IRS Audits: Weaknesses in Selecting and Conducting Correspondence Audits
(Chapter Report, 03/31/99, GAO/GGD-99-48).

Pursuant to a congressional request, GAO reviewed the Internal Revenue
Service's (IRS) program to audit income tax returns through
correspondence, focusing on: (1) the number, results, and duration of
correspondence audits as well as the characteristics of the audited
returns; and (2) processes and requirements that IRS has had for years
to govern correspondence audits.

GAO noted that: (1) several weaknesses were found in IRS' correspondence
audit processes; (2) these weaknesses, individually or in combination,
can erode the integrity of the correspondence audit processes, which are
designed to help ensure that taxpayers pay the correct tax amounts and
are treated properly; (3) during fiscal years 1992-1997, the annual
number and results of correspondence audits conducted by IRS varied
considerably; (4) the number ranged annually from just over 200,000 to
about 1.1 million audits; (5) the rate at which IRS auditors closed
audits without recommending additional taxes ranged from 13 percent to
46 percent; (6) when they did recommend additional taxes, the average
amounts ranged from $1,300 to $2,800; (7) the rate at which taxpayers
did not respond to these recommended additional taxes after being
requested to do so by IRS ranged from 29 percent to 63 percent; (8)
these variations resulted, in part, from an increase in the number of
correspondence audits of returns claiming an earned income credit (EIC);
(9) for the traditional correspondence audits closed in fiscal year
1996, the time between the filing of a return and the start of the
correspondence audit averaged 10 months; (10) it then took 11 more
months before IRS assessed any taxes that were recommended during the
audits; (11) as for the characteristics of these 1996 returns, an
estimated 75 percent had reported adjusted gross incomes of less than
$15,000; (12) in part, this percentage reflects the correspondence
audit's focus on simple tax issues and EIC; (13) IRS had weaknesses in
implementing the correspondence audit requirements for four processes;
(14) not all of the traditional correspondence audits closed in 1996
were manually reviewed (or classified) to identify all issues for audit,
as required by IRS; (15) support for recommended audit findings was not
adequately documented in the audit workpaper files, as required, for
about one-third of the audits; (16) the taxpayer documentation that was
required to justify EIC claims varied from service center to service
center; (17) GAO found weaknesses in the reviews that IRS did on a
sample of closed audits to measure their quality; and (18) in addition
to the weaknesses in implementing the requirements, IRS allowed service
centers to exclude certain types of audits that did not have all
required documentation from being measured against the audit standard on
workpaper documentation.

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

 REPORTNUM:  GGD-99-48
     TITLE:  IRS Audits: Weaknesses in Selecting and Conducting 
             Correspondence Audits
      DATE:  03/31/99
   SUBJECT:  Tax return audits
             Tax administration
             Income taxes
             Tax nonpayment
             Federal taxes
             Government collections
             Internal controls
             Auditing standards
IDENTIFIER:  IRS Audit Information Management System
             IRS Correspondence Examination Quality Management System
             EIC
             Earned Income Credit
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  This text was extracted from a PDF file.        **
** Delineations within the text indicating chapter titles,      **
** headings, and bullets have not been preserved, and in some   **
** cases heading text has been incorrectly merged into          **
** body text in the adjacent column.  Graphic images have       **
** not been reproduced, but figure captions are included.       **
** Tables are included, but column deliniations have not been   **
** preserved.                                                   **
**                                                              **
** 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.                         **
******************************************************************
Microsoft Word - 14z801!.PBF IRS AUDITS Weaknesses in Selecting
and Conducting Correspondence Audits

United States General Accounting Office

GAO Report to the Chairman, Subcommittee on Oversight, Committee
on Ways and

Means, House of Representatives


March 1999 

GAO/GGD-99-48

March 1999   GAO/GGD-99-48

United States General Accounting Office Washington, D. C. 20548

Page 1 GAO/GGD-99-48 IRS' Correspondence Audits

GAO

General Government Division

B-276687 March 31, 1999 The Honorable Amo Houghton Chairman,
Subcommittee on Oversight Committee on Ways and Means House of
Representatives

Dear Mr. Chairman: This report responds to your request that we
review the Internal Revenue Service's (IRS) program to audit
income tax returns through correspondence. We provide information
on the number, results, and duration of correspondence audits as
well as the characteristics of the audited returns and examine the
processes and requirements that IRS has put in place to govern
correspondence audits. This report also includes our
recommendations on improving the correspondence audit program.

We are sending copies of this report to Representative William J.
Coyne, Ranking Minority Member, Subcommittee on Oversight, House
Committee on Ways and Means; Senator William V. Roth, Jr.,
Chairman, and Senator Daniel Patrick Moynihan, Ranking Minority
Member, Senate Committee on Finance; various other congressional
committees; The Honorable Robert E. Rubin, Secretary of the
Treasury; The Honorable Charles O. Rossotti, Commissioner of
Internal Revenue; The Honorable Jacob Lew, Director, Office of
Management and Budget; and other interested parties. Copies will
also be made available to others upon request.

This report was prepared under the direction of Thomas D. Short,
Assistant Director. Other major contributors are listed in
appendix VI. If you have any questions about this report, please
call me or Mr. Short on (202) 512- 9110.

Sincerely yours, James R. White Director, Tax Policy

and Administration Issues

Executive Summary

Page 2 GAO/GGD-99-48 IRS' Correspondence Audits

Between 1992 and 1997, the Internal Revenue Service (IRS) audited
between 1 and 2 million individual income tax returns per year.
Audits are one of the enforcement tools used by IRS in its efforts
to close the tax gap the amount of taxes owed but not voluntarily
paid. IRS' most recent estimate put the tax gap for individuals at
over $125 billion for 1992. IRS also estimates that it eventually
collects about one- quarter of the gross tax gap through its
enforcement efforts.

Since 1996, the most common type of IRS audit of individual
taxpayers has been the correspondence audit conducted through IRS'
10 service centers. As the name implies, these audits are
conducted through the mail, with IRS typically asking taxpayers
for more support regarding one or two simple issues on a tax
return. Audits of more complex tax issues are done at IRS' 33
district offices.

In recent years, Congress has been concerned about IRS' use of its
enforcement tools, particularly about whether the tools are used
fairly and in a manner that does not impose excessive burdens on
taxpayers. Concerns such as these led, in part, to the passage of
the IRS Restructuring and Reform Act of 1998. The act included
provisions affecting audits.

In keeping with the Committee's oversight responsibilities for IRS
enforcement activities, the Chairman of the House Committee on
Ways and Means, Subcommittee on Oversight, asked GAO to review
correspondence audits and IRS' processes for conducting these
audits. Accordingly, this report (1) provides information on the
number, results, and duration of correspondence audits as well as
the characteristics of the audited returns and (2) examines the
processes and requirements that IRS has had for years to govern
correspondence audits. GAO's analysis is based on computerized IRS
data on audits closed during fiscal years 199297, a study
population of traditional correspondence audits closed in 1996
from which we drew a sample, and surveys from compliance officials
at IRS' 10 service centers.

GAO found several weaknesses in IRS' correspondence audit
processes. These weaknesses, individually or in combination, can
erode the integrity of the correspondence audit processes, which
are designed to help ensure that taxpayers pay the correct tax
amounts and are treated properly.

During fiscal years 1992- 97, the annual number and results of
correspondence audits conducted by IRS varied considerably. The
number ranged annually from just over 200,000 to about 1.1 million
audits. The rate at which IRS auditors closed audits without
recommending additional Purpose

Results in Brief

Executive Summary Page 3 GAO/GGD-99-48 IRS' Correspondence Audits

taxes (the no- change rate) ranged from 13 percent to 46 percent.
When they did recommend additional taxes, the average amounts
ranged from $1,300 to $2,800. The rate at which taxpayers did not
respond to these recommended additional taxes after being
requested to do so by IRS (the default rate) ranged from 29
percent to 63 percent. These variations resulted, in part, from an
increase in the number of correspondence audits of returns
claiming an earned income credit (EIC).

For the traditional correspondence audits closed in fiscal year
1996, the time between the filing of a return and the start of the
correspondence audit averaged 10 months. It then took 11 more
months before IRS assessed any taxes that were recommended during
the audits. As for the characteristics of these 1996 returns, an
estimated 75 percent had reported adjusted gross incomes of less
than $15,000. In part, this percentage reflects the correspondence
audit's focus on simple tax issues and EIC.

IRS had weaknesses in implementing the correspondence audit
requirements for four processes. First, not all of the traditional
correspondence audits closed in 1996 were manually reviewed (or
classified) to identify all issues for audit, as required by IRS.
Further, the classification of returns that included complex
business and investment schedules was not always done by qualified
personnel in district offices, as required. Second, support for
recommended audit findings was not adequately documented in the
audit workpaper files, as required, for about one- third of the
audits. Third, the taxpayer documentation that was required to
justify EIC claims varied from service center to service center.
Fourth, GAO found weaknesses in the reviews that IRS did on a
sample of closed audits to measure their quality. For example,
some service centers excluded certain types of correspondence
audits that are required to be included in the sample.

In addition to the weaknesses in implementing the requirements,
IRS allowed service centers to exclude certain types of audits
that did not have all required documentation from being measured
against the audit standard on workpaper documentation. Further,
more than 50 percent of the taxpayers who were audited by
correspondence did not respond to IRS' letters, and IRS has not
analyzed why. GAO is making a number of recommendations to deal
with these and other weaknesses in the correspondence audit
process.

Each year, IRS checks all individual income tax returns filed for
compliance with the tax laws using a variety of tools. These
compliance checks include reviews of some required information,
such as signatures Background

Executive Summary Page 4 GAO/GGD-99-48 IRS' Correspondence Audits

and Social Security numbers, at the time a return is filed;
computerized corrections of what IRS labels math errors;
computerized matching of information, such as income, with the
corresponding information reported to IRS by third parties; and
audits.

Auditors seek additional support from taxpayers for questionable
items on tax returns which IRS calls tax issues. IRS may raise tax
issues regarding any of the items on a return, such as exemptions,
income, deductions, and credits.

IRS does two types of audits of returns filed by individual
taxpayers: correspondence and district office audits.
Correspondence audits deal with simpler issues through
correspondence sent from one of IRS' service centers. District
office audits deal with more complex issues, such as those on
business or investment schedules that are attached to tax returns,
usually through face- to- face meetings in an IRS office or a
taxpayer's place of business. Correspondence audits are conducted
by tax examiners trained in simple tax issues, while district
office audits are conducted by revenue agents and tax auditors
trained in more complex tax issues.

IRS has established audit quality standards and requirements for
selecting, conducting, and reviewing the quality of correspondence
audits. The standards require actions such as considering all
significant issues for audit and adequately documenting audit
recommendations. During audit selection, tax returns should be
reviewed (classified) by IRS personnel qualified to identify all
questionable issues for audit, particularly complex issues on
business and investment schedules. Further, classification should
determine whether the issues can be addressed through
correspondence audits or through district office audits. In
conducting audits, IRS also requires tax examiners to adequately
support their audit recommendations and to document that support
and audit steps in the workpaper files. When auditing claims for
tax credits, such as EIC, tax examiners are required to
substantiate these claims using third- party verification, such as
school records. In reviewing audits, IRS has established two
levels of review. The first is to be carried out by the tax
examiners' supervisors, who are to review some examiners' work
while the audit is in progress. The second review is to be
performed on a sample of closed audits by quality assurance staff
within each service center to measure adherence to the audit
standards.

To start the correspondence audit process, IRS is to send a notice
to the taxpayer that the return has been selected for audit. This
notice is to request certain information or documents needed to
support the issues

Executive Summary Page 5 GAO/GGD-99-48 IRS' Correspondence Audits

identified for audit through classification. If the taxpayer
responds with such support, the tax examiner is to recommend
closing the audit with no changes to the reported tax liability
(no- change audit) and notify the taxpayer through a letter. If
the taxpayer does not provide support, the tax examiner is to
recommend changes to the tax liability and similarly notify the
taxpayer. If the taxpayer does not respond to the recommended tax
changes within 30 days, the tax examiner is to notify the taxpayer
that the recommended tax will be assessed. If the taxpayer does
not respond to this tax assessment notice within 90 days, the tax
examiner is to close the audit as a default by the taxpayer,
meaning the tax assessment becomes final and the taxpayer is
liable for the additional taxes.

During fiscal years 1992- 97, the annual number of correspondence
audits conducted by IRS varied considerably from year to year. The
fewest number of audits 232,000 were closed in 1993. Between 1994
and 1995, the number of correspondence audits tripled from about
353,000 to about 1.1 million but then decreased to 758,000 between
1996 and 1997. The increase between 1994 and 1995 was due, in
large part, to increased emphasis on auditing returns claiming EIC
in response to congressional concerns about EIC noncompliance. The
decrease in 1997 was due to moving many of the EIC returns from
the correspondence audit program to the math- error program.

During fiscal years 1992- 97, correspondence audit results such as
the nochange rates, amount of taxes recommended per audited
return, and default rates also varied considerably from year to
year. For example, the no- change rate increased through 1995 to a
high of 46 percent and then decreased by 1997 to a low of 13
percent. In contrast, the amount of taxes recommended per audited
return declined between 1992 and 1995 to a low of $1,282 and then
increased in 1997 to a high of $2,835. Similarly, the default rate
decreased by 1995 to a low of 29 percent and then increased in
1997 to a high of 63 percent.

For the correspondence audits closed in fiscal year 1996, GAO
found an overall average of 21 months elapsed from the time the
returns were filed until IRS assessed any taxes recommended during
the audit. The time between the filing of a return and the start
of the correspondence audit averaged 10 months. It then took an
average of 11 more months before IRS assessed any taxes
recommended during the audits. Audits of returns for Principal
Findings

IRS' Use of Correspondence Audit

Executive Summary Page 6 GAO/GGD-99-48 IRS' Correspondence Audits

which the taxpayers never responded to IRS' letters usually took
longer to do than other returns.

GAO's review of characteristics of the audited returns during
fiscal year 1996 showed that an estimated 75 percent of the
returns had reported adjusted gross incomes of less than $25,000.
In part, this percentage reflects the correspondence audit's focus
on simple tax issues and EIC. Specifically, these correspondence
audits dealt primarily with a few types of issues reported on
individual tax returns, such as EIC and selfemployment taxes, as
well as taxpayers who did not appear to have filed the required
tax returns. Further, an estimated 20 percent of the audited
returns included complex schedules, such as business and
investment schedules. Such schedules included those filed by
individuals to report business net profits or losses generated
through self- employment and to report capital gains or losses.

IRS had weaknesses in its processes for implementing
correspondence audit requirements as well as for seeking
taxpayers' responses. Specifically, IRS did not consistently
implement all requirements in its processes for (1) classifying
returns to identify issues for audit, (2) documenting the audit
steps and support for audit recommendations, (3) justifying EIC
claims, and (4) reviewing the quality of audits. Also, IRS did not
know why over half the taxpayers audited by correspondence did not
respond during the audit. Although GAO did not measure the effects
of these weaknesses, each weakness can, individually or in
combination, erode the integrity of the correspondence audit
processes, which are designed to help ensure that taxpayers pay
the correct tax amounts and are treated properly.

First, service centers did not follow IRS' requirements for
classifying returns for audits and referring returns with complex
business or investment schedules to districts. IRS' manual
requires tax returns identified for audit to be classified so that
all questionable issues are audited. However, excluding nonfilers,
an estimated 69 percent of the workpapers in the audits closed in
1996 had no evidence of classification. When classification did
occur, tax examiners classified most returns, some involving
complex schedules for which they were not trained. IRS officials
at both the National Office and the service centers said that
classification was not needed for all correspondence audits but
could be needed for returns having complex schedules.

GAO found evidence that less than an estimated 1 percent of the
returns with complex schedules were referred, as required, to
revenue agents and Weaknesses in IRS'

Correspondence Audit Processes

Executive Summary Page 7 GAO/GGD-99-48 IRS' Correspondence Audits

tax auditors in the district offices. Such referrals are important
because IRS generally has one chance to audit a return. If these
complex schedules are not classified for potential audit issues,
tax noncompliance is more likely to go undetected.

Second, the workpapers for the 1996 correspondence audits showed
that up to about one- third of the audits had little or no
evidence of how the tax examiner did the audit or supported any
recommended taxes. IRS' manual requires that tax examiners
adequately support their audit recommendations and document that
support in the audit workpaper files. Analyses showed that the
less time spent on the audit, the less documentation appeared in
the workpapers. When supporting documentation is lacking, it can
create uncertainty about the additional taxes recommended in those
audits.

Third, the type of taxpayer documentation that was accepted to
justify EIC claims varied from service center to service center.
Tax examiners in at least three service centers did not follow
IRS' requirement to substantiate EIC claims through third- party
verification. On the basis of interviews at one of these service
centers, GAO found that this weakness occurred when service center
management eased requirements because of an inexperienced tax
examiner work force. Examiners at the other seven centers
generally required third- party verification when auditing EIC
claims. GAO could not tell from analyzing IRS' audit workpapers
from any service center whether taxpayers received an earned
income credit to which they were not entitled or were denied a
claim to which they were entitled.

And fourth, IRS' two audit quality reviews were unlikely to
identify weaknesses in the audits. The workpapers for the 1996
traditional correspondence audits had evidence of supervisory
review in an estimated 6 percent of the audits. IRS officials said
that one reason for this low percentage is that IRS does not
require supervisors to document their reviews.

Further, although IRS had weaknesses in classifying and
documenting support for correspondence audits, IRS' quality
measurement reviews gave high ratings in these areas. IRS' higher
ratings can be explained, in part, by two types of exclusions from
IRS' reviews, which usually occurred when taxpayers did not
respond to IRS' request for information. First, some service
centers excluded such audits (as well as those not recommending
tax changes) from the quality measurement review samples, even
though the audits are required to be included. Second, IRS allowed
service centers

Executive Summary Page 8 GAO/GGD-99-48 IRS' Correspondence Audits

to exclude certain audits from being measured against the quality
standard on workpaper documentation. These audits closed with no
responses from the taxpayer and no documentation of the audit
steps or support for the audit recommendations. According to IRS
requirements, the workpapers still are to document the rationale
for recommending additional taxes, even if taxpayers did not
respond.

In addition, over half the taxpayers in the traditional
correspondence audits did not respond to IRS' audit letters.
Without taxpayers' responses, IRS cannot be as certain about
actual tax liabilities. IRS had not collected data on why
taxpayers did not respond. IRS officials believed that taxpayers
were less likely to respond to standardized letters that are not
audit- specific and contain legal language that some taxpayers may
not understand. Tax examiners use standardized letters and do not
spend time crafting audit- specific letters, given the number and
simple nature of the audits. Taxpayers who did not respond to
these letters were less likely to pay their additional tax
assessments. Further, we estimated that 75 percent of the
taxpayers who asked IRS to reconsider these assessments or redo
these audits months after the audit closed had not responded to
IRS' letters during the audit.

GAO recommends that the IRS Commissioner improve controls to
better ensure that IRS' correspondence audit processes adhere to
existing audit requirements and standards on

 classifying filed returns, and in particular, referring returns
with complex schedules that may have potential tax changes to
staff with sufficient knowledge to classify them;

 documenting the support for audit findings and recommendations in
the audit workpaper files;

 ensuring consistency in the treatment of audited EIC claims by
collecting and using the information required, including
verification from third parties, to justify the claims; and

 including all types of closed audits across the 10 service
centers in the samples for measuring audit quality.

GAO also recommends that the IRS Commissioner

 require supervisors in the service centers to document their
reviews of audit workpapers;

 eliminate the discretion that service centers have to exclude
audits that lack documentation on the audit steps and support for
audit Recommendations

Executive Summary Page 9 GAO/GGD-99-48 IRS' Correspondence Audits

recommendations from the calculations IRS does to measure
adherence to the audit quality standard on workpaper
documentation; and

 determine the reasons, through statistically valid and cost-
effective means, for taxpayers' not responding to IRS' audit
letters, so that IRS can identify ways to encourage more taxpayers
to respond.

We requested comments on a draft of this report from the
Commissioner of Internal Revenue. Officials representing the
Assistant Commissioner for Customer Service and the IRS
Commissioner's Office of Legislative Affairs provided IRS'
comments in a February 23, 1999, meeting. IRS also documented
these comments in a letter dated March 12, 1999, which is
reprinted in appendix V.

Overall, IRS concurred with all of our recommendations and agreed
to take efforts to implement them. IRS officials provided
elaboration on our recommendation involving IRS' requirement to
classify returns, particularly returns with complex schedules.
Given that the number of returns audited through correspondence
that include complex schedules is relatively low, Customer Service
officials said that they would work with the Assistant
Commissioner (Research and Statistics of Income) on ways to
isolate returns with complex schedules so that they are not
automatically selected for correspondence audits. This would
provide district offices with the opportunity to first look at
these returns before they are audited through correspondence.
Agency Comments

Page 10 GAO/GGD-99-48 IRS' Correspondence Audits

Contents 2 Executive Summary 12 Screening and Classification of
Tax Returns 12 The Correspondence Audit Process 14 Audit Quality
Reviews 16 Objectives, Scope, and Methodology 16 Chapter 1

Introduction 19 Number of Returns Audited and the Results of the

Correspondence Audits 19

Time Taken to Conduct Correspondence Audits 21 Profile of Tax
Issues Audited Through Correspondence 22 Profile of Tax Returns
Audited Through Correspondence 23 Chapter 2

Correspondence Audits From 1992 to 1997

25 Most Tax Returns Audited Through Correspondence

Were Not Classified 25

Audits Did Not Always Have Adequate Documentation 25 Some Service
Centers Accepted Differing Levels of

Justification for EIC Claims 25

Limited Supervisory Reviews and Audit Quality Reviews 25 Minimal
Contact With Nonresponsive Taxpayers 25 Conclusions 25
Recommendations 25 Agency Comments and Our Evaluation 25 Chapter 3

Weaknesses in Processes and Requirements for Selecting and
Conducting Correspondence Audits

Appendix I: Overview of Audit Processes at Service Centers and
District Offices

38 Appendix II: IRS' Auditing Standards and Key Elements

for Correspondence Audits 41

Appendix III: Sampling and Data Analysis Methodology 42 Appendix
IV: Additional Comparisons of IRS Service

Center Workloads 45

Appendix V: Comments From the Internal Revenue Service

48 Appendix VI: Major Contributors to This Report 52 Appendixes

Tables Table 2.1: Number of Returns Audited and Additional Taxes
Recommended for Correspondence Audits

19

Contents Page 11 GAO/GGD-99-48 IRS' Correspondence Audits

Table 2.2: Frequency of Audit Closures for Correspondence Audits
in Total, Fiscal Years 1992- 97

21 Table 2.3: Comparison of Annual Incomes of Taxpayers

Audited Through Correspondence to All Taxpayers Filing Tax Returns

24 Table 3.1: Extent to Which Workpaper Files Had an Audit

Trail and Supporting Documentation on Audited Issues for Returns
in Our 1996 Study Population

25 Table II. 1: IRS' Audit Standards and Key Elements for

Correspondence Audits 41

Table IV. 1: Correspondence Examination Workload by Service Center
for Fiscal Year 1992

45 Table IV. 2: Correspondence Examination Workload by

Service Center for Fiscal Year 1993 45

Table IV. 3: Correspondence Examination Workload by Service Center
for Fiscal Year 1994

46 Table IV. 4: Correspondence Examination Workload by

Service Center for Fiscal Year 1995 46

Table IV. 5: Correspondence Examination Workload by Service Center
for Fiscal Year 1996

47 Table IV. 6: Correspondence Examination Workload by

Service Center for Fiscal Year 1997 47

Figure 1. 1: Simplified Overview of the Correspondence Audit
Process

15 Figure 2. 1: Percentage of Time That Each Issue Occurred

in Traditional Correspondence Audits, Fiscal Year 1996 23

Figure I. 1: Process for Selecting Individual Tax Returns for
Audit

40 Figures

Abbreviations

AIMS Audit Information Management System CEQMS Correspondence
Examination Quality Management System DIF Discriminant function
EIC Earned Income Credit IRS Internal Revenue Service TEFRA Tax
Equity and Fiscal Responsibility Act of 1982

Chapter 1 Introduction

Page 12 GAO/GGD-99-48 IRS' Correspondence Audits

Using a variety of tools, IRS checks all individual income tax
returns filed each year for compliance with the tax laws. These
compliance checks include reviews of some required information,
such as signatures and social security numbers, at the time a
return is filed; computerized corrections of what IRS labels math
errors; 1 computerized matching of information, such as income,
with the corresponding information reported to IRS by third
parties; and audits.

IRS audits individual income tax returns to help ensure that
taxpayers pay their proper tax liability. Audits are one of the
enforcement tools used by IRS in its efforts to close the tax gap
the amount of taxes owed but not voluntarily paid. IRS' most
recent estimate put the individual tax gap at over $125 billion
for 1992. IRS also estimates that it eventually collects about
one- quarter of the gross tax gap through its enforcement efforts.

The audits account for a small share of the returns filed annually
about 1 or 2 percent. Between 1992 and 1997, IRS audited between 1
million and 2 million individual income tax returns per year. In
1996, when about 119 million individual income tax returns were
filed, IRS audited 1.9 million returns 1.1 million at its 10
service centers and about 760,000 at its 33 district offices. In
selecting returns to audit, IRS attempts to identify those that
are most likely to have errors or audit potential owing to
questionable tax issues such as income, deductions, exemptions,
and credits. IRS' process for selecting returns with the highest
audit potential is important because audits can be costly for both
taxpayers and IRS.

In recent years, Congress has been concerned about IRS' use of its
enforcement tools, particularly about whether the tools are used
fairly and in a manner that does not impose excessive burdens on
taxpayers. Concerns such as these led, in part, to the passage of
the IRS Restructuring and Reform Act of 1998. The act included
provisions affecting audits.

IRS uses both computer technology and manual review to select
returns for audit. This selection process results in a
determination of whether a return has audit potential. If the
return does, the process helps identify specific tax issues to be
audited (audit issues) and the audit technique to be used
correspondence sent to taxpayers from IRS' service centers or
meetings with taxpayers or their representatives through IRS'
district offices.

1 For returns with math errors, IRS notifies taxpayers about the
errors and the adjusted tax liabilities through computer-
generated letters. The letters are similar to those used for
correspondence audits. Screening and

Classification of Tax Returns

Chapter 1 Introduction

Page 13 GAO/GGD-99-48 IRS' Correspondence Audits

IRS uses computer programs to screen tax returns filed each year
by individual taxpayers for audit potential. One program utilizes
a computerized series of discriminant function (DIF) formulas to
compute a score for each return. The higher the score, generally
speaking, the greater the likelihood that the return's reported
tax would be changed if audited. Districts have used DIF scores as
a primary source to select most of the returns they audited.
Service centers rarely use DIF scores because centers usually
focus on one audit issue, whereas DIF focuses on returns with
multiple, more complex issues.

Service centers have their own computer programs to identify
returns with issues to be audited through correspondence. The
types of issues to be audited vary from year to year as Congress,
IRS, and the local district offices and service centers focus on
particular aspects of the tax law. For example, in recent years,
earned income credit (EIC) claims and individual retirement
account penalties have been highlighted by the service centers.

Once the screening programs have identified tax returns with audit
potential, the returns are to be reviewed (or classified) by
qualified staff for each district office and service center. IRS
requires the returns to be classified to determine which can be
accepted as filed and, if not accepted, to identify all potential
audit issues. 2 IRS allows tax returns to be classified by
different types of auditors, depending on the complexity of the
returns and the issues. Returns with more complex audit issues
that require personal contacts with the taxpayers are to be
reviewed by revenue agents or tax auditors. Returns with less
complex issues that can be audited through correspondence are to
be sent to service centers to be reviewed by tax examiners. (See
app. I for detailed explanations of screening and classification.)
Once returns are classified and selected for audit, any identified
audit issue is to be documented in the audit workpaper file.

The IRS staff who do the classification have varying degrees of
training and audit experience. Revenue agents are the highest
graded auditors and are required to have at least 24 hours of
college- level accounting courses to audit the most complex tax
returns from corporations, other businesses, and higher income
individuals. Tax auditors, who usually have some accounting
education, audit less complex issues by meeting with taxpayers at
the district offices. Service centers use tax examiners to audit
the simpler individual issues. Tax examiners, who are not required
to have any

2 Identifying all potential audit issues is important because the
Internal Revenue Code, section 7605( b), provides for only one
inspection of a taxpayer's books or records per tax year unless
the taxpayer requests another inspection or unless the Secretary
of the Treasury notifies the taxpayer in writing that an
additional audit is necessary.

Chapter 1 Introduction

Page 14 GAO/GGD-99-48 IRS' Correspondence Audits

accounting education, audit simple issues through correspondence
and are not trained to audit complex issues on business and
investment schedules attached to tax returns.

According to IRS, tax returns assigned to the service centers are
to involve only one or two simple tax issues, such as EIC claims,
that can be audited through correspondence. Correspondence audits
are to exclude complicated tax issues, such as those on business
and investment schedules, and are designed to quickly review a
taxpayer's support for the tax issues.

The time needed to audit returns and establish the appropriate tax
liability depends, in large part, on the number of letters IRS
sends to taxpayers and the sufficiency of the taxpayers'
responses. IRS typically sends two letters one asking for
information or documentation and at least one reporting any
changes to the tax liability reported on the tax return as a
result of the audit. (See fig. 1.1.)

IRS service centers send an initial letter to notify taxpayers of
the issues on the return to be audited and to request information
on those issues within 30 days. If the taxpayer provides the
information requested, a tax examiner is to determine whether it
supports the audit issues. If it does provide the support, the tax
return can be accepted as filed and the audit closed with no
change in the reported tax liability. If the information does not
support the issues, the tax examiner can request further
clarification or documentation. After reviewing all the submitted
information, the tax examiner determines whether additional taxes
are warranted. Should the taxpayer not provide information to
support the audit issues, the tax examiner is to treat the issues
as unsubstantiated and recommend additional taxes.

Whether or not the taxpayer responds to IRS' initial letter, IRS
is to notify the taxpayer in writing of the results of the audit,
such as whether the tax examiner is recommending any changes to
the reported tax liability and, if so, the amount of change. This
IRS letter also gives taxpayers a 30- day period to agree or
disagree with the recommended results. If the taxpayer agrees, any
recommended tax changes are to be assessed as taxes owed. If the
taxpayer disagrees, IRS is required to inform the taxpayer of the
right to file a protest with IRS' Appeals Office to determine
whether the recommended taxes should be assessed. 3

3 IRS' Appeals Office is charged with trying to settle tax
disputes without litigation on the basis of what is fair to the
government and the taxpayer. The Correspondence

Audit Process

Chapter 1 Introduction

Page 15 GAO/GGD-99-48 IRS' Correspondence Audits

a Based on the information provided, the examiner can close the
audit with no changes to the tax liability. b Taxpayers can agree
or disagree with any proposed changes to their tax liability. If
taxpayers

disagree, they can ask IRS' Appeals Office to determine whether
the recommended tax amounts should be addressed as an additional
tax liability. c For various reasons, not all recommended tax
amounts are assessed and not all assessed tax

amounts are collected, regardless of the type of IRS audit.
Source: GAO summary of IRS data.

Figure 1.1: Simplified Overview of the Correspondence Audit
Process

Chapter 1 Introduction

Page 16 GAO/GGD-99-48 IRS' Correspondence Audits

Taxpayers who do not respond to these 30- day letters are to be
sent another notice giving them an additional 90 days to respond
before the recommended taxes are formally assessed. If the
taxpayer does not respond to this 90- day letter, the recommended
taxes are to be assessed, and the matter is to be referred to IRS'
Collection Division. 4

In addition to the formal appeals process, IRS allows taxpayers
who disagreed with the audit results or who did not respond to
IRS' audit letters to ask for a reconsideration of any additional
taxes that were assessed as a result of an audit. IRS extends this
option to taxpayers who provide new information about these
assessed taxes, including those who bypassed opportunities to
resolve their disputes in IRS' Appeals Office or the Tax Court.

IRS has two quality review functions to see whether correspondence
audits comply with its requirements and standards. 5 The audit
standards include such requirements as adequate consideration of
all significant audit issues and workpaper support for all audit
findings. (See app. II, which describes each standard.)

The first review is to be carried out by supervisors who are to
periodically review the tax examiners' work during the audits. The
results of these reviews are to be used in the tax examiners'
performance evaluations, but IRS does not require supervisors to
document the results in the audit workpaper files.

The second review is to be performed by quality assurance staff
within the service centers. The staff are to review a sample of
all closed audits for adherence to IRS' requirements and audit
standards. The results of these reviews are to be entered into the
Correspondence Examination Quality Management System (CEQMS)
database and used to identify procedural and systemic weaknesses
and to develop improvements where needed.

In response to a request from the Chairman, Subcommittee on
Oversight, House Committee on Ways and Means, this report (1)
provides information on the number, results, and duration of
correspondence audits as well as the characteristics of the
audited returns and (2) examines the processes

4 Upon receipt of the 90- day letter, the taxpayer may petition
the Tax Court to resolve the disputed tax liability. 5 IRS also
does Action 61 reviews to determine whether tax examiners are
responsive with taxpayer inquiries. These reviews consider such
items as professionalism, timeliness, and language used. Audit
Quality Reviews

Objectives, Scope, and Methodology

Chapter 1 Introduction

Page 17 GAO/GGD-99-48 IRS' Correspondence Audits

and requirements that IRS has had for years to govern
correspondence audits.

To provide information on the number, results, and duration of
correspondence audits, we gathered data from various IRS sources.
We gathered detailed data on audits closed in fiscal years 1992-
97 from IRS' Audit Information Management System (AIMS), including
the number of audits, amount of additional taxes recommended, and
type of audit closures (e. g., no tax changes). For the duration
and other characteristics, we gathered data from a statistically
valid sample of traditional correspondence audits closed in fiscal
year 1996 by reviewing audit workpaper files and IRS' individual
masterfile data. 6 We did not check the reliability of IRS' data
in AIMS and the masterfile.

In selecting the 1996 sample, we sought traditional correspondence
audits, which required us to exclude certain types of workload.
According to Customer Service officials, most returns audited in
fiscal year 1996 were not the traditional correspondence workload.
Rather, they included work on EIC returns that misreported Social
Security numbers which IRS transferred out of the audit program to
the math- error adjustment program in 1997. 7 Therefore, with the
concurrence of the Customer Service officials, we excluded these
and other types of tax returns from our sample to allow us to
analyze more traditional correspondence audit work. 8

This adjusted the fiscal year 1996 total of 1.1 million tax
returns audited to 335,050 returns, from which we randomly
selected 502 returns. We stratified our sample by whether (1) the
taxpayer responded to IRS' letters and (2) a tax change was
recommended. IRS located 446 of the 502 tax returns and related
audit workpapers for us to analyze. (See app. III for a detailed
description of our sampling methodology.)

To examine IRS' processes and requirements for correspondence
audits, we collected documentation and did interviews at IRS on
those processes and requirements. We then used three
methodologies. First, we reviewed

6 Because our review pertains to traditional correspondence audits
closed in 1996, all results are subject to sampling errors. Unless
otherwise noted, all estimates of percentages in this report have
a 95percent confidence interval of less than plus or minus 10
percentage points.

7 Congress created EIC in 1975 to help workers offset payroll
taxes and stay off welfare. In 1994, Congress expanded EIC
eligibility. 8 For example, we also excluded audits of certain
types of partnership returns as defined under the Tax Equity and
Fiscal Responsibility Act (TEFRA). In these audits, tax
adjustments flowed through service centers to allow necessary
adjustments to be made to individual partners' accounts on the
basis of audits by district office auditors of the partnership.

Chapter 1 Introduction

Page 18 GAO/GGD-99-48 IRS' Correspondence Audits

the 446 tax returns in our sample and related audit workpaper
files. For these returns, IRS' Fresno Service Center also provided
masterfile account data on what happened to the additional taxes
assessed as of April 1998. Second, we received written comments to
a survey from all 10 service center Compliance chiefs on the
processes and requirements. And third, we interviewed Examination
officials at the National Office, as well as the Directors,
Compliance chiefs, Examination chiefs, and selected Examination
officials in 4 of IRS' 10 service centers: Andover, MA; Fresno,
CA; Kansas City, MO; and Memphis, TN. (See app. IV for a
comparison of workloads at the service centers.)

We requested comments on a draft of our report. IRS provided
comments in a meeting on February 23, 1999, and a letter dated
March 12, 1999.

We have incorporated these comments as appropriate. We have also
summarized the comments at the end of chapter 3 and have reprinted
the letter in appendix V.

We performed our audit work between September 1997 and August 1998
in accordance with generally accepted government auditing
standards.

Chapter 2 Correspondence Audits From 1992 to 1997

Page 19 GAO/GGD-99-48 IRS' Correspondence Audits

During fiscal years 1992- 97, the correspondence audit program
experienced distinct shifts. IRS tripled the number of returns
audited from 1992 to 1994- 95 before reducing its correspondence
audit inventory in 1997. A higher percent of the audits closed
with no tax changes during fiscal years 1994- 96, and the percent
of audits that closed with no response from taxpayers more than
doubled from 1995 to 1997.

In terms of the characteristics for the traditional correspondence
audits closed in 1996, an average of about 11 months elapsed
between the start of the audit and the assessment of any taxes
recommended during the audit. Many of the audited returns included
complicated business and investment schedules that were not
audited. Because so many of the correspondence audits in our study
population year dealt with EIC claims, the reported income of most
of the audited taxpayers was below $15,000.

The number of correspondence audits and their results varied
considerably between 1992 and 1997, with three distinct periods. 1
As discussed with Customer Service officials and shown in table
2.1, these periods were 1992- 94, 1995- 96, and 1997.

Fiscal year Number of

audited returns a Direct audit

hours Direct audit

hours per return audited Staff years

Additional taxes recommended (dollars in millions) b

Recommended taxes per return audited

1992 339,120 273,744 0. 8 3,224 $740.1 $2,182 1993 232,237 245,692
1. 1 3,107 576.7 2, 483 1994 352,596 336,992 1. 0 3,223 621.4 1,
762 1995 1,054,573 703,848 0. 7 3,969 1,368.6 1, 298 1996
1,113,646 749,058 0. 7 3,536 1,765.9 1, 586 1997 757,670 572,957
0. 8 3,048 2,170.5 2, 865

a These figures do not include TEFRA returns. b IRS does not
assess or collect all of these additional recommended taxes. Our
earlier work shows, as of September 1997, that IRS assessed 76
percent and collected 43 percent of the additional taxes
recommended in service center correspondence audits that closed in
fiscal year 1997. See Tax Administration: IRS Measures Could
Provide a More Balanced Picture of Audit Results and Costs
(GAO/GGD-98-128, June 23, 1998).

Source: GAO analysis of IRS data.

In the first period (fiscal years 1992- 94), the number of tax
returns IRS audited through correspondence totaled a few hundred
thousand annually. Comparing 1992 to 1994, the amount of taxes
recommended overall and

1 The number of returns audited and the results of correspondence
audits also varied by service center. Appendix IV provides these
numbers and results across the 10 service centers for fiscal years
1992- 97. Number of Returns

Audited and the Results of the Correspondence Audits

Table 2.1: Number of Returns Audited and Additional Taxes
Recommended for Correspondence Audits

Chapter 2 Correspondence Audits From 1992 to 1997

Page 20 GAO/GGD-99-48 IRS' Correspondence Audits

per audited return declined. According to Customer Service
officials, these 3 fiscal years represented IRS' traditional
practices for doing correspondence audits.

In the second period (fiscal years 1995- 96), IRS' correspondence
audit inventory increased to over 1 million returns annually
because of an increase in audits of returns that claimed EIC.
Almost 800,000 of the audits that closed in 1996 involved EIC
returns with missing or invalid Social Security numbers. 2 This
increased audit focus on EIC arose from congressional concerns
about EIC noncompliance. With the threefold increase in audits,
the amount of recommended taxes more than doubled. As a result,
the tax amounts recommended per audited return declined by as much
as about half (from $2,483 in 1993 to $1,298 in 1995). 3

In the third period (fiscal year 1997), IRS decreased the
correspondence audit inventory to about 758,000. Customer Service
officials told us that IRS moved over 700,000 EIC returns with
misreported Social Security numbers to its math- error adjustment
program. To partially substitute for these transferred audits, the
officials said IRS began auditing more individual retirement
accounts and potential nonfilers. 4 With this shift in the audit
inventory in 1997, the number of audited returns and staff years
decreased and the amounts of taxes recommended overall and per
audited return increased compared to 1996. With the new types of
tax issues being audited, the audits were more likely to recommend
tax changes in higher amounts compared to audits of other types of
issues such as EIC.

The large number of audited EIC returns also affected audit
closures. From 1992 to 1994, less than 25 percent of the audits
closed with no changes to tax liability. Further, the default rate
that is, the rate at which audits closed when taxpayers did not
respond to tax assessment notices remained relatively constant at
around 40 percent. With the inclusion of EIC returns in 1995, the
no- change rate doubled to 46 percent, and the default rate
initially dropped to 29 percent. After the EIC returns were
transferred to the math- error adjustment program in 1997, the no-
change rate decreased to the 1992 level, but the default rate more
than doubled 2 Although our study population for fiscal year 1996
excluded this type of EIC return from the audit inventory, IRS
still audited many other types of EIC claims. 3 A contributing
factor was that many of these EIC audits recommended either no
change to the tax liability or lower amounts of tax changes
compared to audits of other tax issues. 4 Potential nonfilers, as
the name implies, are those who may not have filed returns. When
IRS is unable to find a return that matches financial information
reported by employers or financial institutions, it prepares a tax
return for the nonfiler (an SFR or Substitute for Return) and
computes the tax liability based on the reported information.

Chapter 2 Correspondence Audits From 1992 to 1997

Page 21 GAO/GGD-99-48 IRS' Correspondence Audits

compared to 1995, reaching 63 percent. Table 2.2 shows these
shifts in audit closures over the three time periods.

Audited returns closed as Fiscal year Agreed Unagreed No change
Default a Undeliverable b Total c

1992 28% 3% 13% 41% 14% 100% 1993 27 4 16 43 9 100 1994 31 4 23 36
6 100 1995 17 1 46 29 6 100 1996 15 2 36 43 4 100 1997 15 4 13 63
6 100

a IRS closed the audit with a recommended tax assessment, and the
taxpayer did not respond to IRS' letter on the assessment. b IRS
closed the audit with a recommended tax assessment when the letter
on the audit

recommendations was returned to IRS. c Percentages may not sum to
100 due to rounding.

Source: GAO analysis of IRS data.

In the correspondence audit process, many audits are started soon
after the returns are filed, while others are started a year or
more later. For returns involving EIC claims, beginning the audit
soon after the return is filed increases the likelihood that the
issues can be examined before any tax refunds are released. For
other returns filed in the spring, the audits often do not begin
until the following winter. 5 Audits of taxpayers who do not file
tax returns may take a year or more to begin.

For traditional correspondence audits closed in 1996, we found
that the time between the filing of the return and the start of
the audit averaged 10 months, and the time between the start of
the audit and the assessment of any additional taxes took, on
average, over 11 more months. The length of the audit varied,
depending on whether taxpayers responded and how it was closed. On
the one hand, audits took about 7 months 6 when taxpayers
responded to IRS' notice of audit and the audits closed without
recommending additional taxes. On the other hand, when taxpayers
did not respond to IRS' 90- day letters notifying them of any
additional tax assessments and the audits closed with the taxpayer
defaulting, audits took much longer. For example, an average of
about 13 months elapsed between the start of the audit and the
assessment of any additional taxes.

5 This delay is often necessary to allow IRS to enter return data
into its computers, associate tax returns with information returns
(e. g., Forms W- 2 and 1099), process returns filed after April
15, and classify returns for audit potential.

6 The 95- percent confidence interval for this estimate ranges
from a low of 185 days to a high of 248 days.

Table 2.2: Frequency of Audit Closures for Correspondence Audits
in Total, Fiscal Years 1992- 97

Time Taken to Conduct Correspondence Audits

Chapter 2 Correspondence Audits From 1992 to 1997

Page 22 GAO/GGD-99-48 IRS' Correspondence Audits

Seven months of this time were spent simply waiting for the
taxpayers to respond to one or more of IRS' letters.

For these audits, the average time that elapsed from the filing of
the return to the additional tax assessment was 21 months. This
elapsed time is significant in two respects. First, IRS collects
27 percent of taxes assessed through correspondence audits before
the assessment becomes final and another 49 percent within 15
weeks of the final assessment. 7 Second, Congress' 1998
legislation to restructure IRS contains provisions that suspend
the accrual of interest and some penalties if IRS fails to notify
the taxpayer of additional tax liabilities within 18 months after
a return is filed. 8 After 2003, suspension will begin 12 months
after the tax return is filed. Delays in completing the audit,
therefore, could reduce revenue from penalty and interest
assessments. 9

Aside from these considerations, the time to assess any
recommended taxes in these audits may be extended if taxpayers
choose to exercise their right to appeal these audit
recommendations. For audits being appealed or otherwise disputed
by the taxpayer, a long amount of time can elapse before a
determination on the tax assessment can be reached.

The traditional correspondence audits closed in 1996 dealt with a
few types of simpler tax issues rather than the more complex
business and investment tax issues on individual returns. The
types of tax issues audited through correspondence primarily
included EIC claims and selfemployment tax. On average, each
traditional correspondence audit dealt with two tax issues. As
figure 2.1 shows, EIC and related issues, such as the taxpayer's
filing status and dependent exemptions, accounted for over half of
these audited tax issues.

7 Percentages are taken from an earlier analysis of correspondence
audits closed in fiscal year 1992. See GAO/GGD-98-128, June 23,
1998. 8 The legislation is not explicit, however, on which IRS
letter the one recommending additional tax assessments or the one
notifying taxpayers of the tax assessment will serve to notify the
taxpayer. 9 According to Customer Service officials, to reduce the
time to identify any additional assessments through audits as well
as to give taxpayers more options, four service centers will be
testing during fiscal year 1999, the impact of a new letter that
combines the initial letter and the 30- day letter. This new
letter will give taxpayers the choice of agreeing to pay a
specific amount of tax and interest (as well as any penalty
amount) at the start of the audit. If taxpayers do not agree to
pay, they can respond by providing information in support of the
original tax amount that they reported. Profile of Tax Issues

Audited Through Correspondence

Chapter 2 Correspondence Audits From 1992 to 1997

Page 23 GAO/GGD-99-48 IRS' Correspondence Audits

Source: GAO summary of IRS data.

Although IRS' traditional correspondence audits closed in 1996
usually addressed simpler tax issues, an estimated 65 percent of
the audited returns had at least one schedule attached. Many of
these schedules (such as C, D, E, and F) reported more complex
business and investment tax

Figure 2.1: Percentage of Time That Each Issue Occurred in
Traditional Correspondence Audits, Fiscal Year 1996

Profile of Tax Returns Audited Through Correspondence

Chapter 2 Correspondence Audits From 1992 to 1997

Page 24 GAO/GGD-99-48 IRS' Correspondence Audits

issues that were not audited. For the audited returns with
schedules, an estimated

 61 percent had schedules supporting their EIC claims;

 26 percent had a schedule A or B to report itemized deductions or
interest and dividend income;

 22 percent had a schedule C to report business income from
selfemployment;

 10 percent had a schedule D to report capital gains or losses;

 8 percent had a schedule E to report gains or losses from rental
property, partnerships, and other entities; and

 3 percent had a schedule F to report income from farming. In
addition, the individual taxpayers subjected to traditional
correspondence audits that closed in 1996 usually reported annual
incomes of less than $15,000 on their audited returns. In fact, as
table 2.3 shows, an estimated 76 percent of the audited returns
reported taxable incomes of less than $15,000. However, 44 percent
of all income tax returns filed by individuals in 1996 reported
taxable incomes of less than $15,000. The returns audited through
correspondence tend to report income below $15,000 because they
also report simpler tax issues and EIC claims, both of which can
be associated with lower amounts of income.

Percent of individual tax returns 1996 traditional audit
population by 1996 data on filed tax returns by a Range of
reported income Adjusted gross income Taxable income Adjusted
gross income Taxable income

$0 < $15,000 55% 76% 34% 44% $15,000 < $25,000 20 6 18 17 $25,000
< $50,000 13 12 25 24 $50,000 < $100,000 9 4 17 11 $100,000 and
over 3 2 5 4

Total 100% 100% 99% b 100%

a 1996 represents the filing year and not the year the audits were
closed. b Percentages may not sum to 100 due to rounding. Source:
GAO analysis of results from our study population and IRS data.

It is important to recognize that table 2.3 does not compare all
taxpayers with audited returns to all taxpayers who filed returns.
Taxpayers with higher annual incomes and more complex tax issues
are typically audited through IRS district offices using other
types of audit methods than correspondence audits.

Table 2.3: Comparison of Annual Incomes of Taxpayers Audited
Through Correspondence to All Taxpayers Filing Tax Returns

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 25 GAO/GGD-99-48 IRS' Correspondence Audits

The correspondence audit process is one of the tools IRS uses to
help ensure that taxpayers pay their proper tax liability. It is
designed to audit, through the mail, all questionable issues
identified on less complicated tax returns. However, our work
pointed to a number of weaknesses in the processes for
implementing requirements for selecting returns to audit and for
auditing those returns.

We found that service centers did not follow IRS' requirements for
classifying returns and referring returns with complex business or
investment schedules to district offices. IRS manuals state that
after being identified for audit, tax returns should be classified
to ensure that all issues worth auditing are identified. Returns
with complex issues and schedules are to be classified by revenue
agents or tax auditors who have the training and experience with
complex tax issues. Returns with less complicated issues can be
classified by tax examiners. However, service centers used tax
examiners to classify most returns, some of which involved complex
schedules, and rarely referred returns with complex schedules to
an IRS district office.

In examining the administrative files and audit workpapers for the
traditional correspondence audits closed in 1996, we found that,
excluding nonfilers, an estimated 69 percent did not have any
evidence of classification as required. When returns are not
classified by trained staff, potential noncompliance may go
undetected. Moreover, the lack of classification can lead to IRS'
unnecessarily auditing returns or issues and thus burdening
taxpayers.

For the returns that had been classified, we found indications
that tax examiners did most of the classification. Having tax
examiners classify one or two issues on simpler tax returns is to
be expected. However, in 1996, an estimated 20 percent of the tax
returns traditionally audited through correspondence included
complex business and investment schedules. 1 Specifically, IRS
requires returns with more than one business schedule or schedules
C or F to be referred to district offices for classification and
possible audit. We found referrals of such returns to IRS district
offices in less than an estimated 1 percent of the traditional
correspondence audits.

We reviewed a judgmental sample of about 30 files with these
complex returns and found that one- third could have merited
referral to a district.

1 We estimate that 50 percent of the workpapers for these returns
with complex schedules (schedules C, D, E, or F) had no evidence
of classification. The 95- percent confidence interval for this
estimate ranges from 38 to 62 percent. Most Tax Returns

Audited Through Correspondence Were Not Classified

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 26 GAO/GGD-99-48 IRS' Correspondence Audits

For example, we found complex business returns that deducted
sizable expense amounts compared to the reported business income
but no evidence that anyone at IRS looked at the audit potential
of these deductions. Without a control such as classification of
returns with complex schedules by IRS tax auditors or revenue
agents with commensurate training and experience, IRS could not
verify whether returns slated for correspondence audits should
have been referred to district offices.

We surveyed compliance chiefs at all 10 service centers, who
confirmed that not all returns are classified and that tax
examiners do most classifications. For example, five chiefs noted
that their service centers usually classify less than one- third
of the returns they audit. When classification occurs, tax
examiners classify 85 percent or more of the returns according to
six compliance chiefs; the other four centers use tax examiners
but also use tax auditors or revenue agents. Two compliance chiefs
added that tax examiners are trained to audit simple issues and
are not ideal for classifying returns with complex issues. Also,
many chiefs noted that service centers might not have referred
returns with complex issues because they believed the returns
would not meet district office criteria. They said service centers
tend to receive returns because computerized criteria targeted a
simpler issue to audit, usually before districts began their
classification of returns.

In our follow- up on the classification practices, Customer
Service officials told us that not all tax returns needed to be
classified to identify all potential audit issues. They said that
correspondence audits were designed for quickly auditing one or
two issues through the mail and then moving on to other audits
with similar types of issues. As a result, these officials said
that a fuller classification beyond the one or two issues is
unnecessary for simple returns, which most of the audit workload
comprises. However, these officials also said that such a fuller
classification might be useful for returns with complex issues and
schedules.

IRS' manuals and audit quality standards require that tax
examiners adequately support their audit recommendations in the
audit workpaper files. Officials at the four service centers we
visited told us that documents with information on the case
history, audit issue development, and support for recommended
taxes are key documents in developing an audit trail on the
examiners' actions as well as in documenting support. Without an
audit trail or documented support, neither IRS nor the taxpayer
can be assured of the basis for any additional taxes recommended.
Audits Did Not Always

Have Adequate Documentation

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 27 GAO/GGD-99-48 IRS' Correspondence Audits

Our analysis of these and other documents in the workpaper files
for traditional correspondence audits closed in 1996 showed that
about onethird of the audits had little or no evidence of
documents to explain the audit trail. Furthermore, about one-
fourth of the issues audited had little or very little support in
the workpaper files for the recommended additional taxes. In
making our determination, we looked for these key documents or any
evidence to show what the tax examiners did or how they arrived at
certain tax recommendations. When any evidence was in the files,
we assessed the extent to which it documented the tax examiners'
actions, ranging from none or a very little extent to a very great
extent. Kansas City Service Center staff confirmed our
determinations for files with none or a very little extent. (See
table 3.1.)

Extent Audit trail a Supporting documentation b Very great 7% 9%
Great 28 27 Some 31 38 Little 25 17 Very little c 8 None 9 c Total
100 99 Note: Percentages may not sum to 100 due to rounding. a
This analysis is based on an estimated 335,050 tax returns in the
traditional correspondence audits

closed in 1996. b This analysis is based on an estimated 776,200
audited issues on tax returns in the traditional

correspondence audits for which an indication of supporting
documentation was given. c Not applicable.

Source: GAO analysis of results from our study population.

In attempting to determine the reasons for this lack of
documentation, our analyses showed a direct relationship between
the amount of time charged and degree of documented support. That
is, the less time charged to an audit, the less the workpapers
documented the audit findings. (See app. III for a detailed
description of our analyses.) It is possible that some audits may
be so brief that documentation to a little extent may be enough.
For example, Customer Service officials said that audits of
nonfilers or in which taxpayers did not respond might have less
documentation. However, these officials could not describe an
audit for which no or very little documentation would be
acceptable.

At two of the four service centers we visited, we found that
management affected how much time tax examiners spent documenting
their work. For example, the Andover Service Center management
told us they encouraged tax examiners to document their work in
the files. We found

Table 3.1: Extent to Which Workpaper Files Had an Audit Trail and
Supporting Documentation on Audited Issues for Returns in Our 1996
Study Population

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 28 GAO/GGD-99-48 IRS' Correspondence Audits

that Andover examiners spent 1.2 hours per audited return in 1996
and were generally consistent in documenting their work in the
files. However, the Fresno Service Center management eased
documentation requirements in 1996 because of the increase in EIC
work. As such, Fresno tax examiners spent 0.6 hours per audited
return and had a larger proportion of audit files with little or
no audit trail or documented support for the additional taxes
recommended.

Compliance chiefs in the four service centers we visited told us
that the significant increase in the number of audits during 1995
and 1996 added pressure on tax examiners to respond to taxpayers
within prescribed time frames and to close the audits as quickly
as possible. They said service centers with the greatest increase
in workload had even more pressure to quickly audit the returns.
As a result, they said the tax examiners had difficulty
documenting their audit work when they were expected to quickly
audit so many returns.

In addition to the increased workload, the four chiefs also
indicated that service center management diverts tax examiners
from their audit work to assist in nonaudit duties, such as
answering telephone calls during the filing season. According to
various service center officials, the filing season is a critical
time for tax examiners because most of the EIC workload is done
then. Thus, they said these diversions placed additional pressure
on the tax examiners to work quickly.

Because the traditional correspondence audits that closed in 1996
contained a large number of returns claiming EIC, we determined
whether the service centers adhered to IRS' seven- point
eligibility test, including third- party verification to support
the EIC claims. For example, taxpayers claiming EIC are required
to provide information on the children being served by EIC, such
as their birth dates, Social Security numbers, and school records.
IRS sends taxpayers a questionnaire with the initial letter to
solicit documentation, including information from third parties,
that justifies their EIC claims.

We found that the 10 service centers accepted differing levels of
documentation to justify EIC claims. For the traditional
correspondence audits that closed in 1996, at least three service
centers allowed EIC claims based on assurances from the taxpayers
but without documentation or third- party verification. Officials
at one of the three service centers stated that the documentation
requirements were eased because the center's tax examiner work
force was inexperienced and unable to otherwise audit the large
volume of returns within the prescribed time frames. As taxpayers
Some Service Centers

Accepted Differing Levels of Justification for EIC Claims

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 29 GAO/GGD-99-48 IRS' Correspondence Audits

responded to IRS' letters, service center management pushed the
tax examiners to provide timely responses. To do this as quickly
as possible, the service center eased the documentation
requirements. We did not ask IRS to check these EIC claims to
determine whether any of these taxpayers received an EIC to which
they were not entitled or were denied a claim to which they were
entitled.

The other seven service centers required varying degrees of
taxpayer and third- party justification. Of these seven service
centers, at least three required taxpayers to submit
documentation, such as birth certificates and Social Security
cards or third- party residency verification from schools or day
care providers.

IRS has two levels of review to check whether correspondence
audits adhered to established requirements and audit standards.
The first review is to be performed by the tax examiners'
supervisors while the audits are being conducted. A second review
is to be conducted on a sample of closed audits by each service
center's audit quality staff to measure adherence to the audit
quality standards.

We found evidence of supervisory review in an estimated 6 percent
of the workpaper files for the traditional correspondence audits.
In searching these files, we counted any supervisory comment that
pertained to the quality of the audits and the support for the
auditor's recommendation. We also found that the requirement for
supervisory review varied by service center. For example, one
center required supervisors to review all the nochange audits plus
the audit workload of each tax examiner monthly. Another center
required supervisors to review eight audits per tax examiner per
quarter. A third center required reviews of six audits per
examiner for an entire year.

Customer Service and service center officials told us that
supervisors could not have reviewed all of the 1.1 million audits
that were closed in 1996. They explained that the number of audits
that can be reviewed depends on the supervisor's workload, which
typically includes overseeing between 10 and 15 tax examiners,
with each examiner having roughly 50 audits ongoing at any given
time. The officials said they believed that the supervisors
reviewed more than 6 percent of the closed audits. It may be
possible that some supervisors did reviews but just did not
document them. IRS does not require supervisors to document their
reviews against the audit quality standards in the audit
workpapers. Even so, after we shared the results of our analyses
of the 1996 audits, Andover Service Limited Supervisory

Reviews and Audit Quality Reviews

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 30 GAO/GGD-99-48 IRS' Correspondence Audits

Center officials said they would implement a control to ensure
that supervisory reviews were documented in the workpapers.

As for the reviews to measure audit quality, IRS requires each
service center to review a sample of all its correspondence audits
closed annually, regardless of how they closed, for adherence to
IRS' audit standards. For this sample, service centers are
required to rate adherence to audit standards, such as the
existence of documentation in the workpapers on the audit steps,
and support for audit recommendations, such as additional taxes.
The results of these reviews are to be entered into the
Correspondence Examination Quality Management System (CEQMS) and
analyzed to measure the extent of adherence to each quality
standard and to identify problem areas. 2

Further, we found some disparities between the CEQMS data and our
results. For example, the CEQMS data for the correspondence audits
closed in 1996 showed that almost 100 percent of the audits had
been appropriately classified and that 96 percent of the audits
met IRS' standards for workpaper documentation. However, as stated
earlier, our review of the workpapers for traditional
correspondence audits closed in 1996 found that, excluding
nonfilers, an estimated 69 percent had no evidence of
classification, and an estimated 34 percent had little or no
evidence of documents to explain the audit trail. We asked IRS
officials about these disparities. Customer Service officials had
no opinions about the reasons for the disparities. Officials at 6
of IRS' 10 service centers attributed the disparities to
unreliable CEQMS data because service centers are excluding
certain types of audits from the review sample and from being
measured against the audit standard on workpaper documentation.

In following up with service center officials about CEQMS data
reliability, we found that not all service centers were following
IRS' requirements for drawing the review samples. For example,
according to these officials, some centers' samples excluded
closed audits in which the taxpayers did not respond to IRS'
letters because there were few workpapers to review. Even when
taxpayers do not respond, the audit recommendations still need to
be supported. Similarly, some service centers also excluded audits
that recommended no tax changes. Even when recommending no tax
changes, examiners are still required to follow IRS' audit quality
standards and document adherence to those standards.

2 We found evidence of such a quality review for less than 1
percent of the traditional correspondence audits closed in 1996.

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 31 GAO/GGD-99-48 IRS' Correspondence Audits

We also found a similar exclusion when taxpayers did not respond
with information about the tax issue being audited. IRS allowed
service centers to exclude such audits from being measured against
the audit standard on workpaper documentation. Customer Service
officials told us that service centers might instruct their
quality reviewers to record not applicable for the documentation
standard when required documentation is missing if taxpayers do
not respond. Regardless of the issue being audited, less
documentation of the audit trail or audit support would be
expected when taxpayers do not respond. Even so, the service
centers had some reason for contacting the taxpayers about the
potential for additional tax liabilities, and they are required to
document that reason along with the support for the additional
taxes that are to be recommended when taxpayers do not respond.

For the traditional correspondence audits closed in 1996, an
estimated 57 percent of taxpayers did not respond to IRS' audit
letters. IRS has not identified the reasons why these taxpayers
did not respond. We found that tax examiners sometimes tried to
contact these nonresponsive taxpayers beyond the minimal
requirements for correspondence audits an initial letter and a
final letter to close the audit. Examination officials at the four
service centers we visited stated that additional contacts would
increase the time spent on such audits, with no guaranteed
results, and would prevent tax examiners from auditing more
returns.

We asked Customer Service and service center officials why so many
taxpayers did not respond. They said they have not studied the
reasons but offered three possible explanations. First, some
taxpayers might be overwhelmed or intimidated by IRS' letters and
be uncomfortable with responding. Second, some taxpayers might not
understand IRS' letters and not know how to respond. Third, other
taxpayers might know that they owe the additional tax but hope
that their nonresponsiveness would discourage IRS from trying to
collect that tax.

For the traditional correspondence audits closed in 1996, some IRS
letters might have been overwhelming or confusing to taxpayers.
For example, some letters made blanket, standardized requests for
information without specifying the reason for IRS' concerns; one
letter requested up to 14 pieces of information on an alimony
issue without stating the specific information needed. Further, we
determined that almost half the letters might have been hard to
understand because they were written in generic language. An
estimated 11 percent of the letters were tailored to a specific
audit; in the other audits, either the letters had both generic
and tailored Minimal Contact With

Nonresponsive Taxpayers

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 32 GAO/GGD-99-48 IRS' Correspondence Audits

language or the workpaper files did not have the letters for us to
make a determination.

Customer Service and service center officials acknowledged that
their letters were generic and contained legal language that many
taxpayers might not be able to understand. The officials also
stated that tax examiners tend to rely on this generic language
because it is readily accessible in IRS' computers and expedites
the audit. They said that preparing letters tailored to each
taxpayer takes more time and thus limits the number of audits that
can be processed.

We talked to various Customer Service officials about any plans to
gather information from taxpayers on why they do not respond.
Although these officials identified some efforts, as of January
1999, these efforts either were not implemented or had not
provided any data on why taxpayers did not respond to IRS' audit
letters. 3 Customer Service officials told us that gathering
information from these taxpayers could be difficult, but the
benefits of the information collected from them would more than
likely be cost- effective. Without knowing the reasons for
nonresponsiveness, IRS cannot know the extent to which the clarity
of its correspondence, or other factors, contributed to taxpayers'
not responding to IRS' audit letters.

Without taxpayer responses, IRS' determinations of the correct tax
liability are less certain. When taxpayers do not respond, service
centers are to assess any additional taxes that they think are
owed. Our analyses of correspondence audits closed in 1996 also
identified other negative effects when taxpayers did not respond.
The audits usually took longer to complete, and additional taxes
owed took longer to collect compared to audits where taxpayers
responded. Those not responding paid less of their recommended tax
assessment than taxpayers who responded. As of April 1998,
taxpayers who responded paid an estimated 79 percent 4 of the
additional taxes, while those not responding paid an estimated 31
percent.

Further, an estimated 72 percent of those asking IRS to reconsider
the recommended taxes had not responded to IRS' final letter. 5
Occurring

3 The Memphis Service Center proposed a study in 1996 to gather
information on why taxpayers were not responding to IRS' audit
letters. According to Customer Service officials, they have not
approved this proposal because it was not specific enough. In
January 1999, Customer Service officials also said that a customer
satisfaction survey might have some information on why taxpayers
do not respond.

4 The 95- percent confidence interval for this estimate ranges
from 66 to 92 percent. 5 An estimated 29 percent of all taxpayers
with traditional correspondence audits closed in 1996 asked IRS to
reconsider their recommended taxes.

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 33 GAO/GGD-99-48 IRS' Correspondence Audits

months after the audits closed usually when IRS took action to
collect these assessments audit reconsiderations cause IRS to redo
all or part of the audit, adding to the time for IRS and taxpayers
to determine the correct tax liability. With the added time,
taxpayers also incur additional interest charges on the unpaid
assessments that remain after the reconsideration. We found that
IRS closed an estimated 72 percent of the reconsiderations without
reducing the assessments. These effects might have been minimized
if the taxpayers had responded during the audit or had acted on
their appeal rights at the close of the audit.

Customer Service officials described their efforts to improve
responsiveness. Customer Service officials told us they initiated
an effort to revise the correspondence audit letters in August
1998. They also indicated that IRS is contracting with a private
firm to improve the clarity of various IRS letters, starting with
collection notices. These officials said the correspondence audit
letters eventually would be covered under this contract. Finally,
they said that IRS plans to test a toll- free phone system in 1999
to encourage taxpayers who receive audit letters to call for
clarification if needed. Customer Service officials also shared
their plans to measure the effects of each effort on taxpayer
responsiveness. However, Customer Service officials did not know
whether these efforts would improve taxpayer responsiveness
because the efforts were in the initial stages.

IRS uses audits to help ensure that taxpayers pay their proper tax
liability. In reviewing IRS' correspondence audit processes and
requirements, we found weaknesses that undermine its ability to
(1) identify and audit all questionable tax issues on tax returns,
(2) support its audit recommendations for additional taxes, (3)
consistently accept justification for EIC claims, (4) review
adherence to audit requirements and quality standards, and (5)
solicit taxpayers' responses about the tax issues being audited.

Tax returns that are slated for correspondence audits are to be
reviewed by knowledgeable staff to identify all questionable
issues and to ensure that returns with more complex audit issues
have been forwarded to the appropriate auditors. Such a
classification is important because IRS generally has one chance
to audit a tax return. In reviewing the correspondence audits
closed in 1996, we found that an estimated 69 percent had not been
classified. When returns were classified, tax examiners usually
served as the classifiers. Tax examiners are not trained to
identify or audit complex issues, such as investments and business
deductions. However, some of the returns we analyzed included
complex Conclusions

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 34 GAO/GGD-99-48 IRS' Correspondence Audits

schedules, and less than 1 percent of these returns had been
referred to district offices, where revenue agents and tax
auditors are trained to classify and audit the more complex
issues. As a result, these complex returns may have had
questionable issues that were neither classified nor audited.

We found that the workpaper files for about one- third of the
traditional correspondence audits closed in 1996 provided little
or no audit trail for the tax examiners' actions, and about one-
fourth of the issues audited had little or very little support in
the workpaper files for the recommended additional taxes. Without
proper documentation of the workpapers, neither IRS nor taxpayers
can be assured that the recommended additional taxes are accurate
and supported.

Our analyses indicated that as the time spent by tax examiners on
each audit decreased, so did the amount of support in the audit
workpaper files. IRS officials said that the large volume of
returns that had to be audited pressured the tax examiners to work
quickly. Adding to this pressure has been the diversion of tax
examiners to nonaudit work during each filing season.

As an outgrowth of this emphasis on working quickly, some service
centers eased the documentation requirements for claiming EIC. At
least three service centers allowed EIC claims without the third-
party verification of the taxpayer's eligibility. The other
service centers required more extensive verification than these
three centers. We could not determine from the workpapers whether
these audits allowed any taxpayers to claim an EIC to which they
were not entitled.

IRS' two types of quality reviews were unlikely to identify such
weaknesses in the correspondence audit process. We found little
evidence of supervisory review for the traditional correspondence
audits closed in 1996. IRS does not require supervisory reviews to
be documented. Although we found weaknesses in classifying and
documenting support, IRS' quality measurement reviews of closed
correspondence audits gave these areas high ratings. The
discrepancies between our findings and IRS' measurements can be
explained, in part, by some service centers' not following IRS'
requirement to include all types of audit closures in the review
samples. Further, service centers were allowed to exclude audits
in which taxpayers did not respond and that lacked required
documentation on the audit steps and support for recommended taxes
from being measured against the audit standard on workpaper
documentation.

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 35 GAO/GGD-99-48 IRS' Correspondence Audits

Finally, in over half of the traditional correspondence audits
closed in 1996, the taxpayers did not respond to IRS' queries
during the audit. IRS officials had not determined the reasons but
believed that the language in their letters might be a factor.
Audit letters often had generic and legal language that was
difficult to understand. Because of pressures to process large
audit workloads, tax examiners were not likely to take the time to
tailor letters to a specific audit. And when taxpayers did not
respond, IRS had to assess the additional taxes under audit
without knowing for sure if all these taxes were owed. Further,
taxpayers who did not respond paid less of their recommended taxes
owed. Without knowing the reasons for taxpayers' not responding,
IRS cannot know the extent to which its letters, or other factors,
contributed to their nonresponsiveness.

We did not attempt to measure the specific effects of the
weaknesses we found. Nevertheless, each weakness can, individually
or in combination, erode the integrity of IRS' correspondence
audit processes, which are designed to help ensure that taxpayers
pay the correct tax. Improving controls over the audit processes
as well as collecting more information about the weaknesses would
help to accomplish these ends.

GAO recommends that the IRS Commissioner improve controls to
better ensure that IRS' correspondence audit processes adhere to
existing audit requirements and standards on

 classifying filed returns, and in particular, referring returns
with complex schedules that may have potential tax changes to
staff with sufficient knowledge to classify them;

 documenting the support for audit findings and recommendations in
the audit workpaper files,

 ensuring consistency in the treatment of audited EIC claims by
collecting and using the information required, including
verification from third parties, to justify the claims; and

 including all types of closed audits across the 10 service
centers in the samples for measuring audit quality.

GAO also recommends that the IRS Commissioner

 require supervisors in the service centers to document their
reviews of audit workpapers;

 eliminate the discretion that service centers have to exclude
audits that lack documentation on the audit steps and support for
audit recommendations from the calculations IRS does to measure
adherence to the audit quality standard on workpaper
documentation; and Recommendations

Chapter 3 Weaknesses in Processes and Requirements for Selecting
and Conducting Correspondence Audits

Page 36 GAO/GGD-99-48 IRS' Correspondence Audits

 determine the reasons, through statistically valid and cost-
effective means, for taxpayers' not responding to IRS' audit
letters, so that IRS can identify ways to encourage more taxpayers
to respond.

We requested comments on a draft of this report from the
Commissioner of Internal Revenue. Officials representing the
Assistant Commissioner for Customer Service and the IRS
Commissioner's Office of Legislative Affairs provided IRS'
comments in a meeting on February 23, 1999. IRS also documented
these comments in a letter dated March 12, 1999, which is
reprinted in appendix V.

Overall, IRS concurred with all of our recommendations and agreed
to take efforts to implement them. IRS officials provided
elaboration on our recommendation involving IRS' requirement to
classify returns, particularly returns with complex schedules.
Given that the number of returns audited through correspondence
that include complex schedules is relatively low, Customer Service
officials said that they will work with the Assistant Commissioner
(Research and Statistics of Income) on ways to isolate returns
with complex schedules so that they are not automatically selected
for correspondence audits. This would provide district offices
with the opportunity to first look at these returns before they
are audited through correspondence.

IRS also provided us with a technical comment on figure 1.1 to
clarify at what point recommended taxes become assessed. We have
incorporated this comment into the report. Agency Comments and

Our Evaluation

Page 37 GAO/GGD-99-48 IRS' Correspondence Audits

Appendix I Overview of Audit Processes at Service Centers and
District Offices

Page 38 GAO/GGD-99-48 IRS' Correspondence Audits

This appendix illustrates the differences between the audits
conducted at IRS' service centers and district offices.

Approximately 119 million individual taxpayers filed returns in
1996. Some returns are likely to have errors or misinterpretations
of the tax laws. IRS' process for identifying returns with the
most significant errors or misinterpretations is important because
only 1.9 million, or less than 2 percent, of the returns are
audited. IRS uses computer programs to select many of the returns
for audit and manual reviews of the returns (known as
classification) to identify tax issues for audit. The selection
and classification processes help IRS to choose an auditing
technique correspondence sent from service centers or face- to-
face meetings at the district offices.

For fiscal year 1996, service centers did two- thirds of the 1.9
million audits under the Compliance Branch of Customer Service. 1
The remaining audits were done through district offices under IRS'
Examination Division. Audits at the service centers differ from
those at the district offices in terms of the (1) audit techniques
being used and complexity of tax issues being audited (i. e.,
audit issues) and (2) selection of the audited returns.

First, service center audits generally involve one or two audit
issues that are simple enough to be audited through
correspondence, such as the selfemployment tax. District office
audits generally involve multiple audit issues that are complex
enough to require face- to- face meetings with taxpayers. Complex
issues include gains and losses from various kinds of investments
as well as business income and deductions.

Second, the methods to screen and select returns with audit
potential because of apparent noncompliance differ somewhat at the
districts compared to the service centers. These methods include
using various sources to identify returns with audit potential and
classifying those returns to check not only whether they should be
audited but also what issues should be audited. For example,
although centers and district offices use computer programs to
identify tax returns for audit, these programs differ in scope.
The programs usually identify returns with one audit issue for
service centers and with multiple, more complex issues for
district offices.

1 Correspondence audits were conducted under the Examination
Division until about 1995. At that time, IRS transferred this
audit program to the Collection Division until 1996, when it was
moved to Customer Service. Processes to Screen

and Select Tax Returns for Audit

Differences Between Service Center and District Office Audits

Appendix I Overview of Audit Processes at Service Centers and
District Offices

Page 39 GAO/GGD-99-48 IRS' Correspondence Audits

In addition to computer screening, IRS uses about 30 other sources
to select returns with audit potential. 2 District offices tend to
use these sources more than service centers. Sources include
referrals from inside and outside IRS; a program on tax preparers
who are associated with noncompliant returns; projects on groups
of taxpayers that are known or suspected to be noncompliant;
returns that are related in the same or different tax year to a
noncompliant return; and information filed by third parties, such
as banks or employers.

Service centers also do audit work that is discretionary, allowing
each center to identify returns with audit issues for its local
area. Each service center may apply audit resources to
discretionary programs, which can vary from year to year, as a
center finishes its mandated audit workload, such as audits
involving EIC. 3 For example, local programs may involve the
exchange of tax information with states. At least one service
center analyses a variety of database extracts for its area to
determine potentially noncompliant issues, such as the individual
retirement account penalty.

After identifying returns with audit potential, the next step in
the selection process is a review to classify returns that have
been identified as having issues worth auditing. IRS requires the
returns to be manually classified to either accept the return as
filed or to identify all potential audit issues on the return.
Identifying all potential audit issues is important because IRS
generally has one opportunity to audit a tax return.
Classification is also important in determining whether to audit
through correspondence or a face- to- face meeting.

IRS' processes of checking and requesting individual returns for
audit is shown in figure I. 1.

2 Beyond these sources, decisions about which returns to audit
involve taking into account the number and type of auditors,
existing workload by type of audit, and planned audit rate across
groups of taxpayers. The scope of our work did not cover these
decisions and factors.

3 In fiscal years 1992 and 1993, the discretionary workload was
much larger than the mandated workload. During fiscal years 1995
and 1996 (when service centers did many EIC audits), mandated
audits accounted for 75 percent of all workload. In fiscal years
1994 and 1997, discretionary and mandatory workloads were almost
equal.

Appendix I Overview of Audit Processes at Service Centers and
District Offices

Page 40 GAO/GGD-99-48 IRS' Correspondence Audits

Source: GAO summary of IRS data.

Figure I. 1: Process for Selecting Individual Tax Returns for
Audit

Appendix II IRS' Auditing Standards and Key Elements for
Correspondence Audits

Page 41 GAO/GGD-99-48 IRS' Correspondence Audits

This appendix shows IRS' auditing standards as addressed in IRM
4015.5, Exhibit 4010- 2, Center Examination Quality Measurement
Rating Guide. These standards are used to define the quality of
completed casework and to measure the accuracy and effectiveness
of the correspondence examination process.

Audit standard Key elements

All classified items considered. Scope of examination appropriate
to include all significant items for determination of tax
liability. 1. Adequate consideration of

significant items Prior- and subsequent- year returns considered.
Adequate examination and sampling techniques used, proper source
documents requested and evaluated, and taxpayer and third- party
contacts made as needed. 2. Examination depth and

conclusions reached Correct conclusion reached from developed
facts and circumstances. Examination procedures, audit trail,
findings, and conclusions fully disclosed. Workpapers are clear,
concise, legible, labeled, dated, organized, and indexed.
Adjustments in workpapers agree with examination report. 3.
Workpapers support

conclusions Examination activities adequately documented in the
case file. Appropriate report writing procedures followed. 4.
Report writing procedures followed Tax computation was correct and
used the method of

most benefit to the taxpayer. The case file reflected
consideration of and appropriate assertion or abatement of all
applicable penalties as necessary. 5. Penalties properly

considered If they were necessary, penalties computed correctly.
Time charges are relevant to the issues raised and documented in
the case file, as are other problems that affect use of time. 6.
Timely actions and time

charged appropriate All examination actions, such as requests for
documents and information were timely. Appropriate and correct
correspondence used to respond to or notify taxpayer. Statute
controls followed. Case properly assembled in accordance with
local procedures. Case accurately completed with proper forms,
closing instructions, and computer research and updates in
accordance with local procedures. 7. Case administration

Power of attorney and disclosure procedures followed. Source: GAO
summary of IRS data.

Table II. 1: IRS' Audit Standards and Key Elements for
Correspondence Audits

Appendix III Sampling and Data Analysis Methodology

Page 42 GAO/GGD-99-48 IRS' Correspondence Audits

This appendix describes how we identified the study population of
correspondence audits closed during fiscal year 1996 and our
sampling methodology. In addition, it discusses our methodology
for analyzing Individual Master File data for the tax returns in
our study population.

We identified a population of tax returns related to
correspondence audits that closed during fiscal year 1996. We
chose that year because it was the latest completed fiscal year at
the time we began our effort. Our computer analysis of IRS'
databases identified a total population of 1.1 million
correspondence audits closed during fiscal year 1996. With the
concurrence of Customer Service Compliance Branch officials, we
excluded the EIC audits with missing or invalid Social Security
numbers that are no longer in the correspondence audit program.
These officials also advised us to exclude Tax Equity and Fiscal
Responsibility Act (TEFRA) flow- through returns from our
population because these returns were actually audited in district
offices, and service centers only made adjustments to taxpayers'
accounts. By excluding the EIC math- error and TEFRA cases, our
study population of tax returns audited was 335,050. The
Compliance Branch officials told us that this population more
closely represented traditional correspondence audit tax returns.

We selected a stratified random sample of 502 audits from IRS'
AIMS file. The strata were defined by whether the taxpayer
responded to IRS' final letter and the additional taxes
recommended from the audits. These strata variables were selected
because we believed there might be a relationship between them and
the issues we were studying, such as the number of audits, the
time to do the audits, and the results of the audits.

To allocate the sample, audits were first separated into three
groups: (1) those taxpayers who responded to IRS' letters, (2)
those taxpayers who did not respond to IRS' letters, and (3) those
tax returns closed with no change in the reported tax liability.
The number of sample audits allocated to each group was
approximately proportional to the number of audits in the group.
For audited tax returns in the first two groups, the number of
audits sampled in each stratum was a compromise between a
proportional allocation of the number of taxpayers in each stratum
and a proportional allocation based on absolute dollar amounts
represented by audits in that stratum.

Within each stratum, we sorted audits by IRS service center, and
we systematically selected audits for inclusion in the sample.
This ensured that tax returns audited by all 10 service centers
were included in our Sample Selection

Methodology

Appendix III Sampling and Data Analysis Methodology

Page 43 GAO/GGD-99-48 IRS' Correspondence Audits

sample. However, our sample was not designed to statistically
project to the practices of specific service centers.

We received 446 audit files from the 502 tax returns in our
sample, or an 89- percent response rate. The difference is partly
due to the fact that some audit files were being used by IRS
projects and others could not be located. Sampling weights, which
included nonresponse adjustments, were assigned to each tax return
in our sample. As such, the 446 sampled tax returns are
statistically projectable to the 335,050 tax returns audited
nationwide during fiscal year 1996.

Because our study population results come from samples, all
results are estimates that are subject to sampling errors. We
calculated sampling errors for all of the study population results
presented in this report. These sampling errors measure the extent
to which samples of these sizes and structure can be expected to
differ from their total populations. Each of the sample estimates
is surrounded by a 95- percent confidence interval. This interval
indicates that we are 95- percent confident that the results for
the total population fall within this interval. Except where
noted, the 95percent confidence intervals for proportions do not
exceed 10 percentage points. All numeric estimates other than
proportions have sampling errors smaller than 10 percent of the
value of those estimates, unless otherwise noted.

In addition to the reported sampling errors, the practical
difficulties of conducting any case study may introduce other
types of errors, commonly referred to as nonsampling errors. For
example, differences in how a particular question is interpreted,
in the sources of information that are available, or in the types
of files not available, introduce unwanted variability into the
study population results. We included steps in our audit to
minimize such nonsampling errors. For example, we carefully
pretested the data collection instrument, discussed our study
population results with officials responsible for these audits at
the National Office and four service centers we visited, and made
follow- up requests to all of the compliance chiefs at the 10
service centers nationwide to help us accurately interpret the
data.

To assess the strength of association between the degree of
documented support and other characteristics of the audit, such as
the amount of audit time charged and the assessed amount of the
correspondence audit, we conducted correlation and regression
analyses. We also reviewed scatterplots of the relationships
between the variables. We found a relationship between the amount
of audit time charged and the degree of Response Rates

Sampling and Nonsampling Errors for Key Estimates Used in the
Report

Data Analysis

Appendix III Sampling and Data Analysis Methodology

Page 44 GAO/GGD-99-48 IRS' Correspondence Audits

documented support. However, we did not find a statistically
significant relationship between the assessed amount of a
correspondence audit and the degree of documented support of the
audit.

Appendix IV Additional Comparisons of IRS Service Center Workloads

Page 45 GAO/GGD-99-48 IRS' Correspondence Audits

This appendix shows, by IRS service center, correspondence audit
workload by number of returns, dollars recommended and direct
audit hours for fiscal years 1992 through 1997. The following
tables exclude the service center TEFRA workload because
correspondence audit officials indicated that these cases are
atypical and do not reflect the traditional correspondence audit
workload. The data come from IRS' AIMS closed case database. The
service center listing is IRS', which goes from the Northeast to
the West.

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 26,267 8 $41,490,702 6 28,465 10 Brookhaven 28,728 8
87,083,199 12 30,026 11 Philadelphia 20,062 6 40,867,034 5 11,196
4 Atlanta 42,126 12 76,690,155 10 25,958 9 Memphis 29,290 9
75,105,888 10 27,176 10 Cincinnati 31,796 9 75,996,439 10 27,285
10 Kansas City 30,640 9 91,533,907 12 28,122 10 Austin 18,422 5
53,103,406 7 16,731 6 Ogden 29,900 9 85,334,668 12 51,701 19
Fresno 81,889 24 112,859,555 15 27,084 10

Total 339,120 100 $740,064,953 100 273,744 100

Source: GAO analysis of IRS data.

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 18,657 8 $28,419,988 5 22,047 9 Brookhaven 25,855 11
70,660,954 12 32,277 13 Philadelphia 10,500 5 12,025,126 2 7,036 3
Atlanta 21,546 9 85,547,848 15 20,355 8 Memphis 20,716 9
37,425,722 6 21,240 9 Cincinnati 28,280 12 51,125,797 9 36,571 15
Kansas City 22,782 10 59,825,149 10 22,400 9 Austin 40,009 17
101,576,093 18 36,776 15 Ogden 16,031 7 41,278,786 7 30,783 13
Fresno 27,861 12 88,840,216 15 16,205 7

Total 232,237 100 $576,725,679 100 245,690 100

Source: GAO analysis of IRS data.

Table IV. 1: Correspondence Examination Workload by Service Center
for Fiscal Year 1992

Table IV. 2: Correspondence Examination Workload by Service Center
for Fiscal Year 1993

Appendix IV Additional Comparisons of IRS Service Center Workloads

Page 46 GAO/GGD-99-48 IRS' Correspondence Audits

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 23,250 7 $32,291,962 5 34,456 10 Brookhaven 26,514 8
48,404,805 8 30,974 9 Philadelphia 26,459 8 47,026,981 8 25,092 7
Atlanta 21,218 6 64,637,249 10 22,871 7 Memphis 16,635 5
21,335,944 3 28,247 8 Cincinnati 30,364 9 38,977,595 6 49,037 15
Kansas City 22,193 6 49,113,827 8 29,525 9 Austin 44,380 13
97,056,666 16 41,844 12 Ogden 24,414 7 50,020,761 8 34,576 10
Fresno 117,169 33 172,534,464 28 40,372 12

Total 352,596 100 $621,400,254 100 336,994 100

Source: GAO analysis of IRS data.

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 49,781 5 $56,199,233 4 59,808 8 Brookhaven 105,246 10
94,756,307 7 57,184 8 Philadelphia 113,970 11 132,568,342 10
61,950 9 Atlanta 60,844 6 118,095,996 9 51,800 7 Memphis 41,346 4
38,893,068 3 46,507 7 Cincinnati 54,926 5 103,448,080 8 44,518 6
Kansas City 75,131 7 97,576,557 7 47,811 7 Austin 91,562 9
112,925,152 8 72,009 10 Ogden 97,826 9 90,330,991 7 62,908 9
Fresno 363,941 35 523,562,111 38 199,354 28

Total 1,054,573 100 $1,368,355,837 100 703,851 100

Source: GAO analysis of IRS data.

Table IV. 3: Correspondence Examination Workload by Service Center
for Fiscal Year 1994

Table IV. 4: Correspondence Examination Workload by Service Center
for Fiscal Year 1995

Appendix IV Additional Comparisons of IRS Service Center Workloads

Page 47 GAO/GGD-99-48 IRS' Correspondence Audits

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 52,882 5 $72,162,081 4 61,133 8 Brookhaven 92,375 8
175,271,688 10 51,680 7 Philadelphia 93,406 8 142,672,368 8 63,367
8 Atlanta 84,519 8 167,652,022 9 77,920 10 Memphis 64,150 6
85,741,692 5 45,269 6 Cincinnati 57,164 5 155,350,831 9 52,933 7
Kansas City 85,704 8 135,828,778 8 41,258 6 Austin 135,078 12
155,873,552 9 74,807 10 Ogden 113,862 10 160,284,785 9 77,872 10
Fresno 334,506 30 515,079,489 29 202,819 27

Total 1,113,646 100 $1,765,917,286 100 749,058 100

Source: GAO analysis of IRS data.

Service center Returns Percent Additional

taxes recommended Percent Hours Percent

Andover 55,477 7 $156,542,174 7 46,582 8 Brookhaven 74,246 10
180,023,302 8 59,003 10 Philadelphia 50,561 7 141,764,403 7 37,202
7 Atlanta 43,527 6 163,020,469 8 42,868 8 Memphis 51,960 7
162,861,282 8 20,355 4 Cincinnati 60,763 8 299,380,616 14 65,066
11 Kansas City 52,220 7 156,219,970 7 29,728 5 Austin 48,987 6
97,644,618 4 45,553 8 Ogden 100,115 13 231,952,120 11 70,788 12
Fresno 219,814 29 581,062,259 27 155,811 27

Total 757,670 100 $2,170,471,213 100 572,957 100

Source: GAO analysis of IRS data.

Table IV. 5: Correspondence Examination Workload by Service Center
for Fiscal Year 1996

Table IV. 6: Correspondence Examination Workload by Service Center
for Fiscal Year 1997

Appendix V Comments From the Internal Revenue Service

Page 48 GAO/GGD-99-48 IRS' Correspondence Audits

Appendix V Comments From the Internal Revenue Service

Page 49 GAO/GGD-99-48 IRS' Correspondence Audits

Appendix V Comments From the Internal Revenue Service

Page 50 GAO/GGD-99-48 IRS' Correspondence Audits

Appendix V Comments From the Internal Revenue Service

Page 51 GAO/GGD-99-48 IRS' Correspondence Audits

Appendix VI Major Contributors to This Report

Page 52 GAO/GGD-99-48 IRS' Correspondence Audits

Sidney Schwartz, Senior Mathematical Statistician Patricia
McGuire, Senior Computer Specialist Susan Baker, Computer
Specialist Elizabeth W. Scullin, Communications Analyst

Kirk Boyer, Evaluator- in- Charge Joseph F. Lenart, Jr., Senior
Evaluator Thomas Bloom, Senior Computer Specialist Marge Vallazza,
Evaluator General Government

Division, Washington, D. C.

Kansas City Field Office

Ordering Information The first copy of each GAO report and
testimony is free. Additional copies are $2 each. Orders should be
sent to the following address, accompanied by a check or money
order made out to the Superintendent of Documents, when necessary.
VISA and MasterCard credit cards are accepted, also. Orders for
100 or more copies to be mailed to a single address are discounted
25 percent.

Order by mail: U. S. General Accounting Office P. O. Box 37050
Washington, DC 20013

or visit: Room 1100 700 4 th St. NW (corner of 4 th and G Sts. NW)
U. S. General Accounting Office Washington, DC

Orders may also be placed by calling (202) 512- 6000 or by using
fax number (202) 512- 6061, or TDD (202) 512- 2537.

Each day, GAO issues a list of newly available reports and
testimony. To receive facsimile copies of the daily list or any
list from the past 30 days, please call (202) 512- 6000 using a
touch- tone phone. A recorded menu will provide information on how
to obtain these lists.

For information on how to access GAO reports on the INTERNET, send
e- mail message with info in the body to:

info@ www. gao. gov or visit GAO's World Wide Web Home Page at:
http:// www. gao. gov

United States General Accounting Office Washington, D. C. 20548-
0001

Official Business Penalty for Private Use $300

Address Correction Requested Bulk Rate

Postage & Fees Paid GAO Permit No. G100

(268777)

*** End of document. ***