITIATIVES' ACTUAL RESULTS
      ARE ESTIMATES
---------------------------------------------------------- Letter :3.1

IRS developed and uses ERIS to track the results of enforcement
activities.  Assuming ERIS does what it is designed to do, it should
provide the total amount of dollars actually collected as a result of
enforcement activities in fiscal year 1995 (i.e., $31.4 billion). 
However, as IRS acknowledged in the Compliance Initiatives Report,
ERIS does not distinguish between the dollars collected from base
enforcement activities and the dollars collected as a result of the
initiatives.  Similarly, the Automated Financial System, from which
IRS extracted the total number of FTEs spent on enforcement
activities in fiscal year 1995, does not distinguish between staffing
for base activities and for the initiatives. 

Because its systems do not distinguish between base and initiative
activities, IRS, as part of its methodology, developed the formula we
describe in appendix I to allocate the $31.4 billion in enforcement
revenue between base and initiative activities.  Before implementing
its new methodology, IRS briefed us on the allocation formula.  We
said then that because IRS had no initiative-specific data, its
formula was a reasonable approach for identifying initiative results. 
We continue to believe that.  However, because planned (i.e.,
estimated) revenue and staffing levels are an integral part of that
formula, the end results are estimates.  Thus the "actual" initiative
results cited in the Compliance Initiatives Report are not actual
results but are estimates of results.  IRS did not clearly disclose
that fact in the Compliance Initiatives Report.  In tables throughout
the report, for example, IRS refers to "actual," without clearly
explaining the term. 


      IRS' PRODUCTIVITY
      ASSUMPTIONS ARE NOT BASED ON
      EMPIRICAL DATA
---------------------------------------------------------- Letter :3.2

One important set of assumptions embedded in IRS' methodology relates
to the comparative productivity of new staff versus experienced
employees.  IRS used these assumptions in computing planned revenue,
which was an integral part of the methodology.  The two primary IRS
enforcement functions, Collection and Examination, approached the
issue of relative productivity differently.\8 Collection assumed that
new staff were less productive than experienced staff, even after
they were trained, while Examination believed that new staff, once
trained, were as productive as experienced employees.  Neither of
these assumptions was based on empirical data. 


--------------------
\8 According to the Compliance Initiatives Report, these two
functions accounted for 97 percent of the $803.3 million that IRS
estimated was generated by the initiatives.  Of the $803.3 million,
$545.2 million (68 percent) was attributed to Collection.  The other
$237.1 million was attributed to Examination. 


         COLLECTION
-------------------------------------------------------- Letter :3.2.1

According to IRS officials, to estimate the relative productivity of
new Collection staff, Collection Division officials met in a
brainstorming session and decided, based primarily on their
institutional knowledge, that new staff were generally 50 percent as
productive as experienced employees during their first year on the
job.\9 Part of that reduced productivity assumed by Collection is
attributable to the amount of time new staff spend in training and
part to the belief that it takes time for a new employee to become as
productive as an experienced one.  (Collection assumed that new
employees do not reach full productivity until their second or third
year, depending on their position). 

We have no basis to determine whether Collection's productivity
assumptions are correct.  We do know, however, that changes to the
assumptions could significantly alter the reported results of the
compliance initiatives.  To demonstrate the sensitivity of the
reported results to changes in Collection's productivity assumptions,
we arbitrarily adjusted Collection's 50-percent assumption by 5
percentage points in either direction and recalculated the
initiatives' results.  Our recalculation showed that a 5 percentage
point change would either increase or decrease IRS' reported
initiative results of $545.2 million by $42 million (about 8
percent), depending on the direction of the change. 


--------------------
\9 The 50-percent adjustment was used for most Collection FTEs.  The
only exception was for those FTEs allocated to work on cases
involving delinquent returns.  For those FTEs, Collection assumed
that new staff were 64 percent as productive as experienced employees
in doing that work. 


         EXAMINATION
-------------------------------------------------------- Letter :3.2.2

Unlike Collection, the Examination function assumes that new staff
are as productive as experienced staff after they have completed
classroom training.  That assumption applies to all of Examination's
enforcement staff--tax examiners, who audit simple issues by
corresponding with taxpayers; tax auditors, who do more complex
audits generally by meeting with taxpayers at an IRS office; and
revenue agents, who do the most complex audits generally by meeting
taxpayers or their representatives at the taxpayer's home or place of
business. 

Examination officials told us that they did not have any empirical
data to support their assumption that new tax examiners are as
productive as experienced tax examiners because Examination's
information systems do not track individual tax examiner's
accomplishments.  Thus, historical data cannot be separated between
experienced and inexperienced tax examiners.  Instead, Examination
officials justified their assumption by noting that tax examiners
work on relatively noncomplex issues and returns and are able to get
up to speed fairly quickly.  However, according to an IRS official,
after 2 weeks of formal classroom training, tax examiners are
assigned to an on-the-job instructor for 10 weeks.  The need for
on-the-job instructors suggests that IRS does not expect new
employees to be able to handle issues and returns as effectively or
efficiently, and thus be as productive, as experienced employees. 

Examination officials also said that they did not believe that it was
necessary to assume different productivity levels for experienced
revenue agents and tax auditors and new agents and auditors.  The
officials provided two reasons for their position. 

  -- First, Examination adjusts its revenue estimates to consider the
     amount of time that new staff must spend in classroom training. 
     Examination assumes, based on past experience, that the amount
     of time available to do audit work is reduced between 19 and 25
     percent (it varies between revenue agents and tax auditors)
     because of classroom training requirements.  However, that
     adjustment only affects the amount of time new staff have to do
     audits; it does not get at the issue of how productive new staff
     are when they are doing audits. 

  -- Second, Examination's resource allocation model imputes a lower
     marginal yield for each additional return audited by initiative
     revenue agents and tax auditors.  That means, in effect, that
     Examination assumes that each additional return audited will
     generate less revenue than the return audited before it.\10
     However, although IRS assumes a lower yield with each additional
     audited return, it does not consider any differences in the
     efficiency with which experienced staff and new staff audit tax
     returns. 

In that regard, the officials said that, after completion of
classroom training, new staff are expected to close the same number
of cases in any year as experienced employees who work the same types
of cases.  But the productivity of Examination staff is determined
not only by the number of cases they close in a year but also by the
revenue generated from those cases.  Thus, even if new staff were to
close as many cases as experienced staff (Examination officials could
provide no evidence to support that contention), they may or may not
achieve comparable dollar results.  According to an Examination
official, revenue agents and tax auditors, during their first year on
the job, have 10 weeks and 13 weeks, respectively, of on-the-job
training after their classroom training.  The need for on-the-job
training suggests, in our opinion, that new staff may not be prepared
to be as productive as experienced staff, after classroom training. 

Although we did not have data to test the appropriateness of
Examination's productivity assumption, we tested the sensitivity of
IRS' reported initiative results to changes in that assumption by
assuming the following: 

  -- New tax auditors would be 75 percent as productive as
     experienced tax auditors. 

  -- New revenue agents would be 69 percent as productive as
     experienced revenue agents. 

  -- New tax examiners would be 95 percent as productive as
     experienced tax examiners. 

We arrived at the 75-percent and 69-percent figures for tax auditors
and revenue agents, respectively, by using Collection's productivity
assumption of 50 percent and adjusting it to recognize the fact that
Examination's assumption already includes a factor for lost
productivity due to training.  We assumed only a slight fall-off in
productivity for tax examiners because the correspondence audits they
do are straightforward, and thus, the skills needed to do them can be
more quickly learned than the skills needed by tax auditors and
revenue agents.  Use of the different productivity assumptions
reduced the initiative results by $12.9 million--a reduction of about
5 percent from the reported amount of $237.1 million that IRS
estimated was generated by Examination through the initiatives. 


--------------------
\10 These estimates of marginal yield per return are based on data
collected from the 1988 Taxpayer Compliance Measurement Program.  We
did not assess the sensitivity of IRS' initiative revenue estimates
to changes in the precision of the marginal yield estimates. 


         ONLY MINOR ERRORS WERE
         FOUND IN IRS'
         CALCULATIONS
-------------------------------------------------------- Letter :3.2.3

We traced the data in the Compliance Initiatives Report back to
supporting documentation and verified the various calculations
involved in using IRS' methodology.  We found four relatively minor
errors. 

The first two errors involved the categorization of FTEs between
revenue- and nonrevenue-producing FTEs.  A portion of the staffing
increase associated with compliance initiatives is for support staff,
such as clerks and secretaries.  Those staff, unlike revenue
officers, revenue agents, and other frontline staff, do not directly
generate revenue.  Thus IRS, in its calculations, segregated
nonrevenue-producing FTEs from revenue-producing FTEs.  In verifying
that part of IRS' calculations, we found two errors.  IRS mistakenly
(1) categorized about 140 nonrevenue-producing FTEs in the Collection
function as revenue-producing and (2) included 240
nonrevenue-producing Examination FTEs in the formula it used to
allocate enforcement revenue between base activities and initiatives. 
The third error involved the use of an incorrect average yield figure
in computing initiative results for the Compliance Research program. 
The fourth error involved a failure to include about 180 tax
examiners in computing the initiative results for the Collection
function. 

By our calculations, these four errors--which IRS officials
acknowledged--caused IRS' reported yield from the initiatives to be
understated by $2.6 million.  Absent other changes, correction of the
four errors would increase IRS' reported yield to $805.9 million. 


   CAVEATS TO CONSIDER IN
   INTERPRETING IRS' REPORTED
   RESULTS
------------------------------------------------------------ Letter :4

In considering IRS' estimates of the results of the fiscal year 1995
compliance initiatives, there are two other caveats that are
relevant:  (1) the fact that IRS collected a certain amount in fiscal
year 1995 as a result of the initiatives does not necessarily mean
that IRS collected more enforcement revenue in fiscal year 1995 than
in fiscal year 1994 but only that IRS collected more in fiscal year
1995 than it estimated it would have without the initiatives and (2)
the first year's results from the fiscal year 1995 initiatives are
not necessarily indicative of what other compliance initiatives would
generate in their first year. 


      EVEN WITH THE POSITIVE
      RESULTS REPORTED FROM THE
      INITIATIVES, IRS COLLECTED
      LESS IN FISCAL YEAR 1995
      THAN IN FISCAL YEAR 1994
---------------------------------------------------------- Letter :4.1

Although IRS reported that the compliance initiatives resulted in
additional collections in fiscal year 1995 of $803.3 million, it does
not mean that IRS brought in $803.3 million more than it did in
fiscal year 1994.  What it means is that IRS estimated that it
generated $803.3 million more in enforcement revenue in fiscal year
1995 than it had estimated it would generate without the compliance
initiatives. 

According to the Compliance Initiatives Report, IRS collected a total
of $31.4 billion in fiscal year 1995--$30.6 billion from its base
programs and $0.8 billion from the initiatives.  Despite the
estimated additional revenue from the initiatives, however, the
amount of enforcement revenue collected in fiscal year 1995 was less
than the amount collected in fiscal year 1994.  That is, IRS data
indicated that IRS collected about $33.1 billion in fiscal year 1994,
or $1.7 billion more than in fiscal year 1995.  According to IRS
officials, a number of factors could make enforcement revenue
decrease even with a staffing increase in the same year.  Revenue
collected in one fiscal year is a function not only of that year's
staffing but also of prior years' staffing.  Much of the revenue
impact of a staffing increase or decrease occurs in subsequent years
because of the possibility of appeals, litigation, and collection
activity.  In that regard, IRS officials said that enforcement
revenue in fiscal year 1996 increased to $38.0 billion, even with a
staffing decrease, partially as a result of the fiscal year 1995
compliance initiatives. 


      RESULTS OF FISCAL YEAR 1995
      COMPLIANCE INITIATIVES ARE
      NOT NECESSARILY INDICATIVE
      OF THE RESULTS THAT FUTURE
      COMPLIANCE INITIATIVES WOULD
      PRODUCE
---------------------------------------------------------- Letter :4.2

A second caveat to keep in mind is that IRS' results in fiscal year
1995 are not necessarily indicative of the results that would be
achieved in the first year of future compliance initiatives.  The
results achieved for any compliance initiative depend on many factors
that can, and most likely will, vary from one initiative to another. 
Two of those factors, both of which had a significant impact on the
results achieved in fiscal year 1995, are (1) the extent to which new
staff must be hired to fill the positions authorized by the
initiatives and (2) how IRS decides to allocate the initiative
positions among its various enforcement programs. 

Most of the positions funded with the $405 million provided for the
compliance initiatives in fiscal year 1995 were filled not by new
hires but by staff who were already on board.  This happened, at
least in part, because IRS' fiscal year 1995 appropriation, except
for the compliance initiatives, actually resulted in reductions in
IRS' enforcement staffing.  Thus, many of the positions funded by the
$405 million were used to offset that reduction.  According to data
provided by IRS, of the 5,470 initiative positions filled as of
September 30, 1995, 1,145 were filled by new staff.  The other 4,325
were filled by existing employees either through lateral
reassignments or promotions (such as promoting tax auditors to fill
revenue agent positions).  If future initiatives require a greater
proportion of new staff, the results could be different from those in
fiscal year 1995 because new staff (1) require more training, which,
under IRS' current procedures for training new staff, increases the
opportunity cost associated with moving experienced staff off-line to
do the training and (2) generally can be expected to generate less
revenue, at first, than experienced staff. 

Decisions on how to allocate staff among enforcement programs also
affects initiative results.  The change in IRS' estimate of how much
revenue would be generated by the fiscal year 1995 initiatives, which
we noted at the beginning of this report, is an example of how
staffing decisions can affect initiative results.  IRS' first
estimate was $9.2 billion over 5 years and $331 million in fiscal
year 1995.  Then, in an effort to maximize revenues, IRS decided to
allocate more of the $405 million to areas, such as Automated
Collection System sites, that are staffed by lower graded personnel
and to allocate fewer dollars to more costly areas, such as the
Collection Field Function, which is staffed by higher graded revenue
officers.\11 This reallocation enabled IRS to fund many more FTEs
than originally expected and resulted in revised estimates of $9.6
billion over 5 years and $728 million in fiscal year 1995. 


--------------------
\11 This reallocation of resources was partly in response to a
recommendation we had made in a 1993 report entitled Tax
Administration:  New Delinquent Tax Collection Methods for IRS
(GAO/GGD-93-67, May 11, 1993).  In that report, we recommended that
IRS restructure its collection organization to support earlier
telephone contact with delinquent taxpayers and determine how to use
collection staff in earlier, more productive phases of the collection
cycle. 


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

IRS' methodology for computing the results of the fiscal year 1995
compliance initiatives is a significant improvement over past
methodologies.  However, there are productivity assumptions embedded
in the methodology that are not based on empirical data and could
cause the results of an initiative to be overstated or understated. 
We do not have a basis for determining what the correct assumptions
should be, but our sensitivity analyses showed that a change in the
assumptions used could have a significant effect on the reported
initiative results of $803.3 million.  For example, as discussed
earlier, changing Collection's productivity assumption by 5
percentage points would either increase or decrease the reported
results by $42 million and changing Examination's assumptions to be
comparable to Collection's would decrease the reported results by
$12.9 million.  After adjusting IRS' reported results for the effect
of these different productivity assumptions and after increasing the
results to account for the $2.6 million in calculation errors, the
estimated yield from the fiscal year 1995 compliance initiatives
would fall somewhere between $751.0 million and $847.9 million. 


   AGENCY COMMENTS AND OUR
   EVALUATION
------------------------------------------------------------ Letter :6

We requested comments on a draft of this report from the Acting
Commissioner of Internal Revenue.  IRS provided comments in a letter
dated July 24, 1997 (see app.  II).  Overall, IRS agreed with the
findings in the report, which it said confirmed that IRS accurately
tracked and reported on the results of the compliance initiatives. 
However, IRS expressed concerns about two aspects of the observations
in the report. 

First, IRS took issue with our statement that it did not clearly
disclose the fact that the initiative results cited in the Compliance
Initiatives Report were not "actual" results but rather estimates. 
To show that it had clearly disclosed that fact, IRS quoted two
passages from the Compliance Initiatives Report which refer to the
allocation of the revenue to base and initiative activities.  In our
opinion, the statements IRS quoted, while informative to a careful
reader, do not provide sufficient information to make clear that the
results are estimates, especially when every table in the final
report referred to the results as "actuals."\12

Second, IRS observed that, although we state that we did not assess
the reliability of the data in ERIS, (1) after reviewing the
Compliance Initiatives Report and supporting ERIS data, we have found
no indication that ERIS does not do what it purports to
do--accurately accumulate and summarize enforcement revenue; (2) work
done to date on ERIS as part of our financial audit of IRS had not
disclosed any problems; and (3) an ERIS sample we took as part of an
audit of large corporations concluded that ERIS does what it purports
to do.  IRS also observed that we have asked for ERIS data in
conjunction with two recently initiated audits. 

We appreciate IRS' viewpoint; however, (1) the scope of our review of
the Compliance Initiatives Report, as discussed earlier, did not
include a specific assessment of ERIS reliability; (2) we have not
yet done sufficient work on ERIS as part of the financial audit to
reach any overall conclusion about data reliability; and (3) the
scope of our audit of large corporations was also not broad enough to
reach any overall conclusion about ERIS reliability.  Until such time
as ERIS' reliability can be determined, we necessarily continue to
rely on ERIS data for some of our work because they are the best data
available.  However, the standards applicable to our work require
that we disclose that data reliability has not been confirmed. 


--------------------
\12 IRS also quoted similar language in its letter from several other
documents to support this point.  However, we do not believe that
language in other documents is directly relevant to understanding the
content of the final Compliance Initiatives Report because they would
not be available to most readers. 


---------------------------------------------------------- Letter :6.1

We are sending copies of this report to the Committee Chairman; the
Chairmen and Ranking Minority Members of the Senate Committee on
Finance, the House Committee on Ways and Means, and the House
Committee on Government Reform and Oversight; various other
congressional committees; the Secretary of the Treasury; the
Commissioner of Internal Revenue; the Director of the Office of
Management and Budget; and other interested parties.  We will also
make copies available to others upon request. 

Major contributors to this report are listed in appendix III.  If you
have any questions, please contact me on (202) 512-9110. 

Sincerely yours,

Lynda D.  Willis
Director, Tax Policy and
 Administration Issues


IRS' METHODOLOGY FOR CALCULATING
INITIATIVE RESULTS
=========================================================== Appendix I

According to the Compliance Initiatives Report, IRS estimated the
amount of revenue attributable to the fiscal year 1995 compliance
initiatives by means of a general formula that consists of two
components.  The first component is the ratio of total actual yield
per FTE (base and initiative combined) divided by the total planned
yield per FTE (base and initiative combined).  This component
indicates by how much the average tax yield was over or under what
IRS expected.  If IRS accurately predicted the average yield, the
ratio value of the first component would equal 1.  If the actual
yield was more or less than what was predicted, the ratio value would
be greater or less than 1.  The second component multiplies the
number of "actual" FTEs allocated to the initiatives by the planned
yield per initiative FTE. 

The result of the second component is then multiplied by the result
of the first component.  If the first component equaled 1, the yield
attributable to the initiatives would be equal to the number of
initiative FTEs times what yield they were predicted, on average, to
realize.  However, if the first component was less than 1, implying
that the average yield of both base and initiative staff was less
than expected by some percentage, the expected tax yield per
initiative FTE would be reduced by the same percentage.  Conversely,
if the ratio value of the first component was greater than 1, the
second component would be increased by the percentage by which the
actual average yield exceeded what was expected. 

To illustrate, assume the following information: 

Total actual yield -- $70 million
Actual number of FTEs (base and initiative) -- 1,800
Actual yield per FTE ($70 million divided by 1,800 FTEs) -- $38,889

Planned total yield -- $85 million
Planned number of FTEs (base and initiative) -- 1,500
Planned yield per FTE ($85 million divided by 1,500 FTEs) -- $56,667

First component:  $38,889 divided by $56,667 = 0.686 or 68.6 percent. 

Planned initiative yield -- $5 million
Planned initiative FTEs -- 400
Planned yield per initiative FTE -- $12,500
"Actual" initiative FTEs -- 700

Second component:  700 times $12,500 = $8,750,000. 

"Actual" initiative yield:  0.686 times $8,750,000 = $6,002,500. 

The above information is an illustration.  The numbers and thus the
results would vary among initiatives and even within an initiative. 
For example, Examination assigned its initiative staff to various
audit classes ("individual taxpayers who file a Form 1040C showing
total gross receipts of less than $25,000 dollars" is an example of
an audit class).  For each audit class, IRS would apply the above
formula; and for each audit class, the numbers and the results would
be different. 




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



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)


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


   GENERAL GOVERNMENT DIVISION,
   WASHINGTON, D.C. 
------------------------------------------------------- Appendix III:1

David J.  Attianese, Assistant Director
James A.  Wozny, Assistant Director
John Lesser, Evaluator-in-Charge
Charles C.  Tuck, Economist


*** End of document. ***