Tax Administration: IRS' Improved Estimates of Tax Examination Yield Need to Be Refined

Published by the Government Accountability Office on 1990-09-05.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

                                                                 Ii- .---
                                                                      It i t.td---Sh t t’s Gt*lltbral Actwurrl,illg
                            ..__.....-___._,.I_. .-“‘.-..-“..-_-__.....
                       ____..,..                                                                                      Ol’l’iw     --~     ._-_--.
                                                                        Ikport, Lo tGk%air.rnan, Cornmitt,ee on
GAO                                                                     1,Iw Ihltiget,, I J.S. Senate

                                             ..-“..--.-I   ---.-
N~~~,t~~rIrlwr             1!1!~0
                                                                         TAX --
                                                                         IRS’ Improved
                                                                         Estimates of Tax
                                                                         Examination Yield
                                                                         Need to Be Refined



 ____-,... -.ll-.--”        -__.___- __..I.-.-.              --    ..__---
 GAo/G(;I)-!w-l                       I!)
GAO                General Accounting Office
                   Washington, D.C. 20548

                   General Government          Division


                   The Honorable Jim Sasser
                   Chairman, Committee on
                     the Budget
                   United States Senate
                   Dear Mr. Chairman:
                   The Internal RevenueService’s(IRS) budget for fiscal year 1991 includes
                   several initiatives that are intended to generateadditional tax revenues
                   by increasing or modifying IRS’ compliance efforts. Oneinitiative calls
                   for increasing IRS’examination staff by adding 760 revenue agentswho
                   audit the more complex tax returns and 290 support staff, According to
                   IRS,the audits resulting from such a staff increasewill generate addi-
                   tional audit revenuesof $1.1 billion by the end of fiscal year 1996. This
                   report respondsto your request for information on the methodology IRS
                   used to estimate those additional audit revenuesand its plans to track
                   the results of the staffing increase-if the increaseis authorized by

Results in Brief   IRS revised its methodology for estimating audit yield following our 1988
                   report on problems with the methodology IRS was using at that time.’ IRS
                   used its revised methodology to compute the $1.1 billion estimate associ-
                   ated with the proposed fiscal year 1991 staffing increase.Although the
                   revised methodology produces a more realistic expectation of the addi-
                   tional revenuesto be derived from an increasein examination staff, we
                   consider someof the assumptionsIRS used in applying the methodology
                   to the proposed fiscal year 1991 staffing increaseto be overly opti-
                   mistic. IRS assumedthat all the new staff would be on board by the
                   beginning of the fiscal year and that a revised (and less costly) training
                   program would be in place at that time-both of which have proven to
                   be overly optimistic considering their current status. Delays in one or
                   both of those events would delay realization of benefits from the
                   staffing increase.IRS also assumedthat the influx of new staff would
                   free up experiencedagentsto work on higher-yield cases,thus further
                   increasing the audit revenuesgeneratedby increasedstaffing. We ques-
                   tioned that assumption in our assessmentof IRS’ past methodology. IRS
                   agreesthat it needsto further study that issue.

                   ‘Tax Adminiitration: Difficulties in Accurately Ektimating Tax Examination Yield (GAO/
                        - 8. 119, Aug. 8,19@).

                   Page 1                                                     GAO/GGD9&119      TM Administration
                          If the examination staffing increaseis authorized, IRSplans to monitor
                          its impact. The reliability of that monitoring information will depend,in
                          large part, on the validity of the baseline from which IRSbegins tracking
                          results. BecauseIRS had not yet computed its baseline,we could not
                          assessits validity.

                          In our August 1988 report to the Chairman of the SenateBudget Com-
Background                mittee, we assessed(1) the methodology IRS was then using to estimate
                          the additional revenuesassociatedwith an increasein examination staff
                          and (2) IRS’ estimate of the amount of revenue actually generatedin
                          fiscal year 1987 as a result of a congressionally authorized increasein
                          examination staff that year. We raised several concernsabout IRS’ meth-
                          odology for estimating the additional revenuesit expectedto realize
                          from a staffing increaseand reported that IRS’ estimate of the increased
                          revenuesit actually realized in fiscal year 1987 was overstated. The
                          overstatement occurred primarily becauseIRS failed to account for any
                          unrealized revenuesresulting from the use of experiencedrevenue
                          agentsto train new staff.
                          In responseto our concernsand those raised by the Department of the
                          Treasury, the Office of Managementand Budget, and the Congressional
                          Budget Office, IRS revised its estimating methodology.2It then used the
                          revised methodology to estimate the additional revenuesassociatedwith
                          the requested fiscal year 1991 increasein revenue agents.

                          As agreedwith the Committee, our objectives were to provide informa-
Objectives, Scope,and     tion on (1) IRS’ methodology for estimating the yield associatedwith the
Methodology               increasein revenue agentsproposed for fiscal year 1991, including any
                          available information relating to IRS’ plan to revise its revenue agent
                          training program; and (2) IRS’ plans to track the results of that staffing
                          To develop information on IRS’ estimating methodology:
                        . We reviewed past studies and reports on the estimating processand ana-
                          lyzed statistical data on Examination’s plans and accomplishments.

                          21RSconsiders the new methodology to be an interim one while it develops more specific data on
                          audit results that can be used to further improve its estimating assumptions.

                          Page 2                                                      GAO/GGD99-119     Tax Atidnistmtion

                         l We interviewed officials in the Office of Examination Planning and
                           Researchand in IRS’ResearchDivision to obtain information on the esti-
                           mating process.
                         l We analyzed documentation provided by IRS to show how its estimating
                           methodology works and to support the various assumptionsbehind that
                         9 We interviewed officials in the Office of the Assistant Commissioner
                           (Examination) about changesbeing consideredin the training program
                           for revenue agents.
                         . We interviewed officials and group managersin the Examination Divi-
                           sion at 1 of IRS’ 63 district offices (San Francisco)to obtain information
                           on IRS’ estimating assumption relating to the allocation of audit
                           resources.We choseSan Franciscobecauseit was one of the districts we
                           had visited during our assessmentof IRS’ previous methodology and
                           becausethe managersin that district were available to meet with us on
                           short notice. We supplementedwhat we learned in San Francisco with
                           information we had obtained from three other district offices (Atlanta,
                           Chicago,and St, Louis) during an earlier assignmentfor the Committee.
                             To obtain information on IRS’ plans for tracking the results of a fiscal
                             year 1991 examination staffing increase,we interviewed officials in IRS’
                             Budget Division and in the Office of Examination Planning and
                             Research,and we reviewed documentation describing the tracking
                             Our audit work required us to use data generatedby IRS’management
                             information systemsand by its computerized processfor estimating
                             audit yield. We did not assessthe reliability of those data. However, we
                             reviewed the data for reasonablenessand consistencyand attempted to
                             resolve any discrepancies.We did our audit work between April and
                             July 1990 and in accordancewith generally acceptedgovernment
                             auditing standards. IRS provided written commentson a draft of this
                             report. Those commentsare presented and evaluated beginning on page
                             14 and are included in appendix I.

                             After we issued our August 1988 report, IRS reassessedits methodology
IRS Has Improved Its         for estimating the revenue to be generatedby an increasein examina-
Methodology for              tion staff. The reassessmentidentified several problems with the meth-
Estimating Audit
                 Yield       odology and led IRS to concludethat its estimating processproduced
                             projections that “understate the time neededto realize direct benefits”

                             Page 3                                      GAO/GGD-ml19   Tax Adminbtration

                            from staffing increases.3As a result of that reassessment,IRSrevised its
                            estimating methodology. .
                            The revised methodology, which IRS used to estimate audit yield associ-
                            ated with the staffing increaseproposed for fiscal year 1991, provides a
                            more realistic indicator of the revenue implications of such an increase
                            than did IRS’ past methodology. Unlike past years when IRS estimated
                            that the government would realize significant additional revenuesin the
                            first year of a staffing increase,IRS’ new methodology has led it to con-
                            clude that the proposed staffing increasefor fiscal year 1991 will result
                            in decreasedrevenuesin the first year and will not start generating
                            increasedrevenuesuntil fiscal year 1992.
                            In revising its methodology, IRS addressedseveral concernsraised in our
                            August 1988 report including those relating to (1) the productivity of
                            new staff, (2) the audit revenuesthat go unrealized while experienced
                            examination staff train new staff, and (3) IRS’ use of outdated informa-
                            tion to determine how much of the additional tax liability recommended
                            by examination staff as a result of their audits would eventually be

Productivity of New Staff   Oneproblem with IRS’ past methodology was IRS’ assumption that the
                            additional examination time made available by an increasein staff was
                            generally applied to ongoing audits rather than to new audits. IRS’
                            revised methodology, however, assumesthat the additional time will be
                            applied to new audits, which meansthat it will be longer before any
                            audit results are realized. That change more accurately portrays how
                            new staff are used and thus how soon audits worked on by new staff
                            will be closed and how soon the government will realize revenuesfrom
                            those audits.
                            This changeis a major reason why IRS’ revised estimating methodology
                            shows almost no additional taxes, penalties, and interest being collected
                            in the first year of a staffing increasecomparedto the significant first-
                            year collections shown by its previous methodology. As IRS noted in its
                            November 1989 researchreport, for example, application of the revised
                            methodology to a hypothetical increaseof 1,000 revenue agent staff
                            years would show expected first-year collections of $1.1 million com-
                            pared to the $321.4 million shown by IRS’ earlier methodology.

                            31RS reassessment wa8 documented in a research report entitled Evaluation of the IRS System of
                            Projecting Enforcement Revenue (Publication 1601) published by IRS in November 1989.

                            Page 4                                                     GAO/GGD9O-119      Tax Admintstration
Unrealized Audit   A secondproblem with IRS’ previous methodology was its failure to rec-
Revenues           ognizethat the additional taxes, penalties, and interest generatedby
                   new revenue agentshad to be offset by the taxes, penalties, and interest
                   that went unrealized becauseexperiencedagents,who would otherwise
                   be auditing returns, would have to train the new staff. IRS’ revised meth-
                   odology recognizesthat fact.
                   In an April 1990 report to the Committee, we commentedon the extent
                   to which experiencedrevenue agentswere involved in training.about
                   1,100 new hires in five IRSdistrict offices between July 1, 1986, and
                   June 30,1989, and the cost of that involvement in terms of unrealized
                   audit revenues.4We concludedthat the government could end up
                   investing about $980 million in unrealized revenuesto train those 1,100
                   As part of the April 1990 report, we comparedour computation of
                   unrealized audit revenueswith one that IRS had made using its revised
                   estimating methodology. IRShad determined that the training associated
                   with a hypothetical increaseof 1,000revenue agent staff years nation-
                   wide would cost about $627 million in unrealized revenues.We attrib-
                   uted the difference between IRS’ and our computations to differences in
                   methodology. As noted in our April 1990 report, the appropriatenessof
                   one methodology over another might dependon the number of new rev-
                   enue agentsbeing hired and trained, We said that (1) our methodology
                   seemedmore appropriate for the period coveredby our review, when IRS
                   was training about 7,300 newly hired revenue agentsnationwide; and
                   (2) IRS’ methodology might be more appropriate during periods when it
                   is hiring, and thus training, a much smaller number of agents-which
                   would be the caseif the proposed staffing increasefor fiscal year 1991
                   is authorized.

                   4Tax Administration: Potential Audit Revenues Lost While Training New Revenue Agents (GAO/
                       - -77, Apr. 6 1990).

                   page 6                                                   GAO/GGD4MMlB Tax Adminbtratlon

Extent to Which        In our August 1988 report, we noted that IRS used information from its
RecommendedTaxes Are   tracking of audits closedin 1972 to determine how much of the recom-
                       mendedtax would be assessedand how much of the assessedamount
Collected              would eventually be collectedS6    We questioned whether that information
                       was still valid. In revising its estimating methodology, IRS recognized
                       that the 1972 audit results were no longer applicable and used data from
                       the late 1980sto translate recommendationsinto eventual collections.
                       Using the more recent data, IRS refined its approach by developing sepa-
                       rate assessmentand collection rates for audits of individuals, small cor-
                       porations (those with assetsof lessthan $250 million), and large
                       corporations (those with assetsof $260 million or more). In the past, IRS
                       used the samerates for all types of audits. IRS also reassessedthe timing
                       of eventual collections and determined that it took much longer for
                       those collections to comeinto the Treasury than IRS had assumedin its
                       previous methodology, which would causethe amount of interest even-
                       tually collected to increase.
                       Tables 1 and 2 show how the assessmentand collection information
                       used by IRS in its revised methodology compared to that used previously.
                       As describedby IRS, the comparison in table 1 shows that eventual
                       assessmentsof tax and penalties are much lower than previously esti-
                       mated and that eventual collections of tax, penalties, and interest are
                       about the sameas was previously estimated, with interest being “a
                       much larger component of total receipts.“” Table 2 shows that the timing
                       of collections is much slower, reflecting, according to IRS, “longer times
                       in the examination, appeals/litigation and collection processes.”

                       %S uses different terms to measure audit yield. “Recommended” is the amount of additional tax and
                       penalties examination staff decide a taxpayer owes after auditing the taxpayer’s return. “Assessed”
                       is the smount of tax and penalties IRS decides is due and payable from the taxpayer, which often
                       differs from the recommended amount because of reductions resulting from the taxpayer’s appeal of
                       the audit findings. “Collected” is the amount IRS receives in payment of the assessedamount,
                       including any penalties and interest.

                       ‘In computing its yield estimate for the fiical year 1991 staffing increase, IRS assumed an annual
                       interest rate of 12 percent. According to IRS officials, they used 12 percent because it was the effec-
                       tive rate when they developed the estimate, and they have no way of knowing what future rates wlll
                       be. To the extent future rates deviate from 12 percent, IRS estimate will be over- or understated. The
                       current rate, which will remain ln effect through September 30,1990, is 11 percent.

                       Page 6                                                        GAO/GGD99-119       Tax Administration
Table 1: Comoarlron of Aa8errment and Collection Rates Used In IRS’ Past and Revised Estimatina Methoddoaieb
As a percent of recommended tax and Denalties
                                                                                          Revised method
                                                                                                    .Small                          Large
                                                          Past method      Individuals       corporatione                    corporations
Recommended tax and penalties                                     100.0           100.0               100.0                            100.0
Assessed tax and penalties                                         71,2            67.2                56.9                             44.5
Assessed tax, penalties, and interest                             108.0           100.8               104.5                             97.3
Collected tax, penalties, and interest                           102.4            103.2              114.2                             106.3
                                           Source: IRS’ November 1989 research report (Publication 1501)

Table 2; Comparison of Data on the Timing of Collectlonr Used in IRS’ Past and Revised Estimating Methodologies
As a percent of recommended tax and Denalties
                                                                                             Revised method
                                                                                                       Small                         Large
                                                            Past method       Individuals       corporations                  corporations
Year 1 (year of staffing increase)                                    76.5             3.9                 0.0                           0.0
Year 2                                                                11.9            22.0                 7.2                           0.0
Year 3                                                                 4.3            28.5                35.1                          18.3
Gar 4                                                                  3.2            18.5                24.1                          25.7
iGar 5-k                                                               6.5            30.3                47.8                          62.3
Collected tax, oenalties. and interest                              102.4            103.2               114.2                         106.3
                                           Source: IRS’ November 1989 research report (Publication 1501)

                                           After assessingthe documentation behind IRS’ revised methodology, it
                                           was clear to us that IRS had put forth considerableeffort to develop
                                           more current information on the rates at which recommendationsget
                                           assessedand assessmentsget collected, and the timing of those assess-
                                           ments and collections. IRS appearedto use the best data available in
                                           revising its methodology, but those data leave much to be desired due to
                                           limitations with IRS’ managementinformation systems,More specifi-
                                           cally, those systemsdo not provide data that are compatible across
                                           functional lines, thus preventing IRS from tracking audit casesthrough
                                           examination, appeals,and collection. Casetracking would provide pre-
                                           cise information on assessmentaJpdcollection rates and the length of
                                           time that transpires between examination and collection. To develop
                                           that information for its revised methodology, IRS relied on whatever rel-
                                           evant data it could obtain from various information systemsand the col-
                                           lective judgments of employeesit consideredknowledgeablein their
                                           respective operations.

                                           Page 7                                                      GAO/GGMMl-I   19 Tax Adminbtration
                        As discussedmore fully in our June 1990 report to the Committee’s
                        Ranking Minority Member, IRS is developing a system that is intended to
                        track the results of enforcement casesthrough all the relevant functions
                        from start to final resolution.7As noted in IRS’ November 1989 research
                        report, it will probably be several years before sufficient data are avail-
                        able from that system for IRS to incorporate into its estimating

IRS’ Expectations for   Using its revised methodology (with someadjustment, as discussed
                        later) IRS estimated that the examination staffing increaseproposed in
the Proposed Fiscal     the fiscal year 1991 budget would result in (1) decreasedaudit revenues
Year 1991 Staffing      of $18.2 million that year, (2) a modest increaseof $2.4 million in fiscal
                        year 1992, and (3) sizeableincreasesranging from $300 million to $740
Increase Are Based on   million in the next 13 years. IRS’ determination that appreciable reve-
SomeQuestionable        nues will not be realized in the short term from the increasedstaff is a
Assumptions             much more realistic assessmentthan would have been produced by IRS’
                        previous methodology. IRS’ ability to achieve even those reduced expec-
                        tations, however, may be adversely affected by the fact that (1) new
                        staff will not be coming on board as soon as assumed,(2) training pro-
                        gram revisions that IRS assumedwould limit the amount of unrealized
                        audit revenuesand increasethe productivity of new staff may not be in
                        place when new staff start coming on board, and (3) IRS’assumption as
                        to the kinds of returns it will be able to audit becauseof the additional
                        examination staff years appears to be inconsistent with information we
                        received from IRS’field offices.

When Will New Revenue   IRSdevelopedits estimate of the revenuesto be generatedby the fiscal
Agents Comeon Board?    year 1991 staffing increasein December1989. At that time, IRS assumed
                        that the 760 new revenue agent positions would be filled at or near the
                        beginning of the fiscal year. IRS officials said they basedthis assumption
                        on the expectation that the budget/appropriations processfor fiscal
                        year 1991 would be completed soon enough for IRS to bring the new staff
                        on board by October 1. Considering recent history, that assumption
                        seemsto have been overly optimistic. IRS’ appropriations for the last 3
                        fiscal years were enacted into law on December22, 1987; September22,
                        1988; and November 3, 1989, respectively.

                        7Tax Administration: IRS Needs More Reliable Information on Enforcement Revenues (GAO/
                            _QO_86, June 20,lOOO).

                        Page 8                                                   GAO/GGDsO-119     Tax Administration

                           Consideringthe current status of budget deliberations, IRS’ appropria-
                           tion for fiscal year 1991 will probably not be enactedbefore the end of
                           September 1990. By then, it may be too late for IRS to recruit college
                           accounting students who are graduating before December1990.In the
                           more competitive areasof the country, IRS may have to wait for the May
                           1991 graduating class.
                           IRS’experiencewith respect to a revenue agent increaseauthorized for
                           fiscal year 1987, although much larger than the increaseproposed for
                           1991, provides an indication of the difficulties that can confront IRSin
                           trying to bring staff on board to fill newly authorized positions. For
                           fiscal year 1987, IRS was authorized an increaseof about 1,600 revenue
                           agents and was given advancehiring authority in July 1986. Even
                           though IRS was able to start hiring about 3 months before the start of the
                           fiscal year, it was only able to fill about half of the 1,600 authorized
                           positions by January 1987 -6 months after it started hiring actions.
                           A delay in bringing new agentson board in fiscal year 1991 will reduce
                           the amount of additional revenuesIRS estimated would be generatedin
                           the next 6 years. According to IRS officials, for example, if IRShad devel-
                           oped its estimate on the assumption that the 760 positions would be
                           filled by January 1, 1991, instead of October 1, 1990, its estimate of $1.1
                           billion in revenuesthrough fiscal year 1996 would decline to $963.3

When Will Changesto IRS’   As noted earlier, a major reasonwhy little, if any, additional revenue is
Training Program Be        generated during the first couple of years after an increasein examina-
                           tion staff is the fact that potential audit revenuesgo unrealized when
Implemented?               experiencedrevenue agents,who would have been auditing returns, are
                           used instead to train the new staff. When IRS revised its estimating
                           methodology to start recognizingthe impact of those lost audit reve-
                           nues, it becameapparent that the impact was considerable.As discussed
                           in our April 1990 report, IRS concludedthat the training associatedwith
                           an increaseof 1,000 revenue agent staff years would cost about $627
                           million in unrealized revenuesover 10 years. To reduce unrealized audit
                           revenues,IRShas been considering somechangesto its training program.
                           In estimating that the increasedexamination staffing proposed for fiscal
                           year 1991 would lead to about $1.1 billion in additional revenuesover 6
                           years, IRSassumedthat two changesto the training program would be in
                           place at the start of the fiscal year. Those two changesinvolve (1) using
                           somecontract trainers in lieu of revenue agentsand (2) restructuring

                           Page 9                                       GAO/GGD40-119   Tax Administration

the training program in an attempt to give new agentsthe training they
need in a shorter time and thus increasethe time they spend auditing. In
estimating the yield to be derived from the proposed fiscal year 1991
staffing increase,IRS assumedthat implementation of those changes
would save about $93 million in training costsover 6 years,
According to Examination officials, they plan to allow each of IRS’ seven
regions to contract for its own instructors in fiscal year 1991, with the
expectation of moving to a nationwide contract in fiscal year 1992.
Dependingon each region’s ability to find acceptablecontract instruc-
tors, the ratio of revenue agent instructors to contract instructors in any
particular training class could vary from what has been the typical con-
figuration of three agents and no contractors to one agent and two con-
tractors (there will always be at least one agent in each class according
to the officials). The hiring of fewer contract instructors would result in
a lower yield than IRS had estimated in computing the revenuesto be
derived from the fiscal year 1991 staffing increase.At that time, IRS
assumeda classwould always involve one agent and two contract
instructors. As of August 3, 1990, IRS’plans to use contract trainers
seemedseveral months from implementation. An Examination official
said that (1) IRScannot request bids on a contract until it knows what
the training workload is going to be-something it will not know until
its appropriation is enacted-and (2) oncebids are requested,it could be
2 to 3 months before contract award.
New revenue agentsnow receive four phasesof training spread out over
26 months. Each phase comprisesseveral weeks of classroominstruc-
tion followed by another several weeks of on-the-job training. The first
two phasesdeal with tax laws that pertain to individual taxpayers; the
third phase deals with laws pertaining to corporate taxpayers; and the
fourth phase deals with, among other things, partnerships. IRS is consid-
ering several changesto that training structure. According to an Exami-
nation official, IRS is considering (1) reducing the number of training
phasesfrom four to three by combining the last two; (2) reducing class-
room training from a total of 20 weeks to about 16 weeks; (3) reducing
on-the-job training from a total of 41 weeks to 19 weeks; and (4)
expanding the length of time between phase 2 and new phase 3 from 26
weeks to 60 weeks, thus allowing trainees to spend more time auditing
returns. This proposed restructuring goesbeyond what IRS assumedin
preparing its $1.1 billion estimate and thus, if implemented, could cause
IRS’estimate to be understated. IRS officials said that training time can be
reduced without adversely affecting quality by doing such things as

Page 10                                      GAO/GGD-90-119   Tax Admlnbtration

                            eliminating redundancy between phasesand using lessformal settings,
                            like workshops, to convey somematerial that had beencoveredin class.
                            BecauseIRShad not begun awarding training contracts and because
                            restructuring plans have not been finalized, it is clear that the training
                            changeswill not be implemented by the start of the fiscal year, as IRS
                            assumedin developing its estimate. There may be little impact, however,
                            because,as discussedabove,new staff will not be coming on board by
                            the first of the year. We have no basis for determining when these or
                            other changesto the training program might be implemented, whether
                            they will be on-line when newly authorized revenue agentsstart coming
                            on board, and what their effect will be on training costsand audit

How Will the Influx of      Oneconcernwe had with IRS’ previous estimating methodology that
Additional Examination      remains a concernwith the revised methodology is IRS’ assumption that
                            the increasedaudit time provided by the additional staff will be allo-
Staff Affect the Workload   cated amongnumerous tax c1asses,B    including onesin which new rev-
of Experienced Staff?       enue agentsdo not typically work. In developing its estimate of the
                            revenue to be derived from the proposed staffing increasefor fiscal year
                            1991, for example, IRSassumedthat the increaseof 760 revenue agent
                            staff years would translate to about 284 additional staff years that it
                            could devote to auditing returns, after allowing for time that would be
                            charged to training and other nonaudit activities. It then assumedthat
                            about 16 of those additional staff years would be devoted to auditing
                            returns in individual tax classes,with the other 269 devoted to returns
                            in more complex, higher-yield tax classes,such as returns filed by
                            In explaining this allocation, IRS National Office officials said that as
                            new agents-who generally work the less complex, low-yield individual
                            returns-come into IRS,experiencedrevenue agentscan move into more
                            complex, higher-yield tax classes.It is IRS’position, therefore, that it is
                            appropriate to attribute a portion of the yield from those higher-yield
                            classesto the increasedstaff. TheseNational Office officials said that
                            although they believe this movement of experiencedstaff occurs,they
                            do not have empirical data to support the assumption.

                            8A tax class is a groupingof returns on the basis of return type, such as individual or estate; source
                            of income, such as farm or business; and level of income.

                            Page 11                                                        GAO/GGDIH)-119     Tax Administration

                           District office managersand officials we talked with had a different per-
                           ception. Managersof revenue agents and Examination officials in IRS’
                           San Francisco District Office and Examination officials in the Atlanta,
                           Chicago,and St. Louis District Offices said that experiencedrevenue
                           agents do not audit the type of low-complexity returns assignedto first-
                           year trainees. They said that the low-complexity returns are usually
                           examined by tax auditors (IRSemployeeswho audit less complex tax
                           returns than those audited by revenue agents).The managerssaid that
                           were it not for the trainees, those low-complexity returns would not be
                           audited by their groups of experiencedagents.The district officials also
                           said that it takes about 2 years from the beginning of an examination
                           staff expansion before they seean increasein the staff years used in the
                           audits of the more complex, higher-yield tax classes.By this time, the
                           new staff will have beentrained to audit the more complex, higher-yield
                           returns and the experiencedrevenue agentswill have returned to
                           auditing from their training assignments.
                           If IRShad allocated all of the proposed fiscal year 1991 increasein exam-
                           ination staff years to individual tax classesin the first year of the
                           increase,its estimate of $1.1 billion in revenue through fiscal year 1996
                           would have declined to $668 million,

The Reliability of IRS’    Onepositive step being taken by IRS with respect to the various revenue-
                           generating initiatives proposed in its fiscal year 1991 budget is the
Tracking of Fiscal         development of methodologiesby which it can track the results actually
Year 1991 Results Will     achieved under each of those initiatives. Congress’budget deliberations
                           would be enhancedif it had reliable information on the amount of rev-
Depend on the              enue generatedby the various initiatives. IRS could also use the informa-
Validity of Its Starting   tion to verify the reliability of its estimates. The methodology IRS plans
Point                      to use to track the results associatedwith the increasein revenue agents
                           seemsto be fairly straightforward. IRS plans to compare the number of
                           revenue agent staff years and associatedyield that would have been
                           achievedwithout the increase(referred to as the baseline) with the staff
                           years and yield actually achieved-the difference being attributed to
                           the increasedstaff. IRS used the samebasic methodology when it esti-
                           mated the yield that had beengenerated in fiscal year 1987 as a result
                           of the increasedexamination staff authorized for that year. We raised
                           several concernswith IRS’ fiscal year 1987 estimate in our August 1988
                           report, including the fact that IRS had, in our opinion, overestimated
                           yield by reducing the baseline against which results were compared.

                           Page 12                                     GAO/GGD-99-119 Tax Administration
Before the start of fiscal year 1987,IRS expectedthat the authorized
staffing increasefor that year would raise its revenue agent staffing
level by 1,618 staff years-from a baseof 14,227at the beginning of the
year to 16,846by year’s end. In fact, however, IRS’ revenue agent
staffing level rose to only 14,939staff years by the end of fiscal year
1987-an increaseof 712 years, or 906 years lessthan expected.If IRS
had held its staffing baseconstant at 14,227staff years, its computation
of additional yield would have beenpredicated on an increaseof 712
revenue agent staff years. Instead, as discussedmore fully in our
August 1988 report, IRS decidedthat about 90 percent of the 906 unreal-
ized staff years applied to the base.As a result, IRS (1) reduced its base
staffing level from 14,227to 13,414;(2) determined that the 1,525 staff
year difference between the adjusted base(13,414) and the end-of-year
staffing level (14,939) representedthe increasedstaff authorized by
Congress;and (3) used that increaseof 1,626years to compute the rev-
enue effects of the authorized staffing increase.
We recognizethat events, like the need to absorb a pay increaseor a
managementdecision to divert staff from Examination to handle a
staffing shortage in Taxpayer Service,can lead to reductions in Exami-
nation’s basestaffing level during the fiscal year. Someof these reduc-
tions may be necessarybecauseof factors, such as a hiring freeze caused
by the need to absorb a pay increase,that are beyond IRS’ control. Other
reductions, however, represent discretionary managementdecisionsto
changestaff allocations.

It is important that IRSdiscloseto Congressthe extent of and reasonsfor
any erosion of the base.It doeslittle good, in our opinion, for Congress
to approve additional staff with the intent of generating additional rev-
enue if that benefit is eroded by reductions to the base.For Congressto
properly assessthe effects of staffing increases,therefore, it needsto
know how and why the basehas changed.
Before the beginning of fiscal year 1991, IRS will be computing the base-
line against which it will track the results of a fiscal year 1991 staffing
increase.Until it doesand until it has specified under what circum-
stances,if at all, it will adjust the baseline and how, if at all, it will
reflect the impact of such adjustments in its tracking results, we have no
basis for assessingthe methodology’s validity or for determining
whether the tracking results will be more meaningful than the results
reported for 1987.

Page 18                                      GAO/GGMO-119 Tax Administration

                      IRS’revised methodology for estimating the yield from an increasein
Conclusions           examination staff is a significant improvement over the methodology IRS
                      used in the past. It addressesmost of the concernsraised in our earlier
                      assessmentof the past methodology and produces a much more realistic
                      projection of what the government can expect as a result of the staffing
                      increase.In applying that methodology to the proposed staffing increase
                      for fiscal year 1991, however, IRS made assumptionsabout the hiring
                      and training of new agentsthat have proven to be overly optimistic.
                      Also, IRS needsto develop information to support its assumption that
                      hiring new agentsallows IRS to assignmore experiencedagentsto more
                      complex, high-yield cases.If empirical data do not bear this out, IRS’esti-
                      mates should be revised to reflect only that revenue generatedfrom
                      audits in the tax classestypically worked by new hires.
                      IRSplans to monitor the results of the fiscal year 1991 staffing increase,
                      assumingit is authorized. Reliable information on results will be valu-
                      able not only to Congressin deliberating future proposed increasesbut
                      also to IRS as a check on the validity of its estimating methodology. IRS
                      will be establishing a baseline against which results will be monitored.
                      Until then, we cannot begin to determine whether the monitoring will
                      produce reliable information.

                      To further improve its methodology for estimating the revenue to be
Recommendationto      derived from an increasein examination staff, IRSshould develop empir-
the Cornmissionerof   ical data to show whether the influx of new examination staff allows
Internal Revenue      more experiencedagentsto work higher-yield casesand in what time

                      The Commissionerof Internal Revenuecommentedon a draft of this
Agency Comments and   report by letter dated August 29, 1990, (seeapp. I). The Commissioner
Our Evaluation        agreedthat there are someareasin which IRS neededto fine tune its
                      processfor estimating examination revenues.He specifically agreedthat
                      IRS neededto study the extent to which the influx of new staff allows
                      more experiencedstaff to do more complex, higher-yielding audits and
                      the timing of any such impact. The Commissionersaid that such fine
                      tuning would not materially changethe revenue estimate in the Presi-
                      dent’s budget for fiscal year 1991, He noted that the revenue estimate in
                      the President’s budget still appears achievable if Congressacts quickly
                      to approve IRS’ appropriation and to limit the impact of sequestration.
                      The Commissionerwent on to say that greater certainty in the budget
                      processwould enable IRS to make commitments to June graduates, bring

                      Page 14                                      GAOp3GDfJO.119   Tax Admhiatmtion

them on board timely, and scheduletraining in a way that maximizes
We agreethat the soonerIRS’appropriation is enacted and the less IRSis
affected by sequestration,the better IRS’ chancesof generating revenues.
Even then, however, we believe that IRScould have a much harder time
meeting its revenue estimate than indicated by the Commissioner.As
noted earlier, for example, if empirical data do not support IRS’ assump-
tion that the influx of new staff enablesexperiencedstaff to work
higher-yield cases,IRS’ S-year estimate of $1.1 billion in revenuescould
be reduced to as low as $668 million.
In discussingthe portion of our report dealing with changesto the rev-
enue agent training program, the Commissionernoted that cumbersome
procurement rules and budget uncertainties affected IRS’ability to move
ahead quickly with planned contract training. He also indicated that we
were concernedabout the fact that IRS was changing its original con-
tracting plan and might be using more revenue agents as instructors
than was originally assumed.We have no basis for judging the relative
merits of IRS’ original and revised plans. Our only purpose in mentioning
the changewas to alert the reader that one of the assumptionson which
the revenue estimate in the President’sbudget was basedhas been
The Commissioneralso agreedthat the reliability of IRS’ monitoring of
the revenuesgeneratedby the staffing increasedependsto a great
extent on the validity of the baseline.He emphasized,however, that the
baseline(amount of revenue that would have beencollected absentthe
staffing increase)can only be estimated. We agreethat the baseline can
only be estimated. Our concernis with the estimate’s validity and with
subsequentchangesto the baseline.

As arranged with the Committee, unlessyou publicly announceits con-
tents earlier, we plan no further distribution of this report until 2 days
from the date of issuance.At that time, we will send copiesof this
report to the Secretary of the Treasury, the Commissionerof Internal
Revenue,and other interested parties.

Page 16                                      GAO/GGDSMlS   Tax Adminbtration

The major contributors to this report are listed in appendix II, Please
contact me on 2756407 if you or your staff have any questions con-
cerning the report.
Sincerely yours,

Jennie S. Stathis
Director, Tax Policy and
  Administration Issues

Page 16                                     GAO/GGD4Mb119   Tax Administration
Page 17   GAO/GGD-90419   Tax Administration

Appendix I                                                                                              20
Comments From the       GAO Comment                                                                     24

Internal Revenue
Appendix II                                                                                             26
Major Contributors to
This Report
Tables                  Table 1: Comparisonof Assessmentand Collection Rates                             7
                            Used in IRS’ Past and RevisedEstimating
                        Table 2: Comparison of Data on the Timing of Collections                         7
                            Used in IRS’ Past and RevisedEstimating


                        IRS            Internal RevenueService

                        Page 18                                   GAO/GGDIw)-119   Tax Administration
Page 19   GAO/GGD-90-119   Tax Administration
Appendix I

CommentsFrom the ZntermalRevenueService”

Note: GAO comments
supplementing those in the
report text appear at the
end of this appendix.                                              DEPARTMENT             OF THE TREASURY
                                                                       INTERNAL         REVENUE          SERVICE
                                                                            WASHINGTON,           D.C.    20224

                             Mr. Richard        L. Fogel
                             Assistant       Comptroller                   General
                             United     States     General                 Accounting          Office
                             Washington,       DC 20548

                             Dear      Mr.     Fogel:

                                       We have          reviewed          your     draft       report         concerning             IRS’s      revised
                             methodology              for     estimating           examination             yield.           We were         pleased         to
                             see that           GAO agrees            that     we have        made substantial                     improvements              in
                             our     estimating             methodology            and that          we are making                 progress         in
                             assuring           that      we are held            accountable             for      our      performance.               We
                             also      agree        that      there       are some areas               in which            we need to fine
                             tune      our estimating                 process,         but       believe          that      our estimates
                             reflected            in the        President’s            budget        for      Fiscal        Year       1991 would
                             not materially                 change        because        of these          findings.              Given       this      and
                             other       improvements               in restructuring                 our      training           programs,          our
                             original           estimates           appear       to be achievable                    if Congress            acts
                             quickly          to approve            our     appropriations               and to limit                the    impact       of
                             sequestration.                   It should          also      be noted           that       any substantial
                             delay       will       postpone          revenue        anticipated              for      the     second       and
                             subsequent             years.

                                       Revenue          estimates        aside,        we should       not      lose      sight       of the
                             fact      that      what       is most      important           is a long-term            commitment           to
                              increase        IRS resources.                 Also,      the     Examination          initiative
                             discussed           in this         report      is only        one part        of our overall               FY 1991
                             revenue        initiative.               4ny discussion             of enforcement               revenue
                             estimating            must       be put     into      this     context       so Congress            will     have
                             the benefit             of your        advice      when considering                the    important          issues
                             of raising            revenue         and funding          this     agency       to collect           that

                             --              1 Entry         on    Gut,/      Date      Unrealistic                for   New Agents

                                       IRS’      revenue        estimate         assumes            that      750 new revenue                agent
                             positions          would         be filled        at or near               the    beginning            of the      fiscal
                             year.        Although            we did      not    base         the estimate            on the assumption
                             that      advance         hiring       would      be authorized,                  we did         assume that           IRS’
                             appropriations                would      provide         funding           for    this      initiative           as of
                             October         1, 199C.           Because        the       initiative            assumed          nat:onwide
                             hiring        :n our        63 districts            and because                we can take             many
                             recruitment             steps      in anticipation                   of actual         hiring,           it was
                             reasonable            to assume          that     we could             hire      750 new revenue                agents
                             very      quickly         after      approval         by Congress.                  We continue             to believe
                             we could          do that.           Although         first-year               revenue        effects         may have
                             to be adjusted                 to reflect         any delays,                second      and later            years
                             would       not     be greatly           affected          if Congress              acts      quickly         to approve
                             our     appropriations.

                                        Page 20                                                                          GAO/GGD-90419            Tax Administration

               Ckmmemta From the Internal                Revenue Ssrvice


    Mr.     Richard         L.     Fogel

               If   IRS could        rely      on a process           that   delivered            a budget
    timely        without        the   threat       of sequestration,              we believe             we would
    be able         to deliver         the     revenue       projected       for     the     initiative.
    Greater         certainty        in the        budget      process      would      enable         us to make
    commitments             to June       graduates,         bring      them on board             timely,      and
    schedule          training       in a way that             maximizes       productivity.                The
    current         “stop       and go” budgeting              process      has a continued
    detrimental              impact    on our ability              to hire,      train       and retain
    quality        employees.             Possible        sequestration          further          jeopardizes
    planned         new initiatives              as well       as current        revenue          processing

    Chanqes         to    IRS’       Examination             Trainina          Proqram

            GAO also           cites        our    planned            improvements               in training               as
    another       factor          affecting          potential              revenue         from       the      initiative.
    As noted,         IRS initiated                these          improvements               in part         to maximize
    early     revenue          producing           activities               of trainees              and to free              up our
    more experienced                  revenue        agents           for     more       productive            work.          GAO
    agrees      that       these        improvements                will,       when fully             effected,             go a
    long    way toward              optimizing             potential            revenue          increases             in the
    early     years        of a revenue               initiative,               but      does      not     believe           we will
    be ready        to contract               out    training             by the         beginning           of the fiscal
    year    as our         estimates            assumed.              While       we agree           that      there         will     be
    a delay       of as much as two to three                                months,         we firmly             believe         that
    we will       be ready            with      contract            trainers          as soon          as new hires               are
    on board.            In addition,              we believe               that      our      planned         restructuring
    of training            programs           will      increase            revenue         producing             activities
    from    the     initiatives               even      more than             originally             projected.

              GAO expressed                concern         that      our   original           plan      for     contracting
    out     training         was being              changed        to use two rather                  than      one revenue
    agent       per training               class.          If    true,     this       of course           would      increase
    opportunity             costs        by taking            more agents           from      revenue         producing
    work.         Although          analysis           has indicated             that       we may use two
    revenue         agents        rather          than     one per class               in some locations
    because         of the        difficulty             of hiring         outside          instructors,             we
    believe         this      change         will      only       impact     a few of the                smaller
    locations            and will          not      have      a substantial              impact       on our revenue
    estimates.              Further,           we believe            the   restructuring                 improvements
    noted        above      may more           than      offset        any potential              effect        of this

             Cumbersome      procurement           rules,    coupled       with    budget
    uncertainties,          do impact        on our ability          to move ahead         quickly
    with     planned    contract       training.          However,      if     Congress    were    to
    approve        our appropriations            request     and remove         the threat      of
    sequestration,          we believe         we would      be able       to hire      and begin     to

               Page 21                                                                          GAO/GGD9@119Tax Adminbtralion

                             Appendix I
                             Comments From the Internal           Revenue Service


               Mr.     Richard         L.     Fogel

               train      new employees                for  the    initiative              within        two to three
               months.         The ability               to plan     in advance             for      the    long  term          would
               greatly       aid  us in             knowing     when we could                hire      and contract             for

               Migration          of        Work

                            Of continuing            concern         to GAO was our                 assumption           that
                increased           audit      time     provided           by the         additional          staff        would       be
               allocated            among numerous               tax     classes,          with       new hires          freeing         up
               experienced              agents       to do more complex                    and higher            yielding
               examinations               beginning          in the        first       year       (known      as “migration”).
See comment,   While         we do not         believe         that      the      period        1985 - 1989 gives                  a
                realistic           picture        of what         could         be expected            from     any new
                initiative,             we agree        to study           the      impact        and timing           of the
               migration            when new hires               are     brought          on board.           As noted           by GAO,
               we have already                 taken       into      account,           in our        estimates,           the     reduced
               opportunity              costs      created         when we use experienced                         revenue         agents
               to train           new employees.                 For this           reason,         we believe           that      the
               effects           noted      by GAO for           the     periods          1985 - 1989 reflect                    factors
               that         have    already        been taken            into      account          in our estimates.                    We
               do,       however,         welcome       GAO’s        assistance             in conducting              further

See comment.             GAO also       compares     planned        versus       actual        staffing          for     the
               same period.             We believe      that      comparing          planned          versus       actual
               staffing       neither       proves     nor disproves             migration            in the first
               year.       Exceeding        or underrealizing              planned        staffing           can be caused
               by various         other     factors      including,          but     not     limited         to,     delays  in
               training,       hiring       freezes,       and difficulties                in recruitment.

               Reliability             of     the     Baseline

                         IRS agrees          with       GAO that          the     reliability            of estimates          of
               revenue         generated          by any initiative                   depends        to a great         extent          on
               the    validity         of the          baseline,           i.e.,      we must        accurately         estimate
               the    enforcement            revenue           that      would      have        been   collected         in the
               absence         of the      initiative.                We must          reemphasize,            however,      that
               this     can only         be estimated;                it     cannot         be “tracked.”              The
               Assistant          Commissioner               (Finance)/Controller                    and the Assistant
               Commissioner            (Examination)                are working               to formalize         the method

                             Page 22                                                                    GAO/GGD-99-119          Tax Administration

                  Appendix I
                  Comments From the Intemal                 Revenue Service


        Mr.    Richard           L.      Fogel

        for      estimating              what     would     have     happened      in     the     absence      of   the

                 Thank         you       for     the   opportunity         to   provide         our   comments        on
        this     report.

                 Best         regards.

                  Page 22                                                                       GAO/GGB9&119        Tax Administration
              Appendix I
              Commenta From the Internal   Revenue Service

              The following is GAO’S comment on the Internal RevenueService’sletter
              dated August 29, 1990.

              1. Thesereferencesare to data that were in the draft report but are not
GAO Comment   in the final report.

              Page 24                                        GAO/GGDW119   Tax Administration
Appendix II

Major Contributors to This Report

                                    David J. Attianese, Assistant Director, Tax Policy and Administration
General Government                   Issues
Division, Washington,
                                    RoyceL. Baker, Evaluator-in-Charge
Kmsas         City   Re@ona1        Shirley A. Franklin ,Evaluator

(268460)                            Page26                                      GAo/GGIMO419
._   ..I_   ____._,.,_._   “.__.   I_--..-_.-   -..-   ---.“.--...   ._   “_   “-l,_    .   .   ..__   -   ___   _.   ._..”        .,_   __l”_.“_.-.-_-                  __-

                                                                                       Orclc~ring                                        lrlf’ornlal              iorl

                                                                                       i I.S. (kttvritl   Acwttttt ittg Of’fiw
                                                                                       I’. 0. IhbX co 15
                                                                                       (;;lit h(brst)urg,   MI) 20877

                                                                                       Ortt(~rs                               Itlily                      ;IlsO tW t>t;lccVt by calling   (202)   275-624   I.