This is the accessible text file for GAO report number GAO-03-378 
entitled 'Tax Administration: IRS Should Continue to Expand Reporting
on Its Enforcement Efforts' which was released on January 31, 2003.

This text file was formatted by the U.S. General Accounting Office 
(GAO) to be accessible to users with visual impairments, as part of a 
longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov.

United States General Accounting Office:

GAO: 

Report to the Chairman, Subcommittee on Oversight, Committee on Ways and
Means, House of Representatives: 

January 2003: 

Tax Administration: 

IRS Should Continue to Expand Reporting on Its Enforcement Efforts: 

GAO-03-378: 

GAO Highlights: 

Highlights of GAO-03-378, a report to the Subcommittee on Oversight, 
Committee on Ways and Means, U.S. House of Representatives: 

Why GAO Did This Study: 

Reported declines in the rate at which the Internal Revenue Service 
(IRS) audits (also referred to as “examines”) individual income tax 
returns have raised concerns that taxpayers may have a false perception 
of the true level of IRS’s tax enforcement efforts. In addition, many 
observers are concerned these reported declines may reduce taxpayers’ 
motivation to voluntarily pay their taxes. 

Because of these concerns, GAO was asked to review a number of issues 
surrounding IRS’s enforcement efforts. GAO determined the trends in the
percent of returns filed that are audited (contact rate) compared with 
similar data on taxpayer contacts through other enforcement programs 
for fiscal years 1993 through 2002. In addition, GAO reviewed whether
IRS’s reporting on its enforcement programs should be expanded. 

What GAO Found: 

IRS’s often-cited audit rate has been declining for several years, as 
shown below. However, the audit rate portrays only a portion of IRS’s 
efforts to enforce tax laws and not all of those efforts have been 
declining. For IRS’s three nonaudit enforcement programs, the contact 
rates in 2002 compared to 1993, after year to year variations, declined 
for one, essentially remained the same for one, and significantly 
increased for one—math error. A complete math error contact trend is 
unavailable because IRS did not capture one type of data on a 
substantial number of contacts prior to 1997. For years where complete 
data are available, IRS has not included all math errors in external
reports. IRS officials agreed that all types of errors are identified 
under the same math error authority and should be similarly counted and 
reported. 

IRS annually reports extensive data on audits but only limited, or no, 
data on its other enforcement programs. This limited reporting does not 
provide policymakers or taxpayers information on the full extent of 
IRS’s enforcement efforts. To the extent that taxpayers do, as is 
widely believed, take the level of enforcement into account when self-
reporting their tax obligations, the audit rate alone may mislead them. 
IRS officials believe that more reporting is desirable and intend to 
report readily available, but incomplete, information on nonaudit 
programs in future reports. 

Figure: IRS Enforcement Program Contact Rates, Fiscal Years 1993 
through 2002: 

[See PDF for image] 

This figure is a multiple line graph depicting IRS Enforcement Program 
Contact Rates, Fiscal Years 1993 through 2002. The vertical axis of the 
graph represents Rate (percent) from 0 to 8. The horizontal axis of the 
graph represents years from 1993 to 2002. Five contact rates are 
depicted on the graph: Math error; Math error (revised); Nonfiler; 
Document matching; and Audits. 

Source: GAO analysis of IRS data. 

Note: Data for revised math error contacts is not available for fiscal 
years 1993 to 1996. The revised math error line above includes all math 
error contacts by IRS for the relevant years. 

[End of figure] 

What GAO Recommends: 

GAO recommends that IRS determine whether future reporting on its other 
enforcement programs can be more complete and comparable to reporting 
on audits. GAO also recommends that IRS correct underreporting of math
error contacts. 

In commenting on a draft of this report, IRS agreed with our 
recommendations. It has already begun to identify additional data to
report on its enforcement programs. 

[hyperlink, http://www.gao.gov/cgi-bin/getrpt?GAO-03-378]. 

To view the full report, including the scope and methodology, click on 
the link above. For more information, contact Michael Brostek at (202) 
512-9110 or BrostekM@gao.gov. 

[End of section] 

Contents: 

Letter: 

Results in Brief: 

Background: 

Scope and Methodology: 

Similarities and Differences among IRS Enforcement Programs Depend on 
the Type of Audit: 

Enforcement Program Contact Rates Did Not Follow Consistent Patterns: 

Limited Evidence Suggests IRS Enforcement Programs Do Increase 
Compliance; No Measures Available on Enforcement Program Burden; 

IRS’s Public Reporting on Its Enforcement Programs Is Incomplete: 

Conclusions: 

Recommendations to the Acting Commissioner of Internal Revenue: 

Agency Comments: 

Appendix I: Development of the Math Error Program: 

Appendix II: Use of Information Returns and Document Matching at IRS: 

Appendix III: Individual Audit and Other Enforcement Program Data: 

Appendix IV: Soft Notices and Voluntary Compliance Agreements: 

Appendix V: Comments from the Commissioner of Internal Revenue: 

Appendix VI: GAO Contacts and Staff Acknowledgments: 

GAO Contacts: 

Staff Acknowledgments: 

Related GAO Products: 

Tables: 

Table 1: Operational Dimensions of IRS Enforcement Contacts: 

Table 2: Data Published on IRS Enforcement Programs for Individual 
Taxpayers, Fiscal Year 2001: 

Table 3: Legislative Authority on Math Error Provisions for Individual 
Tax Returns: 

Table 4: Major Legislation Affecting the Information Returns Program: 

Table 5: Major Types of Information Returns Filed for Tax Years 1983 
and 2000: 

Table 6: Number and Rates of Individual Audit and Other Enforcement 
Contacts, Fiscal Years 1993 through 2002: 

Table 7: Number and Rates of Individual Audit and Document Matching 
Contacts by Income Level, Fiscal Years 1993 through 2002: 

Table 8: Number and Rates of Individual Audits by Type of Audit and 
Income Level, Fiscal Years 1993 Through 2002: 

Table 9: Average Direct Staff Hours by Type of Audit and For Document 
Matching Cases, Fiscal Years 1993 through 2002: 

Figures: 

Figure 1: Average Direct Staff Hours by Type of Audit and for Document 
Matching Cases, Fiscal Years 1993 through 2002: 

Figure 2: IRS Enforcement Contact Rates, Fiscal Years 1993 through 
2002: 

Figure 3: IRS Individual Audit and Document Matching Contact Rates by 
Income Level, Fiscal Years 1993 through 2002: 

Figure 4: Individual Audit and Other Enforcement Contacts, Fiscal Years 
1993 through 2002: 

Figure 5: Number of Individual Audit and Document Matching Contacts by 
Income Level, Fiscal Years 1993 through 2002: 

Abbreviations: 

EmTRAC: Employer-designed Tip Reporting Alternative Commitment: 

GAO: General Accounting Office: 

IRC: Internal Revenue Code: 

IRS: Internal Revenue Service: 

MSA: Medical Savings Accounts: 

NRP: National Research Program: 

RRA: IRS Restructuring and Reform Act of 1998: 

SSA: Social Security Administration: 

TIN: Tax Identification Numbers: 

TRAC: Tip Reporting Alternative Commitment: 

TRDA: Tip Rate Determination Agreement: 

[End of section] 

United States General Accounting Office: 
Washington, DC 20548: 

January 31, 2003: 

The Honorable Amo Houghton: 
Chairman, Subcommittee on Oversight: 
Committee on Ways and Means: 
House of Representatives: 

Dear Mr. Chairman: 

The United States tax system is based on self-reporting and voluntary
compliance by taxpayers. The Internal Revenue Service (IRS) uses various
enforcement programs to check the accuracy of tax returns and contacts
taxpayers if problems are found. IRS makes the contacts through four
major enforcement programs that have existed for many years. 

* Math Error Program: While tax returns are being processed, this 
program uses IRS computers to identify and generate notices to contact
taxpayers about obvious errors such as mathematical errors, omitted or
inconsistent data, or other inconsistencies on the basis of other data
reported on the return or to IRS. These errors must be corrected to
process a tax return. 

* Document Matching Program: This program matches information on
selected tax issues (usually income) reported on tax returns by 
individual taxpayers and reported on information returns by employers, 
banks, and other payers of income. Document matching also matches 
information returns (schedule K-1) filed by pass-through entities—such 
as partnerships, trusts and S-corporations—to individual tax returns. 
IRS may contact taxpayers about any reporting discrepancies. 

* Nonfiler Program: This program identifies and contacts potential
nonfilers of tax returns by using data from information returns and
previously filed income tax returns. The contacts can ask for the 
missing return or offer an IRS-generated return to substitute for the 
missing return. 

* Audit Program: Also referred to as “examination,” under this program, 
an IRS auditor checks compliance in reporting income, deductions, 
credits and other issues on tax returns, as well as in paying the 
correct tax liability. Audit contacts can be made through 
correspondence or in face-to-face meetings with taxpayers at an IRS 
office or a taxpayer location. 

Widely reported substantial declines in the rate at which IRS audits
income tax returns have triggered concern that the declines could reduce
taxpayers’ motivation to voluntarily pay their taxes. Many view IRS’s
enforcement programs as critical support to our voluntary system—they
help provide taxpayers with confidence that their friends, neighbors, 
and business competitors are paying their share of taxes. 

Because of your concerns that the declining audit rate may give 
taxpayers a misleading perception of the true level of IRS’s tax 
enforcement efforts and encourage some taxpayers not to comply, you 
requested that we review a number of issues related to how IRS enforces 
tax laws and publicly reports on those efforts. This report: 

* compares IRS’s enforcement programs in terms of their legal authority,
and operational characteristics (including IRS staff time), and 
describes what is known about taxpayer perceptions of the enforcement 
programs; 

* summarizes enforcement contact trends overall and by taxpayer income
and their causes from 1993 to 2002; 

* determines what IRS knows about the effect of its enforcement programs
on individual taxpayer compliance and the burdens taxpayers experience
when contacted under the programs; 

* assesses whether and, if so, how IRS should expand reporting on its
enforcement programs. 

To address these objectives, our work included interviewing IRS 
officials and reviewing documents on the similarities and differences 
across IRS enforcement programs. To identify trends in IRS’s 
enforcement programs including trends by taxpayer income levels, we 
analyzed IRS data from fiscal years 1993 through 2002.[Footnote 1] To 
analyze impacts on taxpayers, and tradeoffs of reporting more data 
about the enforcement programs, we interviewed IRS officials and 
reviewed any available research on how the programs affect individual 
taxpayers’ compliance and burden. (See our scope and methodology 
section for details on our approach.) In addition, you asked us to 
analyze trends in the number of contacts by the specific programs. 
Appendix III provides this information.

Results in Brief: 

Whether audits and other enforcement programs vary from each other
depends on a number of factors. With regard to legal characteristics, 
audits and other enforcement programs are all authorized to contact
taxpayers about apparent noncompliance and to determine and adjust
taxpayers’ tax liability. However, audits have the broadest authority to
detect possible noncompliance, significant powers to obtain information,
and the most restrictions on how IRS is to interact with taxpayers. With
regard to operational characteristics, the extent to which audits are
operationally similar to or different from other enforcement programs
varies depending on the type of audit. In general, audits done in 
taxpayer locations and IRS offices are not similar operationally to 
other enforcement programs. Audits done through correspondence with the
taxpayer, while still different, are more operationally similar to the 
other programs. IRS officials were unaware of any research on whether
taxpayers perceive differences among IRS’s enforcement programs. 
However, looking at audits and other enforcement programs from the
taxpayers’ perspective, IRS officials and officials we interviewed who
represent taxpayers believe that taxpayers may not perceive distinctions
among many of the enforcement programs. 

In fiscal years 1993 through 2002, the enforcement program contact rates
often did not follow consistent patterns from one program to another or
from year to year within programs.[Footnote 2] Comparing just fiscal 
years 1993 to 2002, the contact rates for the audit and document 
matching programs dropped 38 percent and 45 percent, respectively, 
while the nonfiler program contact rates stayed about the same. Only 
the math error program had a contact rate significantly higher in 2002 
than in 1993. However, this growth covers only a portion of math error 
contacts because data on one type of math error contact does not exist 
for years before 1997, and have not been reported by IRS as math 
errors. These unreported math error contacts total about 2 million 
annually. IRS officials agreed that these other contacts should be 
counted and reported as math errors. Excluding these contacts, the math 
error contact rate was 33 percent higher in 2002 than in 1993. For 
individuals, audit rates for taxpayers in higher and middle-income 
ranges were significantly lower in 2002 than in 1993, while the rate 
for the lowest income range was virtually the same in 2002 and 1993. 
Document matching program contact rates ended significantly lower in 
2002 than 1993 for taxpayers in all three income ranges. Income data 
for the contact rates in the math error and nonfiler programs were not 
available. 

The divergent trends among the enforcement programs in their contacts
with taxpayers are attributable to several factors, including statutory
changes that expanded the types of issues IRS could address with
nonaudit programs, declines in IRS enforcement staffing, and priorities 
in using staff. For example, the rise in math error contacts is at 
least partly attributable to a 1996 statutory change that enabled IRS 
to check hundreds of thousands of missing or invalid social security 
numbers through the math error program rather than audits. Declines in 
the audit program are generally attributable to statutory changes that 
reduced the availability of IRS staff to do audits and increased the 
time needed to do audits. In addition, declines in the number of staff 
assigned to work enforcement cases coupled with the priority IRS gives 
to staffing the math error program—because such errors must be resolved 
before tax returns can be processed—have contributed to declines in 
contacts with taxpayers in the audit, nonfiler, and document matching 
programs. 

IRS has limited evidence on the effects of its enforcement programs on
taxpayer compliance and no evidence on the burdens taxpayers experience 
when contacted under the programs. Although widespread agreement exists 
that enforcement programs are critical to ensuring voluntary 
compliance, IRS officials only identified one study that attempted to 
measure the effect of individual enforcement programs on compliance. 
This IRS study, using various data for 1982 through 1991, estimated 
that the audit and document matching programs had some positive effects 
on whether taxpayers filed returns and reported relevant information. 
No measures are available on the burdens placed on individual taxpayers 
due to IRS’s enforcement contacts and IRS does not currently plan to 
start any studies to measure these burdens. 

IRS’s annual public reporting on its enforcement programs for individual
taxpayers does not provide a complete perspective on its efforts to 
enforce the tax laws. IRS annually reports extensive data on audits but
limited or no data on other enforcement programs. This limited reporting
does not provide policymakers or taxpayers information on the full 
extent of IRS’s enforcement efforts. To the extent policymakers and 
taxpayers focus on audits due to IRS’s limited reporting, they may not 
understand that long-term declines in the audit rate are in part due to 
the movement of some tax issues from audits into other enforcement 
programs, and that these programs contact far more taxpayers about 
compliance issues than does the audit program. Of two options for 
expanding reporting that we identified, reporting more data on each 
enforcement program would avoid several disadvantages of combined 
reporting under an expanded definition of an audit. Expanding the 
definition of an audit would combine some enforcement activities that 
are so disparate—such as audits conducted at taxpayer locations of 
complex issues versus simple corrections of inadvertent math 
errors—that the consolidated reporting could be misleading. IRS 
officials plan to expand public reporting for fiscal year 2002 on IRS’s 
major enforcement programs to the extent that data are available and 
cost effective to extract. They do not plan to determine whether IRS 
can cost effectively develop additional data to enable future reporting 
to more completely represent program results and to facilitate program 
comparisons. 

We are recommending that IRS determine whether data can be cost 
effectively developed to make future reporting on its other enforcement
programs more complete and comparable to reporting on audits. We also
recommend that IRS correct underreporting of math error contacts. In
commenting on a draft of our report, IRS agreed with our 
recommendations and has already begun to identify additional data to
report on its enforcement programs. 

Background: 

Each year, IRS screens all individual tax returns and selects a small
percentage in which to contact taxpayers about potential noncompliance.
Prior to doing automated checks of tax returns, IRS had relied on its 
audit program to contact taxpayers about apparent inaccuracies in 
reporting income, deductions, and other issues on their tax returns. 
For example, to verify interest income or dependent exemptions claimed 
by taxpayers, IRS auditors had to contact taxpayers, request and review 
documentation. Thus, if IRS audited the returns of 5 percent of all 
taxpayers, it could at most check on the accuracy of interest income 
for 5 percent of taxpayers. 

Since the 1970s, IRS’s ability to verify some items on individual 
returns expanded as IRS’s capacity to use automated processes grew and 
as Congress enacted laws requiring third parties (like banks, mortgage
finance firms, etc.) to provide information returns to taxpayers and 
IRS on income paid. These steps enabled IRS to more universally and 
efficiently check taxpayer compliance for those tax issues covered by 
information returns. For example, with the initiation of information 
returns for interest income and the development of IRS’s automated 
capacity, IRS began to check whether every taxpayer for whom it had 
received an applicable information return had accurately reported that 
interest on their tax return. 

As a result, for some wage earners who claim no deductions, IRS can
review the accuracy of all, or nearly all, items reported on their tax 
return to the extent that third parties correctly filed all information 
returns. In these cases, IRS effectively receives information that 
should be in taxpayers “books and records” and no longer needs to use 
auditors to obtain such information from taxpayers’ records for these 
selected issues. 

Concurrent with these expansions in IRS’s ability to check the accuracy 
of certain issues on taxpayers’ returns, the number of taxpayer returns 
that IRS audited began to decline. For example, between fiscal years 
1981 and 1992, the number of document matching contacts rose from 1.2 
million to 3.8 million and the number of audits dropped from 2.5 
million to 1.1 million. 

Several GAO reports have discussed IRS audits, other enforcement
contacts, and taxpayer burden as follows: 

* In 1996, we reported that audit rates fell from 1988 to 1993 and then 
rose to a high of 1.67 percent in 1995.[Footnote 3] In 2001, we 
reported that audit rates had steadily dropped from 1996 to 2000, 
declining to 0.49 percent.[Footnote 4] 

* During 2000, we reported that IRS made almost 10 million nonaudit
contacts of taxpayers in 1998 through about 6 million math error 
notices, 2 million document matching notices, and 2 million soft 
notices. We recommended that IRS analyze the data collected for each of 
the three major nonaudit contact programs to improve taxpayer 
compliance and taxpayer service.[Footnote 5] 

* During 2000, we reported on IRS’s efforts to estimate taxpayer 
compliance burden for prefiling, filing, and postfiling activities. We 
found that IRS was developing two models that, when combined, should 
provide more reliable estimates of compliance burden for wage earners. 
[Footnote 6] 

Scope and Methodology: 

To compare audit and other enforcement programs, we obtained 
information on their legal and operational characteristics. For legal
authority, we reviewed the Internal Revenue Code (IRC) and IRS 
documents. For program operations, we reviewed IRS documents and
interviewed responsible IRS officials to understand how each program
works. For the average time spent on each type of contact, we analyzed
available IRS data for fiscal years 1993 through 2002. For how taxpayers
perceive IRS’s enforcement programs, we interviewed IRS officials;
reviewed tax research studies and press articles; and contacted four 
large national organizations representing attorneys, certified public
accountants, enrolled agents, and tax preparers, as well as the largest 
tax return preparation firm, and IRS’s national taxpayer advocate. 

To summarize trends in the number and rate of individual taxpayer audits
and other enforcement contacts in total and by taxpayer income, we
analyzed available IRS data from fiscal year 1993 to fiscal year 2002 on
each type of contact. To compute the audit contact rate, we used IRS’s
method, which equals the proportion of IRS audits closed in a fiscal 
year compared with returns filed in the previous calendar year. IRS has 
not stated a method for computing math error, document matching, and
nonfiler rates. For the document matching and nonfiler programs, we
compared the proportion of notices sent in a fiscal year to return 
filings in the previous year because these contacts generally occur in 
the year after a return is filed. For the math error program, we based 
the contact rate on the proportion of math error notices to the returns 
filed in that year because the notices are sent to taxpayers as IRS 
processes tax returns. For the math error, document matching, and 
nonfiler programs, we based the contact rate on the number of initial 
notices sent to taxpayers rather than closures because (1) it is the 
broadest measure of IRS’s enforcement efforts with taxpayers and (2) 
the math error and document matching programs usually conclude the 
contact with the taxpayer within a few months after the initial notice 
is sent. Nonfiler contacts can take considerably longer to close, 
making it difficult to know which tax year to use in computing a 
contact rate. We used the number of initial notices so that the 
nonfiler program could be measured on a consistent basis with the 
document matching and math error programs. To understand the reasons 
for the trends, we analyzed our previously issued reports and IRS 
reports and interviewed IRS staff for each enforcement program. 

To determine how audit and other enforcement programs affect individual
taxpayer compliance and burden, we obtained and reviewed available data
such as IRS studies and reports, our previous reports, and other 
research. We also interviewed responsible IRS officials. 

To assess whether IRS should expand reporting on its enforcement 
efforts, we analyzed the types of enforcement data already publicly 
reported on an annual basis by IRS. We also analyzed the tradeoffs for 
two options we identified—expanding the definition of audit to include 
the other enforcement programs and reporting more data on each program. 
We used much of the information from the previous objectives and 
interviews with IRS officials. 

For all objectives, our work focused on the four major enforcement
programs identified by IRS—math error, document matching, nonfiler, and
audit. We attempted to verify the completeness and accuracy of IRS’s 
data but could not reconcile all differences given time constraints. As 
a result, we either did not report some data or disclosed limitations 
in the data being reported. Further, in analyzing audit and document 
matching rates by income level, we did not adjust the income levels for 
the effects of inflation over the 1993 to 2002 period because detailed 
data on taxpayer income was not available during the timeframes for the 
assignment. All data used in the report are final except for the number 
of tax returns filed in 2002. Because this number is preliminary, the 
final math error contact rate for fiscal year 2002 may differ somewhat 
from what we report. In addition, you asked us to analyze two newer IRS 
efforts—voluntary compliance agreements and soft notices. Appendix IV 
describes the two newer efforts. 

We did our work at IRS’s national office in Washington, D.C., and 
offices in New Carrolton, Maryland, and Atlanta, Georgia, between 
August 2002 and December 2002 in accordance with generally accepted 
government accounting standards. We requested comments on a draft of 
this report from IRS (see app. V). 

Similarities and Differences among IRS Enforcement Programs Depend on
the Type of Audit: 

Whether audits and other enforcement programs vary from each other 
depends on a number of factors. With regard to legal characteristics,
audits and other enforcement programs are all authorized to contact
taxpayers about apparent noncompliance and to determine and adjust
taxpayers’ tax liability. However, audits have the broadest authority to
detect possible noncompliance, significant powers to obtain information,
and the most restrictions on how IRS is to interact with taxpayers. With
regard to operational characteristics, the extent to which audits are
operationally similar to or different from other enforcement programs
varies depending on the type of audit. Audits done in taxpayer locations
and IRS offices are not similar operationally to other enforcement 
programs and audits done through correspondence with the taxpayer,
while still different, are more operationally similar to some of the 
other programs. IRS officials were unaware of any research on whether
taxpayers perceive differences among IRS’s enforcement programs. 
However, looking at audits and other enforcement programs from the
taxpayers’ perspective, IRS officials and officials we interviewed who
represent taxpayers believe that taxpayers may not perceive distinctions
among many of the enforcement programs. 

Legal Characteristics: Audits Have the Broadest Scope, Significant 
Powers to Obtain Information, and the Most Detailed Restrictions: 

In a general sense, the IRC provisions for enforcement are similar in 
that they authorize IRS to contact taxpayers about apparent 
noncompliance and to determine and adjust taxpayers’ tax liability. 
However, the IRC provisions grant IRS the authority to review all 
matters that may affect a taxpayer’s tax liabilities under audits but 
only certain specified tax issues under other enforcement programs. The 
IRC also establishes more rules—including significant powers to obtain 
information as well as restrictions on those powers—that govern the 
nature of audit contacts with taxpayers than for the other programs. 

The IRC does not explicitly limit the tax issues to be covered by an 
audit, unlike for the other enforcement programs. Under the authority 
of section 7602, audits can cover any issue on a tax return, including 
those that the other programs cover.[Footnote 7] In contrast, the IRC 
specifies the scope of legal authority for the three other enforcement 
programs. For example, after five statutory expansions since 1976, math 
error authority[Footnote 8] now covers 11 tax issues (see app. I). 
Document matching—which grew primarily through the 1980s as Congress 
authorized more information reporting—now covers over 20 types of 
individual income as well as certain tax credits and deductions (see 
app. II). [Footnote 9] The IRC also specifically authorizes the 
nonfiler program to pursue unfiled tax returns that should have been
filed.[Footnote 10] 

The IRC also establishes more rules governing IRS’s contacts with
individual taxpayers under the audit program than it does for the math
error, document matching, and nonfiler programs. These rules give IRS
significant powers to obtain information needed to determine an
individual’s tax liabilities when doing an audit and, in turn, places
restrictions on the use of those powers. If resolving issues raised 
under the other enforcement programs requires that IRS auditors become 
involved, the contacts with taxpayers become audits subject to these 
greater powers and restrictions. For example, if a taxpayer who 
receives a math error notice files a claim for IRS to abate the tax 
assessment, IRS could audit that claim. Similarly, if a taxpayer 
responds to a document matching notice with materials that cannot be 
readily and immediately used to settle the discrepancy, the case could 
be referred to audit staff. 

The greater powers that IRS has under audit compared with the other
programs include the authority to examine books and records and take
testimony for purposes of determining the tax liability of a tax 
return. IRS also has the power to use a summons to compel taxpayers and 
third parties to provide books and records, and to enter premises to 
examine objects subject to taxation.[Footnote 11] 

Given these greater powers, the law also places more restrictions on
audits to protect taxpayer rights.[Footnote 12] For example, the law 
restricts IRS from doing unnecessary audits or generally doing more 
than one inspection of taxpayers’ books for each tax year. The law also 
governs the time and place of an audit and burden of proof on IRS. In 
addition, the Internal Revenue Service Restructuring and Reform Act 
(RRA) of 1998 (P.L. 105-206) added several requirements, such as 
informing taxpayers of their rights during audits.[Footnote 13] 

On the other hand, RRA also added a provision that creates a legal
similarity for all four enforcement programs because it affects any IRS
employee, including those making audit or nonaudit contacts. Section 
1203 of RRA lays out the conditions under which any IRS employee is to 
be fired for any of 10 specific acts or omissions. Many of these 
conditions involve contacts with taxpayers—such as harassing taxpayers 
or taxpayer representatives, violating their civil rights, or 
threatening to audit a taxpayer for personal gain. These restrictions 
were intended to protect taxpayers in their interactions with IRS. 

Another legal provision creates a similarity between audits and two of 
the three other programs. Except for the math error program, when IRS
proposes a change in taxpayers’ liabilities, it is required to send a 
notice[Footnote 14] informing taxpayers of their rights, such as the 
right to appeal additional taxes that IRS proposes. The IRC does not 
provide taxpayers a right to appeal assessments created under math 
error authority because that authority generally applies to obvious 
errors made by taxpayers on their returns. However, IRS informs 
taxpayers receiving a math error notice that they may file a claim to 
ask IRS to abate (reduce) the tax assessment if they believe IRS erred. 

Operational Characteristics: Audits at IRS Offices or Taxpayer 
Locations Differ the Most from Other Enforcement Programs: 

The extent to which audits are operationally similar to or different 
from other enforcement programs varies depending on the type of audit.
Compared with other enforcement programs, audits done in taxpayers’
locations or IRS offices are more likely to deal with multiple and 
complex issues, require more skill and judgment by IRS employees, 
require a greater number of interactions with taxpayers, and take more 
IRS staff time. Correspondence audits also tend to differ from other 
enforcement programs in these operational characteristics but to a 
lesser degree, and in some cases correspondence audits and document 
matching contacts with taxpayers can be very similar in these 
characteristics. Table 1 provides an overview of key operational 
dimensions across the enforcement programs. 

Table 1: Operational Dimensions of IRS Enforcement Contacts: 

Contact: Audits; 
What triggers the contact: 
* computer analysis of potentially noncompliant returns; 
* referrals from inside or outside IRS; 
* projects on specific areas of known noncompliance; 
How many contacts occur between IRS and the taxpayer: likely requires 
multiple exchanges via notices/letters, telephone, and/or face-to-face 
meetings; 
How much skill and judgment is required by IRS staff: skilled review of
simpler issues and more sophisticated analysis of complex issues; 
When is the initial contact sent to the taxpayer: usually within 1 year 
after the return is filed but may occur later as long as IRS finishes 
the audit within 3 years after the return is filed. 

Contact: Document Matching; 
What triggers the contact: computer identification of error based on 
information received from third parties on income paid; 
How many contacts occur between IRS and the taxpayer: one notice from 
IRS and possible exchanges via letters or telephone; 
How much skill and judgment is required by IRS staff: some 
skill/judgment; 
When is the initial contact sent to the taxpayer: within 1 year after 
the return was filed. 

Contact: Non-filer; 
What triggers the contact: 
* computer identification of those who did not file or who stopped
filing; 
* referrals from inside or outside IRS; 
How many contacts occur between IRS and the taxpayer: one notice from 
IRS and probable exchanges via letters, telephone, or meetings; 
How much skill and judgment is required by IRS staff: some 
skill/judgment; 
When is the initial contact sent to the taxpayer: usually within 1 year 
after the return was to be filed. 

Contact: Math Error; 
What triggers the contact: computer identification of error using 
taxpayer information on their tax returns or forms; 
How many contacts occur between IRS and the taxpayer: one notice from 
IRS; 
How much skill and judgment is required by IRS staff: little 
skill/judgment; 
When is the initial contact sent to the taxpayer: as part of the initial
processing for the tax return. 

Source: GAO analysis. 

Note: Of the three types of audits, the simplest usually covers one to 
two tax issues handled by a lower-graded auditor and correspondence. 
More complex audits are done by meeting with taxpayers in IRS field 
offices. The most complex audits are done through field visits to 
taxpayer locations. 

[End of table] 

Reviewing these operational dimensions helps highlight similarities and
differences across the four types of enforcement. 

* Contact triggers: All enforcement contacts use computers to identify a
potential compliance issue. However, audits are more likely to be
triggered by other means such as a special compliance project or 
referrals from inside or outside of IRS. 

* Number of contacts: Once any potential compliance issues are found, 
the fewest contacts with taxpayers to resolve the issues are likely 
under the math error and document matching programs because they have 
relatively simple issues. Correspondence audits might need more than 
one contact, depending on the complexity of the issue(s) being audited 
and taxpayers’ responses. The number of contacts in the nonfiler 
program can vary depending on whether taxpayers respond to an IRS 
notice by filing a return, or explaining why a return was not 
required.[Footnote 15] Some may not respond, possibly leading IRS to 
send a second notice or create a substitute return and send it to that 
taxpayer. Audits at IRS offices or taxpayer locations are likely to 
have the most contacts with taxpayers through meetings, notices, or the 
telephone because they tend to cover several, more complex issues. 

* IRS staff skill and judgment: Audit contacts—especially those done in
IRS offices or in taxpayer locations—require the most staff skill and
judgment to analyze taxpayers’ testimony and books and records. Being
more automated and usually dealing with simpler issues, other 
enforcement programs rely on less staff skill and judgment. Document
matching staff might have to analyze taxpayers’ explanations for why 
they do not owe more tax but are to refer the case to the audit program 
if an explanation requires detailed analysis or includes books and 
records. The nonfiler program requires limited skill and judgment when 
automated processes send the notices or generate substitute returns. 
More skill and judgment is required when IRS staff manually create 
substitute returns or when taxpayers respond to a notice by saying that 
they do not have to file a return. 

* Timing of initial contact: Math error contacts are made as the return 
is being processed and identify errors that must be corrected to finish
processing the return. Document matching and nonfiler contacts usually
occur within 1 year after the return is filed or is to be filed. Audits 
usually start within 1 year after a return is filed but can start later 
as long as IRS finishes the audit within 3 years after the return is 
filed. Another operational characteristic is the average time spent by 
IRS staff. Figure 1 shows that audits use more staff time per case than 
document matching contacts.[Footnote 16] For fiscal years 1993 through 
2002, the staff time ranged from roughly an hour per document matching 
case to up to 30 hours per field audit. (see Table 9 in app. III for 
details.) IRS does not separately track the time spent on math errors 
from the rest of the returns filing process or on nonfilers from other 
work done by collection staff. 

Figure 1: Average Direct Staff Hours by Type of Audit and for Document 
Matching Cases, Fiscal Years 1993 through 2002: 

[See PDF for image] 

This figure is a multiple line graph depicting the average direct staff 
hours by type of audit and for document matching cases, fiscal years 
1993 through 2002. The vertical axis of the graph represents Hours from 
0 to 35. The horizontal axis of the graph represents years from 1993 to 
2002. Four types of audits are depicted by lines: Field audit; Office 
audit; Correspondence audit; and Document matching. 

Source: GAO analysis of IRS data. 

Note: Fiscal year 2002 data are estimated. 

[End of table] 

Looking across all of these operational dimensions in general, audits 
that take place in IRS offices or in taxpayer locations differ the most 
from other enforcement programs. They differ primarily because they are 
more likely to deal with multiple complex tax issues, require more 
skill and judgment by IRS employees, require a greater number of 
interactions with taxpayers, and take more IRS staff time. Although 
correspondence audits do differ from other enforcement programs on 
these characteristics, they do not differ as much from other 
enforcement programs as do the audits in IRS field offices or taxpayer 
locations. The closest similarity between correspondence audits and the 
other programs is with the document matching program. In comparison to 
the document matching program, correspondence audits in some cases may 
deal with about the same number of issues, have the same number of 
interactions with taxpayers, and require similar skill, judgment, and 
time on the part of IRS staff. Correspondence audits are less similar 
to contacts under the math error and nonfiler programs than to document 
matching. Math error contacts deal with straightforward issues that 
must be corrected as a return is processed as opposed to contacting the 
taxpayer up to one year later about issues in the return. Nonfiler 
contacts deal with taxpayers who have not filed a return at all as 
opposed to seeking to correct issues related to a filed return. 

In addition to the four major enforcement programs, starting in the mid-
1990s, IRS created two new programs intended to help individual 
taxpayers file accurate tax returns. IRS sends so-called “soft notices” 
on duplicate claims for dependent exemptions and missing self-employment
tax reporting. The soft notices do not require taxpayers to take action 
but are intended to educate them about the potential errors and 
encourage them to correct their returns, if necessary. The other new 
program is the voluntary compliance agreements program. These 
agreements are negotiated with certain employers with the goal of 
increasing their employees’ compliance in reporting tip income. As 
discussed in appendix IV, while these programs attempt to improve 
compliance, they have significant differences from the four major 
enforcement programs and IRS has little data on their use. However, IRS 
was able to provide us with data that it sent taxpayers 1.2 million 
soft notices on duplicate dependent claims in 2002. 

No Research Identifies Whether Taxpayers Perceive Distinctions in 
Enforcement Programs but Officials Believe That Distinctions May Not Be 
Made: 

IRS officials were not aware of any research, and our search of the tax
literature and press did not uncover research, on whether taxpayers
perceive distinctions between audits and other enforcement programs. Of
the four major enforcement program contacts, IRS officials said that 
they could see how some taxpayers might view two types of contacts—
document matching and correspondence audits—as similar in that both
tend to cover one or two tax issues that are fairly simple, contact
taxpayers through the mail, and give taxpayers the same appeal rights.
Otherwise, these officials did not see how taxpayers could view the
enforcement contacts as similar, especially the math error contacts. 

Although they had not surveyed taxpayers, officials we interviewed from
six groups that represent taxpayers or help prepare their tax returns
believed that many individuals perceive no distinction among the 
programs. For example, one representative attributed this to the 
conclusion that taxpayers view all IRS notices as stating the same
thing—that the taxpayer owes more taxes. 

Enforcement Program Contact Rates Did Not Follow Consistent Patterns: 

From fiscal years 1993 through 2002, the rates for the four enforcement
programs often did not follow consistent patterns from one program to
another or from year to year within programs. Comparing just 1993 with
2002, the contact rates for two programs—audits and document 
matching—were significantly lower, the rate for math errors was 
significantly higher, and the rate for nonfilers was essentially the 
same. By taxpayer income level, the audit rate for higher and middle 
income taxpayers generally declined over the 10-year period—with the 
sharpest declines for higher income taxpayers. The rate for the lowest 
income taxpayers increased sharply between 1993 and 1995 and then 
generally fell, ending virtually the same as in 1993. The document 
matching contact rate by income class followed very similar patterns 
with the rates for all income levels dropping over the 10-year period. 
The enforcement contacts increased or decreased because of several 
reasons, including statutory changes, staffing declines, and priorities 
in the use of staff among the programs. 

Enforcement Program Contact Rates Varied from Program to Program and 
Often from Year to Year within Programs: 

As figure 2 shows, the math error program contact rates rose or fell 
from year to year; however, it’s the only enforcement program that had a
significantly higher contact rate in fiscal year 2002 than in 1993. 
This is true even without counting certain math error contacts for 
which IRS lacks data over the 10-year period. Using only the math error 
count, which is consistent throughout the 10 years, the math error 
contact rate rose 33 percent (from 3.59 percent to 4.79 percent). 
Document matching contact rates went down and up at various times but 
ended 45 percent lower (from 2.37 percent to 1.30 percent) in 2002 
compared to 1993. The nonfiler rates also went up and down but ended in 
2002 about where they were in 1993. Comparing 1993 to 2002, the audit 
contact rate dropped 38 percent (from 0.92 percent to 0.57 percent), 
even though it rose significantly between 1993 and 1995. Over the 10 
years, the math error rate exceeded the rate for each of the three 
other programs, and the audit rate was the lowest rate, except in 
fiscal years 1995, 1996, and 1997. The trends in the number of contacts 
in all four programs generally follow the trends in the rates. Appendix 
III provides details about the contact numbers and rates for all four 
programs. 

Figure 2: IRS Enforcement Program Contact Rates, Fiscal Years 1993 
through 2002: 

[See PDF for image] 

This figure is a multiple line graph depicting IRS Enforcement Program 
Contact Rates, Fiscal Years 1993 through 2002. The vertical axis of the 
graph represents Rate (percent) from 0 to 8. The horizontal axis of the 
graph represents years from 1993 to 2002. Five contact rates are 
depicted on the graph: Math error; Math error (revised); Nonfiler; 
Document matching; and Audits. 

Source: GAO analysis of IRS data. 

Notes: The math error (revised) line includes math error contacts from 
masterfile notices that IRS’s reports had excluded. Data on such math 
errors were not available prior to fiscal year 1997. The figure does 
not include additional contacts for math errors related to the rate 
reduction credit during 2002. 

Fiscal year 2002 data is estimated. 

[End of figure] 

The trend line in Figure 2 shows a revised math error contact rate that
includes masterfile[Footnote 17] notices IRS had been sending 
throughout this period but had not been reporting as math errors. In 
data made publicly available on math error contacts,[Footnote 18] IRS 
had excluded roughly 2 million masterfile math error contacts annually 
for fiscal years 1997 through 2002.[Footnote 19] When included, the 
math error contact rate increases (e.g., from 4.97 percent to 6.5 
percent in 1997). These math error contacts arise from IRS’s match of
filed tax returns to its masterfile accounts to identify tax returns 
that misreport taxes already paid such as in previous years or 
estimated tax payments. IRS officials said that the masterfile errors 
were not reported as math errors because they are identified through a 
different process at a later time compared to other math errors during 
the processing of tax returns. In our discussions, these officials 
agreed that both types of errors are identified under the same math 
error authority, are indistinguishable to taxpayers being contacted, 
and should be similarly counted and reported. 

IRS also did not include in its published report about 8 million 
notices sent in fiscal year 2002 to correct errors in tax returns 
reporting the rate reduction credit.[Footnote 20] If these notices had 
been included, the math error contact rate would have nearly doubled to 
12.5 percent. The Economic Growth and Tax Relief Reconciliation Act of 
2001 authorized tax rate reductions as well as an advance tax refund, 
called the rate reduction credit. Because the rate reduction credit 
applied for only 1 year, this error is unlikely to recur according to 
IRS officials. As a result, we did not include this information in 
figure 2. 

Audit and Document Matching Contact Rates Across Income Levels Have 
Generally Declined: 

Figure 3 shows that the contact rates generally declined in the audit 
and document matching programs for all taxpayer income levels between
fiscal years 1993 and 2002. For the math error and nonfiler programs, 
data on contact rates by income level were not available, and IRS 
officials said that it would take some time and effort to develop math 
error and nonfiler contact rates by income levels. (See table 7 in app. 
III for details on contact rates by income levels.) 

Figure 3: IRS Individual Audit and Document Matching Contact Rates by 
Income Level, Fiscal Years 1993 through 2002: 

[See PDF for image] 

This figure contains two multiple line graphs depicting IRS Individual 
Audit and Document Matching Contact Rates by Income Level, Fiscal Years 
1993 through 2002. The vertical axis of the graphs represent Audit 
rates from 0 to 5 percent (first graph) and Document matching rates 
from 0 to 5 percent(second graph). The horizontal axis of both graphs 
represents years from 1993 to 2002. Lines depicting the following three 
income levels are represented on each graph: $200,000 and over; $25,000 
to under $100,000; and Under $25,000. 

Source: GAO analysis of IRS data. 

Note: Data by income level for the math error and nonfiler programs are 
not available. 

[End of figure] 

As figure 3 shows, because the audit contact rate declined (from 3.89 
percent to 0.86 percent) for higher income (more than $100,000)
individuals and remained virtually the same (from 0.77 percent to 0.78 
percent) for the lowest income (less than $25,000) individuals between 
fiscal years 1993 and 2002, the rates for the highest and lowest income 
individuals essentially converged in 2001 and 2002. Over the same time, 
the document matching contact rate generally declined for all three 
income groups with fairly similar year-to-year patterns and with higher
income individuals being contacted at the higher rate. 

Statutory Changes, Fewer Staff, and Resource Priorities Explain Trends 
in Math Error, and Other Enforcement Contact rates: 

The divergent trends between the growing rate of math error contacts and
the declining or relatively stable rates for the other enforcement 
programs can be attributed to how the programs have been affected by 
statutory changes, fewer enforcement staff, and priorities for using 
available staff. Math error contacts grew over the fiscal year 1993 
through 2002 period in part because Congress expanded the types of tax 
issues covered by the math error authority. For example, in 1996, 
Congress gave IRS authority to shift a number of earned income tax 
credit issues from its audit program to its math error program in that 
year. As a result, in 1997, IRS shifted over 700,000 cases involving 
missing or invalid social security numbers (SSN) on tax returns from 
the audit program to the math error program. Under this and other 
statutory expansions, IRS was making hundreds of thousands of math 
error contacts with taxpayers by 2002 that were not made in 1993. 

A second statutory change played a role in the diverging trends among 
the enforcement programs. In RRA, Congress took steps to better ensure 
that taxpayer rights were protected by revising certain audit 
processes, such as informing taxpayers about their rights and generally 
how they were selected for audit. According to IRS officials, the 
changes contributed to the decline in audits because IRS auditors had 
to spend more time to handle nonaudit duties, to be trained in new 
procedures and taxpayers’ enhanced rights, and to do new tasks. 
[Footnote 21] Those changes contributed to reductions in the number of 
audits that each auditor completed, meaning they were less productive 
in closing audit cases. 

Finally, declines in enforcement staffing and priorities for using 
staff also contributed to trends in enforcement program contacts. IRS 
has reported that from 1993 to 2001, enforcement staffing levels 
declined about 24 percent. These staffing declines affected not only 
the audit program but also the document matching and nonfiler programs 
because those programs require that IRS staff screen most notices 
before they are sent and follow up when taxpayers respond to 
notices.[Footnote 22] Given declining staff resources, IRS has 
restricted the number of notices sent when it finds probable 
noncompliance under the document matching and nonfiler programs. In 
contrast, IRS allocated enough resources over this period to the math 
error program to continue sending these notices. IRS officials said 
that IRS must resolve math errors to process tax returns and adjust the 
tax liability so that taxpayers are in compliance.[Footnote 23]

Limited Evidence Suggests IRS Enforcement Programs Do Increase 
Compliance; No Measures Available on Enforcement Program Burden: 

Although widespread agreement exists that enforcement programs help
ensure voluntary tax compliance, evidence is limited about the degree to
which enforcement overall or by type of program affects taxpayer
compliance. No studies are available that measure the burdens that
taxpayers experience when contacted under IRS’s enforcement programs. 

Over the years, many tax practitioners and academics have suggested that
enforcement programs are critical for ensuring voluntary compliance.
However, measuring the effects of enforcement programs on compliance
is a difficult task. IRS officials identified only one study that 
attempted to estimate the effects of its enforcement programs on 
compliance; no more recent work is underway or planned to measure these 
effects. Relying on an econometric analysis of taxpayer behavior—using 
various assumptions, IRS and non-IRS data for 1982 through 1991, and 
alternative measures of compliance—this IRS study estimated the effects 
of various IRS programs across the general taxpayer population. The 
study suggested that audits had a positive impact on compliance in 
reporting information on tax returns and that document matching had a 
positive effect on compliance in filing required returns. We did not 
have time to analyze the reasonableness of the study’s approach, 
assumptions, and results. 

To obtain current information on taxpayers’ compliance in filing tax
returns and reporting correctly on them, IRS developed its National
Research Program (NRP).[Footnote 24] This program is designed to yield 
reliable estimates of the compliance levels of individual taxpayers 
while addressing concerns about the burden such a measurement program 
can impose on taxpayers. NRP’s design was completed in fiscal year 
2002, and IRS will be auditing taxpayers’ returns under the program 
during fiscal year 2003.[Footnote 25] IRS plans to use the NRP results 
to update tools to select individual tax returns for audit, to allocate 
resources, to estimate the impacts of legislative and administrative 
changes on voluntary compliance and tax revenue, and to identify 
potential ways to improve voluntary compliance. Although NRP should 
yield useful data, it was not designed to measure the effect that each 
major enforcement program could have on voluntary compliance. 

In addition, IRS has been working to produce more comprehensive 
estimates of burden that individual taxpayers face in meeting their tax
obligations. IRS developed a system in 1984 for estimating the burdens
taxpayers face in filing IRS forms, and began efforts during the 1990s 
to create a better model for estimating such compliance burdens. IRS
recently announced that the new burden model is ready to be tested and
likely will replace the old model during fiscal year 2003. Although the
model should provide better estimates of individual taxpayer burdens in
completing and filing tax returns, it is not designed to estimate the
postfiling burdens related to IRS’s enforcement efforts. IRS expects to
model these postfiling burdens but does not yet know when that phase of
its burden estimation project will begin. 

IRS’s Public Reporting on Its Enforcement Programs Is Incomplete: 

IRS’s public reporting on its enforcement programs for individual 
taxpayers does not provide a complete perspective on its efforts to
enforce tax laws because that reporting heavily focuses on audits. IRS’s
audit rate is often cited in the press and is often the focus of 
congressional and other debates concerning how well IRS is enforcing 
the tax laws. However, over time the audit rate has become increasingly 
less complete as a measure of IRS’s efforts to enforce tax laws because 
IRS’s other enforcement programs have expanded their coverage of issues 
once covered under audits. 

At least two options exist for expanding reporting: changing the 
definition of audits to include other enforcement efforts and reporting 
more data on each enforcement program separately. The second option 
would achieve more complete and balanced reporting without incurring 
some of the disadvantages that could come from expanding the definition 
of audits. IRS officials plan to expand public reporting for fiscal 
year 2002 on IRS’s major enforcement programs to the extent that data 
are available and cost effective to extract. 

Limited Public Reporting on IRS’s Enforcement Programs: 

IRS publishes extensive data on audits but only limited data on other
enforcement programs in its Data Book.[Footnote 26] Table 2 summarizes 
the data annually published on the enforcement programs involving 
individual taxpayers. 

Table 2: Data Published on IRS Enforcement Programs for Individual 
Taxpayers, Fiscal Year 2001: 

Enforcement program: Audit; 
IRS published data: Number and/or rate of audits by: 
* type of tax (e.g., individual income, gift); 
* type of return filed (e.g., Form 1040, Form 1040A); 
* taxpayer income class; 
* type of audit (e.g., field, correspondence); 
* type of auditor (e.g., tax auditor, revenue agent); 
* whether the tax liability changed; 
* total and average amount of recommended additional tax; 
* whether taxpayers agreed with the recommended tax change; 
Selected data on audits: 
* resulting in tax refunds; 
* preventing tax refunds on the basis of taxpayers’ efforts to recoup 
taxes previously assessed or paid. 

Enforcement program: Document Matching; 
IRS published data: 
* number of information returns received; 
* number of taxpayer contacts; 
* amount of additional tax assessed. 

Enforcement program: Nonfiler; 
IRS published data: 
* number of taxpayer delinquency investigations completed; 
* number of initial nonfiler notices sent; 
* additional assessments for the substitute for returns program. 

Enforcement program: Math error; 
IRS published data: No data are published. 

Source: IRS Data Book, fiscal year 2001. 

[End of table] 

As shown, IRS publishes no data on IRS’s math error program—which
affects millions of individual taxpayers annually. Compared with audits,
public reporting on the document matching and nonfiler programs is much
more limited. IRS officials said that IRS publicly reports more data on
audits because IRS has had a separate audit case-tracking system for 
many years that produces such data. Also, they said that requests to 
publicly report more data on the other programs had not been made. 

IRS has not changed Data Book reporting on its enforcement programs to
keep up with changes in its enforcement programs over the years. With 
its focus on audits, the reporting may lead others to focus on audits 
and thereby to have an incomplete understanding of IRS’s enforcement 
efforts. For example, trends in the audit rate alone are difficult to 
use to assess IRS’s enforcement presence because that rate does not 
measure the same phenomenon today as it did earlier. Even within the 10-
year period we reviewed, some of the tax issues that formerly had been 
checked only under the audit program migrated into the other 
enforcement programs. This type of migration was more pronounced in the 
1980s as the document matching program expanded substantially. 

Although the scope of what IRS does under audits has changed 
considerably over the past few decades, and even within the past 10 
years, the audit rate remains an often-cited statistic when Congress 
and others consider how well IRS is enforcing the tax laws. For 
instance, during annual oversight hearings on IRS’s performance, 
members of Congress often raise questions about changes in the audit 
rate. Over the past several years, these hearings have included 
concerns about the declining audit rate and its possible affect on 
taxpayers’ compliance. The IRS Commissioner also expressed concern 
about the decline in audits. However, the Commissioner said that he did 
not believe the audit rate needed to increase to the same level as a 
number of years ago because IRS has other programs to enforce the tax 
laws that were not available, or as broad in scope, in past years. 

To the extent that IRS’s audit rate is the major source of information
available to taxpayers on IRS’s enforcement efforts, the public cannot 
be fully aware of the extent to which IRS enforces tax laws and thus may
misjudge the chances that noncompliance is likely to be detected. 
Taxpayers who are aware only of the audit rate would not be aware that
IRS often contacts more taxpayers under each of its other enforcement
programs—and IRS always contacts far more taxpayers in these other
programs combined—than it does under the audit program. As discussed
earlier, although the degree to which enforcement encourages voluntary
compliance is difficult to measure, it is widely believed that public
knowledge about enforcement efforts helps prompt higher levels of
voluntary compliance. Although he did not specifically cite possible
increases in voluntary compliance, in a letter issued in March 2001, 
the IRS Commissioner said that only focusing on audits substantially 
understates IRS’s capacity to find errors.[Footnote 27] 

While greater awareness of the scope of IRS’s efforts to enforce the tax
law may encourage compliance, it could also increase taxpayers’ 
awareness of the trends in these efforts. It is not clear how taxpayers
would interpret and react to the differing trends among IRS’s 
enforcement programs. For the period from 1993 through 2002, trends in 
IRS’s individual enforcement programs often varied from year to year as 
well as between the programs. Therefore, the compliance signals to 
taxpayers from publicizing data on the trends in these other programs 
probably would be different—and more mixed—than the signal they receive 
based exclusively on the audit rate. 

In addition, awareness of the fuller range of IRS’s enforcement efforts 
may not affect compliance of all groups of taxpayers equally. This 
could occur, for example, when the contact rates under the enforcement 
programs differ, as they do under the audit and document matching 
programs for different income groups. Further, to the extent that 
taxpayers know that IRS can only understand their tax situation through 
a traditional audit, their compliance might be less affected by fuller 
reporting on IRS’s other enforcement efforts. The IRS Commissioner has 
said that the decline in the traditional audit rate is of concern in 
part because a growing portion of taxpayers and a growing amount of 
income is not well identified through such programs as document 
matching and nonfiler. 

Expanding Public Reporting on Enforcement Programs Does Not Require 
Redefining Audits: 

Of two options we identified for expanding public reporting on IRS’s
enforcement efforts, providing data on each major program separately
avoids certain disadvantages of aggregating data into one broad audit
program. After we discussed the tradeoffs of these options with IRS
officials, they said they plan to expand public reporting for each of 
the nonaudit enforcement programs. 

One option for expanding reporting on IRS’s enforcement programs would
be to define all of IRS’s enforcement programs to be audits for 
statistical reporting. If the programs were all defined to be audits, 
IRS might report a consolidated “audit rate” that would represent all 
of IRS’s contacts with taxpayers. Consolidated reporting might also be 
done on such things as the additional tax revenues identified through 
the contacts and the staff time invested by IRS. 

This option could have several advantages. For instance, it would 
provide for more complete reporting on IRS’s overall enforcement 
efforts in a single “rate.” Another advantage to expanding the 
definition of an audit is that the major enforcement programs have an 
overall similarity in what they intend to achieve. Moreover, some 
document matching and math error checks now cover some tax issues that 
had been covered under audit authority. Thus, because audits do not 
measure the same thing over time, expanding the definition would create 
a more consistent measure of the extent to which IRS is enforcing tax 
laws. 

However, combining all enforcement programs under one definition poses
a number of potential disadvantages. For example, IRS’s legal authority
and operational rules, as well as taxpayers’ rights, vary across
enforcement programs. If all enforcement programs were called audits,
IRS staff and taxpayers could become confused about the rights and
restrictions that govern contacts with taxpayers. Labeling all 
enforcement programs as audits might confuse taxpayers about whether 
IRS could examine their books and records for a specific tax year (an 
action taken under IRS’s current audit authority) if they had already 
been contacted under document matching and/or math error programs. 

If all enforcement programs were called audits and aggregate reporting
was done, IRS would face a challenge in ensuring that taxpayers and
others are not misled. For example, a single audit rate would cover the
range from intense audits covering multiple tax issues to the 
correction of simple math errors arising from inadvertent 
miscalculations by taxpayers. Given the higher number of math errors 
being detected by IRS over time, if taxpayers interpreted a revised 
audit rate as representing the former rather than the latter situation, 
they would be misled about IRS’s true level of tax return scrutiny. 
Another challenge for IRS would be in reporting audit results like tax 
dollars assessed and time spent per audit. Considerable variability 
already exists in these results for audit—e.g., field audits take 
significantly more staff hours than correspondence audits. These 
differences would be more extreme under a consolidated audit reporting 
system that included document matching, nonfiler, and math error 
contacts. Finally, the IRS would need to account for potential double 
counting because taxpayers can be contacted through more than one
enforcement program for the same return. [Footnote 28] 

In addition, labeling all IRS enforcement programs as audits might 
suggest that the programs are in some sense substitutable in detecting
noncompliance and encouraging voluntary compliance. Although the
document matching and nonfiler programs do replace part of what had
previously been done by auditors, these programs do not completely
substitute for audits. Math error program contacts are even less of a
substitute for audit. Combining all of these efforts suggests an
equivalence—one math error contact with a taxpayer is equivalent to a
complex, intense audit of a taxpayer books and records—that is not
correct. Therefore, if audits dropped even further than they have in 
recent years, but math error contacts rose even faster, some might 
assume that IRS is doing better at enforcing tax laws while others 
might disagree. 

Because of such disadvantages, IRS officials said that they do not favor
changing the audit definition to include the other enforcement programs 
at this time. Specifically, they said any changes would create 
confusion about IRS’s enforcement activities and could distort any 
comparisons because the programs significantly differ. 

Instead of expanding the audit definition, IRS has already expressed
support for greater reporting on the full range of IRS’s enforcement
efforts. For example, in 2001, the IRS Commissioner stated that IRS’s 
goal is to make public reporting on nonaudit enforcement efforts as
informative and meaningful as possible.[Footnote 29] This approach 
generally avoids the disadvantages associated with reporting IRS’s 
enforcement efforts under one consolidated, redefined audit program. At 
the same time, it would provide more complete reporting to the public. 

In December 2002, IRS officials told us that they plan to try to report 
more data on other enforcement programs to the extent that the data are
available and cost-effective to extract. Officials expect that this 
expanded reporting will begin with the fiscal year 2003 Data Book if 
the necessary statistical tables cannot be produced in time for the 
2002 edition that is to be published in early calendar year 2003. The 
expanded information will also be available on the IRS Web site. 

These officials said that they would attempt to report the number of 
cases closed, the staff time expended, and the tax amounts adjusted for
document matching and for the automated substitute for return program
(ASFR) component of the nonfiler program. For document matching, IRS
plans to account for not only the cases in which taxpayers were 
contacted but also in which IRS staff resolved the apparent income 
discrepancy without contacting taxpayers. For ASFR, IRS is planning to 
adjust the data for cases in which IRS abated the additional tax 
amounts assessed after taxpayers later filed a tax return. For both 
programs, IRS officials said that reporting the data by the taxpayer’s 
level of income is doubtful. For math errors, IRS officials said that 
they could report the number of notices, staff time, and tax amounts 
assessed but that reporting other data is questionable either because 
the data are not collected or are difficult or costly to extract. IRS 
has no plans to analyze whether changes could be made to cost-
effectively extract or collect other data to facilitate understanding 
of and comparisons among these nonaudit enforcement programs. 

Conclusions: 

Although research is not conclusive about the extent to which taxpayers
comply with the law based on their perception of whether noncompliance
will be caught, it is widely believed that those perceptions do 
contribute to the overall level of compliance by taxpayers. On the 
basis of this belief, observers in Congress and elsewhere have been 
concerned as IRS’s oft cited audit rate has declined in recent years. 

To an unknown, but real extent, the long-term decline in the audit rate 
is attributable to the movement of some tax issues from IRS’s audit 
program into its other enforcement programs. This movement has been 
facilitated by changes in technology, and has enabled IRS, for some tax 
issues, to more universally check whether taxpayers have accurately 
reported their tax liabilities. Although much of the movement of IRS’s 
audits into other programs occurred during the 1970s and 1980s, this 
trend continued during the fiscal year 1993 through 2002 period. 

Given these changes in IRS’s enforcement operations, policymakers in
Congress and elsewhere, as well as taxpayers, would be better informed
about the scope of IRS’s efforts to enforce tax laws if IRS were to 
expand its annual public reporting to include the full range of its 
enforcement programs. Toward this end, some interest has been expressed 
in having IRS report a new audit rate that would aggregate IRS’s various
enforcement programs into a total, revised audit rate. Although such a
measure would attempt to provide a more comprehensive picture of IRS’s
overall effort to detect compliance problems, the advantages of doing so
do not clearly outweigh potential disadvantages. For instance, expanding
the definition of an audit would package enforcement activities that 
are so disparate that the consolidated reporting could be misleading. 

However, policymakers and taxpayers could be better informed about of
the extent of IRS’s efforts to enforce the tax laws without combining 
data on all of IRS’s enforcement programs into one set of aggregate 
measures. IRS’s commissioner set fuller reporting of IRS’s enforcement 
efforts as an IRS goal, and IRS officials plan to move to fuller 
reporting of enforcement program results, perhaps as early as in the 
2002 IRS Data Book, which will be published in early calendar year 
2003. IRS officials expect that this expanded reporting will use only 
readily available data on the enforcement programs. 

Because the document matching, math error, and nonfiler programs now
cover many tax issues formerly covered by audits and they annually
contact far more taxpayers than audits, expanded reporting on these
programs, using readily available data, is an appropriate first step.
However, the readily available data for the nonaudit programs is
incomplete compared to data reported on audits. For example, the data do
not cover all nonfiler contacts or the results of the programs by 
taxpayer income. IRS has no plans to determine whether it could cost-
effectively extract or collect additional data in order to more 
completely present program results, and facilitate comparisons across 
the programs or with any new programs, as they evolve. 

In the case of the math error program, total data that includes math 
errors identified during initial processing of tax returns as well as 
errors found in comparing tax return data to data in IRS’s masterfiles 
should be reported. Excluding data on math errors found in comparing 
returns to IRS’s masterfiles materially understates the volume of math 
error contacts with taxpayers. 

Recommendations to the Acting Commissioner of Internal Revenue: 

The Acting Commissioner of Internal Revenue should: 

* determine whether additional data on each nonaudit program can be cost
effectively extracted or collected to make future annual reporting on
enforcement programs more complete and comparable; 

* provide information on all types of math error contacts when 
publishing data on IRS’s math error program. 

Agency Comments: 

The Acting Commissioner of Internal Revenue provided written comments
on a draft of this report in a January 27, 2003, letter, which is 
reprinted in appendix V. The Commissioner agreed with our 
recommendations. We are heartened that IRS has already begun to 
identify additional data to report on its enforcement programs. Given 
the differing nature of IRS’s enforcement programs, we encourage IRS to 
provide information that is as comparable as possible among the 
programs. 

As arranged with your office, we plan no further distribution of this 
report until 30 days from the date of its issue, unless you publicly 
announce its contents earlier. After that period we will send copies to 
the Chairman and Ranking Minority Member, House Committee on Ways and 
Means; and Chairman and Ranking Minority Member, Senate Committee on 
Finance. We will also send copies to the Acting Secretary of the 
Treasury; Acting Commissioner of Internal Revenue; the Director, Office 
of Management and Budget; and other interested parties. Copies of this 
report will be made available to others on request. In addition, the 
report will be made available at no charge on the GAO Web site at 
[hyperlink, http://http://www.gao.gov]. 

If you have any questions, please contact me or Tom Short on (202) 512-
9110. Key contributors to this report are acknowledged in appendix VI. 

Sincerely yours, 

Signed by: 

Michael Brostek: 
Director, Tax Issues: 

[End of section] 

Appendix I: Development of the Math Error Program: 

As early as the first codification of the Internal Revenue law in 1939,
Congress granted IRS “math error” authority so that IRS does not have to
provide the taxpayer with a statutory notice of deficiency[Footnote 30] 
for math errors. In general, these are errors that must be corrected 
for IRS to process the tax return. A 1976 statutory revision defined 
the authority to include not only mathematical errors but other obvious 
errors such as omissions of data needed to substantiate an item on a 
return. In the 1990s, Congress extended the authority five times to 
help determine eligibility for certain tax exemptions and credits. 
Table 3 summarizes the legislative authority on math error provisions 
for individual tax returns. 

Table 3: Legislative Authority on Math Error Provisions for Individual 
Tax Returns: 

Basis of Authority: Internal Revenue Code; 
Provision: Provided a basic “math error” exception to the deficiency 
procedures whereby the Service could notify a taxpayer that on account 
of a mathematical error an amount of tax in excess of that shown on the 
return was due without first sending a notice of deficiency, which 
gives the taxpayer the right to judicial review; 
Year: As early as 1939. 

Basis of Authority: Tax Reform Act of 1976 (P.L. 94-455); 
Provision: Expanded the definition of math errors to include (1) an 
error in addition, subtraction, multiplication, or division shown on 
the return; (2) incorrect use of an IRS table if the error is apparent 
from the existence of other information on the return; (3) inconsistent 
entries on the return; (4) an omission of information required to be
supplied on the return in order to substantiate an item on that return; 
and; (5) entry of a deduction or credit item in an amount which exceeds 
a statutory limit which is either (a) a specified monetary amount or 
(b) a percentage, ratio, or fraction if the items entering into the 
application of that limit appear on that return; 
Year: 1976. 

Basis of Authority: Small Business Job Protection Act of 1996 (P.L. 104-
188); 
Provision: Extended math error authority to the omission of a correct 
Taxpayer Identification Number (TIN) required for the dependent care 
credit or the deduction for personal exemptions; 
Year: 1996. 

Basis of Authority: Personal Responsibility and Work Opportunity
Reconciliation Act of 1996 (P.L. 104-193); 
Provision: Extended math error authority to the omission of a TIN for 
the earned income tax credit; 
Year: 1996. 

Basis of Authority: Taxpayer Relief Act of 1997 (P.L. 105-34); 
Provision: Extended math error authority to the omissions of correct 
TINs for the child tax credit and the higher education tuition tax 
credit, and to information required for the earned income tax credit for
taxpayers who previously made improper claims; 
Year: 1997. 

Basis of Authority: Tax and Trade Relief Extension Act (1999 
Appropriations Act (P.L. 105-277)); 
Provision: Extended math error authority to the inclusion of a TIN on a 
return which allows IRS to determine ineligibility for the dependent 
care credit, child tax credit, or earned income tax credit on the basis 
of the statutory age restrictions of those credits; 
Year: 1999. 

Basis of Authority: Economic Growth and Tax Reconciliation Act of 2001 
(P.L. 107-16); 
Provision: Extended math error authority to include an entry on a 
return claiming the earned income tax credit when, according to the 
Federal Registry of Child Support Orders, the taxpayer is not the 
custodial parent of the child being claimed. This provision, effective 
January 1, 2004, includes a sunset provision of December 31, 2010; 
Year: 2001. 

Source: GAO analysis of legislation. 

[End of table] 

According to IRS officials, math error authority applies to obvious 
errors where most taxpayers do not dispute IRS’s decisions. However, if
taxpayers do disagree with the changes in taxes assessed, they can 
request an abatement (reduction) of the additional taxes. The math 
error process also generates lower administrative and other costs 
because it is highly automated and requires little contact with 
taxpayers, according to IRS officials. 

[End of section] 

Appendix II: Use of Information Returns and Document Matching at IRS: 

IRS has endorsed the concept of matching information returns to income
tax returns for the purpose of identifying unreported income since the
1960s. Prior to the 1960s, employers had reported on wages paid to
employees by the name of the employee. To facilitate matching, Congress
required a TIN—generally a social security number for individual
taxpayers—that is unique to each taxpayer, unlike a name. IRS and those
filing information returns (i.e., payers of income) need accurate TINs 
for the system to work well. [Footnote 31] Filing of information 
returns on magnetic media or other electronic means combined with 
greater IRS computer capacity also has facilitated the matching 
process. 

In 1962, Congress recognized that underreporting of nonwage income,
such as interest and dividend income, was a serious problem. To correct 
it, Congress required information reporting on interest and dividend 
income. Congress substantially expanded information reporting 
requirements during the 1980s and added a few requirements during the 
1990s. Table 4 lists each major statute expanding information returns 
authority. 

Table 4: Major Legislation Affecting the Information Returns Program: 

Statute: Public Law 87-397 (enacted 1961); 
Description: Required taxpayers to provide IRS and payers of income 
with a TIN and established penalties for failure to do so. 

Statute: Revenue Act of 1962 (P.L. 87-834); 
Description: Required information returns reporting for interest and 
dividend income. Required payers to furnish copies of the information 
returns to those receiving interest and dividend income. 

Statute: Combined Old-Age, Survivors, and Disability Insurance-Income 
Tax Reporting Amendments of 1975 (P.L. 94-202); 
Description: Directed IRS and Social Security Administration (SSA) to 
implement an annual wage reporting system, which enhanced IRS’s machine 
processing efficiency because SSA had the equipment and capacity, which 
IRS did not, to process a large volume of Forms W-2. 

Statute: The Economic Recovery Tax Act of 1981 (P.L. 97-34); 
Description: Expanded the requirement that payers furnish all types of 
information returns to the taxpayer receiving a payment. Increased the 
penalties for failure to provide copies of such returns to the taxpayer 
and to IRS. 

Statute: The Tax Equity and Fiscal Responsibility Act of 1982 (P.L. 97-
248); 
Description: Expanded information reporting to include state and local 
income tax refunds, and proceeds from brokers and barter exchanges. 
Mandated 10 percent withholding on interest, dividends, patronage 
dividends, and original issue discount. Expanded and increased 
penalties for failure to (1) file information returns, (2) provide
copies to payees, and (3) provide a payer or payee TIN. Required backup 
withholding at a 15 percent rate in some instances where a payee failed
to provide a correct TIN to a payer. Authorized the Secretary of the 
Treasury to prescribe regulations to define which returns are to be 
filed on magnetic media. 

Statute: Interest and Dividend Tax Compliance Act of 1983 (P.L. 98-67); 
Description: Repealed the mandatory withholding requirements of the Tax 
Equity and Fiscal Responsibility Act of 1982. Expanded and revised 
backup withholding to include a 20 percent rate if (1) the payee does 
not furnish the payer with a TIN, (2) IRS notifies the payer that the 
TIN is incorrect, (3) the payee underreports interest or dividend 
income and IRS notifies the payer, or (4) the payee does not properly 
certify that he or she is not subject to backup withholding for 
interest and dividend income and that the TIN provided to the payer is 
correct. Strengthened TIN and failure to file penalties. Expanded the 
magnetic media filing requirements. 

Statute: Tax Reform Act of 1984 (P.L. 98-369); 
Description: Required information reporting for foreclosures and 
abandonments of property which secure indebtedness and for mortgage 
interest. Provided penalties for failure to file and furnish such 
information returns. 

Statute: Tax Reform Act of 1986 (P.L. 99-514); 
Description: Required real estate brokers to file an information return 
on any real estate transactions. Required federal executive agencies to 
file information returns on persons receiving contracts from them. 
Required persons making royalty payments aggregating $10 or more during 
any calendar year to file information returns on such payments and 
provide a copy of such return to the taxpayer who receives such 
royalties. Required TINs for dependents claimed on tax returns. 
Increased maximum penalties for failure to file information returns and 
to provide copies to taxpayers from $50,000 to $100,000. Added penalty 
for including incorrect information or for omitting required 
information on information returns. 

Statute: Technical Corrections to Tax Reform Act of 1986 (P.L. 100-
647); 
Description: Required that information returns filed by partnerships 
having tax-exempt partners to include reporting of unrelated business 
taxable income. 

Statute: Omnibus Budget Reconciliation Act of 1989 (P.L. 101-239); 
Description: Imposed a uniform penalty of $50 per offense to a maximum 
of $250,000 per year on any person who fails to file timely and correct 
information returns, and to a maximum of $100,000 per year on any 
person who fails to (1) furnish correct payee statements or (2) meet 
other requirements. 

Statute: Omnibus Budget Reconciliation Act of 1993 (P.L. 103-66); 
Description: Required certain financial entities (such as Federal 
Deposit Insurance Corporation, Resolution Trust Corporation, and 
National Credit Union Administration) to file information returns on 
discharges of indebtedness of $600 or more. 

Statute: Health Insurance Portability and Accountability Act of 1996 
(P.L. 104-191); 
Description: Established Medical Savings Accounts (MSAs). Required 
information returns for MSAs. Provided penalties for failure to 
file/furnish the returns. 

Statute: Tax Relief Extension Act of 1999 (P.L. 106-170); 
Description: Required information reporting for indebtedness discharged 
by any organization for which a significant trade or business is the 
lending of money. 

Source: GAO analysis. 

[End of table] 

IRS did not perform extensive document matching until 1974 when IRS
established a program to match information returns against tax return 
data to identify potential income underreporting. Even so, IRS used 
labor intensive, paper-driven methods. For example, clerks had to 
manually create case files for each potential underreporter, and IRS 
staff had to review the case files to determine if income was 
underreported. Clerks entered the results of these file reviews into 
systems, which generated notices to taxpayers. In 1987, IRS began to 
automate the document matching process. At that time IRS established 
the Automated Underreporter Program that allows access to computerized 
information, reducing the need for hard copy documents and clerks, and 
enabling a faster response to taxpayer inquiries. 

By tax year 2000, almost 1.5 billion information returns were filed with
IRS. Table 5 lists the major types of information returns filed for 
1983 and 2000.[Footnote 32] 

Table 5: Major Types of Information Returns Filed for Tax Years 1983 
and 2000: 

Form Number: 1098; 
Title: Mortgage Interest Statement; 
Source: Banks and Mortgage Companies; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 80.2. 

Form Number: 10198-E; 
Title: Student Loan Interest Statement; 
Source: Educational Institutions and Financial Institutions; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 9.6. 

Form Number: 1098-T; 
Title: Tuition Payments Statement; 
Source: Educational Institutions; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 19.8. 

Form Number: 1099-A; 
Title: Acquisition or Abandonment of Secured Property; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 0.4. 

Form Number: 1099-B; 
Title: Proceeds from Broker and Barter Exchange Transactions; 
Source: Brokers; 
Information Returns Filed (millions) 1983: 10.0; 
Information Returns Filed (millions) 2000: 329.4. 

Form Number: 1099-C; 
Title: Cancellation of Debt; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 0.8. 

Form Number: 1099-G; 
Title: Certain Government and Qualified State Tuition Program Payments; 
Source: State Governments; 
Information Returns Filed (millions) 1983: 36.0; 
Information Returns Filed (millions) 2000: 63.7. 

Form Number: 1099-DIV; 
Title: Dividends and Distribution; 
Source: Brokers, Corporations; 
Information Returns Filed (millions) 1983: 82.0; 
Information Returns Filed (millions) 2000: 130.6. 

Form Number: 1099-INT; 
Title: Interest Income; 
Source: Banks; 
Information Returns Filed (millions) 1983: 296.0; 
Information Returns Filed (millions) 2000: 261.1. 

Form Number: 1099-MISC; 
Title: Miscellaneous Income; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: 39.0; 
Information Returns Filed (millions) 2000: 77.7. 

Form Number: 1099-OID; 
Title: Original Issue Discount; 
Source: Banks, Corporations, and Other Financial Institutions; 
Information Returns Filed (millions) 1983: 2.0; 
Information Returns Filed (millions) 2000: 4.8. 

Form Number: 1099-PATR; 
Title: Taxable Distributions Received from Cooperatives; 
Source: Cooperatives; 
Information Returns Filed (millions) 1983: 2.0; 
Information Returns Filed (millions) 2000: 1.6. 

Form Number: 1099-R; 
Title: Distributions from Pensions, Annuities, Retirement or Profit-
sharing Plans, IRAs, Insurance Contracts, Etc. 
Source: Various Entities; 
Information Returns Filed (millions) 1983: 6.0; 
Information Returns Filed (millions) 2000: 65.8. 

Form Number: 1099-S; 
Title: Proceeds from Real Estate Transactions; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 2.9. 

Form Number: 1099-SSA; 
Title: Social Security Benefits; 
Source: Social Security Administration; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 48.4 

Form Number: 1099-RRB; 
Title: Railroad Retirement Benefits; 
Source: Railroad Retirement Board; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 0.6. 

Form Number: 5498; 
Title: IRA Contribution Information; 
Source: Banks, Brokers, and Insurance Companies; 
Information Returns Filed (millions) 1983: 18.0; 
Information Returns Filed (millions) 2000: 94.9. 

Form Number: 5498-MSA; 
Title: Medical Savings Accounts or Medicare Plus Choice MSA 
Information; 
Source: Trustees or custodians of MSAs or Medicare Plus Choice MSAs; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 0.1. 

Form Number: CTR; 
Title: Currency Transaction Report; 
Source: Financial Institutions and Shareholders and Beneficiaries; 
Information Returns Filed (millions) 1983: [A]; 
Information Returns Filed (millions) 2000: 14.4. 

Form Number: K-1; 
Title: Partner’s Share of Income, Credits, Deductions, Etc. 
Source: Partnerships; 
Information Returns Filed (millions) 1983: 15.0; 
Information Returns Filed (millions) 2000: 19.1. 

Form Number: W-2; 
Title: Wage and Tax Statement; 
Source: Employers; 
Information Returns Filed (millions) 1983: 165.0; 
Information Returns Filed (millions) 2000: 247.2. 

Form Number: W-2G; 
Title: Certain Gambling Winnings; 
Source: Gaming Establishments; 
Information Returns Filed (millions) 1983: 1.0; 
Information Returns Filed (millions) 2000: 5.8. 

Form Number: W-2P; 
Title: Annuities, Pensions, Retired Pay, or IRA Payments; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: 18.0; 
Information Returns Filed (millions) 2000: N/A. 

Form Number: Other; 
Title: Various; 
Source: Various Entities; 
Information Returns Filed (millions) 1983: 1.0; 
Information Returns Filed (millions) 2000: 1.7. 

Total: 
Information Returns Filed (millions) 1983: 691.0; 
Information Returns Filed (millions) 2000: 1,480.6. 

Source: IRS and Statement of Johnny C. Finch, Senior Associate 
Director, General Government Division, GAO, Before the Subcommittee on 
Commerce, Consumer, and Monetary Affairs, Committee on Government 
Operations, House of Representatives, on IRS’ Information Returns 
Matching Program, April 29, 1986. 

[A] Information return not required for tax year 1983. 

[End of table] 

In 2002, IRS re-instituted matching[Footnote 33] of income reported by 
flow-through entities such as trusts, partnerships, and S-corporations 
on Schedule K-1 to income reported on tax returns by the related 
partners and beneficiaries. Schedule K-1 shows the income distributed 
to partners and beneficiaries, who receive a copy as well as IRS. 
According to IRS, information provided on Schedule K-1 is important for 
determining whether recipients of flow-through income have properly 
reported that income on their tax returns. IRS expects the matching of 
Schedule K-1 data to increase accurate reporting of trust income on 
future tax returns by providing information that IRS can use to detect 
possible unreported income and to induce taxpayers to voluntarily 
comply. Under K-1 matching, IRS sent 69,097 notices to taxpayers in 
2002 for tax year 2000. In most cases, the taxpayers did not owe 
additional tax for various reasons (e.g., taxpayers reported the income 
differently than expected). IRS does not yet have complete results from 
this new matching program. IRS officials told us that K-1 matching has 
been suspended for one year to analyze the matching criteria and 
results. 

[End of section] 

Appendix III: Individual Audit and Other Enforcement Program Data: 

Table 6: Number and Rates of Individual Audit and Other Enforcement 
Contacts, Fiscal Years 1993 through 2002: 

Audit contacts: Field audits; 
1993: 250,712; 
1994: 364,016; 
1995: 338,605; 
1996: 252,430; 
1997: 209,781; 
1998: 168,054; 
1999: 124,518; 
2000: 91,586; 
2001: 77,950; 
2002: 88,896. 

Audit contacts: Office audits; 
1993: 505,539; 
1994: 456,216; 
1995: 458,880; 
1996: 509,434; 
1997: 505,834; 
1998: 383,366; 
1999: 235,625; 
2000: 145,975; 
2001: 115,971; 
2002: 111,695. 

Audit contacts: Correspondence audits; 
1993: 302,715; 
1994: 405,475; 
1995: 1,121,952; 
1996: 1,179,696; 
1997: 803,628; 
1998: 625,021; 
1999: 715,789; 
2000: 366,657; 
2001: 529,241; 
2002: 538,779. 

Audit contacts: Revenue officer examiner audits; 
1993: [A]; 
1994: [A]; 
1995: [A]; 
1996: [A]; 
1997: [A]; 
1998: 16,339; 
1999: 24,341; 
2000: 13,547; 
2001: 8,594; 
2002: 4,543. 

Audit contacts: Total audits; 
1993: 1,058,966; 
1994: 1,225,707; 
1995: 1,919,437; 
1996: 1,941,560; 
1997: 1,519,243; 
1998: 1,192,780; 
1999: 1,100,273; 
2000: 617,765; 
2001: 731,756; 
2002: 743,913. 

Returns filed (previous calendar year): 
1993: 114,718,900; 
1994: 113,754,400; 
1995: 114,683,400; 
1996: 116,059,700; 
1997: 118,362,600; 
1998: 120,342,400; 
1999: 122,546,900; 
2000: 124,887,100; 
2001: 127,097,400; 
2002: 129,948,400. 

Audit rate: 
1993: 0.92; 
1994: 1.08; 
1995: 1.67; 
1996: 1.67; 
1997: 1.28; 
1998: 0.99; 
1999: 0.90; 
2000: 0.49; 
2001: 0.58; 
2002: 0.57. 

Other enforcement contacts: Math errors; 
1993: 4,088,000; 
1994: 4,059,000; 
1995: 6,102,000; 
1996: 4,750,771; 
1997: 5,983,944; 
1998: 5,668,906; 
1999: 6,552,290; 
2000: 5,751,462; 
2001: 6,082,967; 
2002: 6,265,455. 

Other enforcement contacts: Math errors (masterfile notices); 
1993: [B]; 
1994: [B]; 
1995: [B]; 
1996: [B]; 
1997: 1,834,232; 
1998: 1,894,170; 
1999: 1,965,405; 
2000: 2,010,514; 
2001: 2,026,802; 
2002: 2,061,830. 

Other enforcement contacts: Math errors (revised); 
1993: [C]; 
1994: [C]; 
1995: [C]; 
1996: [C]; 
1997: 7,818,176; 
1998: 7,563,076; 
1999: 8,517,695; 
2000: 7,761,976; 
2001: 8,109,769; 
2002: 8,327,285. 

Other enforcement contacts: Document matching; 
1993: 2,723,830; 
1994: 2,645,075; 
1995: 2,711,086; 
1996: 1,930,326; 
1997: 931,354; 
1998: 1,726,098; 
1999: 1,770,695; 
2000: 1,353,545; 
2001: 1,161,901; 
2002: 1,687,800. 

Other enforcement contacts: Nonfiler; 
1993: 1,603,969; 
1994: 1,931,781; 
1995: 1,756,325; 
1996: 1,302,432; 
1997: 1,917,212; 
1998: 2,313,633; 
1999: 1,890,794; 
2000: 1,251,375; 
2001: 1,371,401; 
2002: 1,882,475. 

Math error rate: 
1993: 3.59; 
1994: 3.54; 
1995: 5.26; 
1996: 4.01; 
1997: 4.97; 
1998: 4.63; 
1999: 5.25; 
2000: 4.53; 
2001: 4.68; 
2002: 4.79. 

Math error rate (revised): 
1993: [D]; 
1994: [D]; 
1995: [D]; 
1996: [D]; 
1997: 6.50; 
1998: 6.17; 
1999: 6.82; 
2000: 6.11; 
2001: 6.24; 
2002: 6.36. 

Document matching rate: 
1993: 2.37; 
1994: 2.33; 
1995: 2.36; 
1996: 1.66; 
1997: 0.79; 
1998: 1.43; 
1999: 1.44; 
2000: 1.08; 
2001: 0.91; 
2002: 1.30. 

Nonfiler rate: 
1993: 1.40; 
1994: 1.70; 
1995: 1.53; 
1996: 1.12; 
1997: 1.62; 
1998: 1.92; 
1999: 1.54; 
2000: 1.00; 
2001: 1.08; 
2002: 1.45. 

Source: GAO analysis of IRS data. 

Note: To compute individual audit and other enforcement rates, we used 
two methods. We used IRS’s method for computing audit rates, which 
equals the proportion of IRS audits closed in a fiscal year as compared 
to returns filed in the previous calendar year. For example, as shown 
in the table above, the audit rate for 1993 is computed by dividing 
total audits (1,058,966) by the number of returns filed in the previous 
calendar year (114,718,900) to compute the audit rate (0.92). IRS has 
not stated a method for computing math error, document matching, and 
nonfiler rates. For the document matching and nonfiler programs, we 
used the IRS audit rate method because document matching and nonfiler 
contacts generally occur in the year after a return is filed. For the
math error program, we compared the math errors notices to the returns 
filed in that year because identifying math errors is part of IRS’ 
returns processing system. For example, the math error rate for
1993 is computed by dividing the number of math errors (4,088,000) by 
the number of returns filed in 1993 (113,754,400) to compute the math 
error rate (3.59). Note: IRS estimates that the number of returns filed 
in 2002 is about 130,905,000. Final data for fiscal year 2002 were not 
available at the time of publication of this report. We used the 
estimate of 130,905,000 returns filed to compute the math error rate 
and the math error rate (revised) for fiscal year 2002. 

[A] IRS did not publish data on revenue officer examiner audits prior 
to 1998. 

[B] Data for math error masterfile notices do not exist prior to fiscal 
year 1997. 

[C] The number of math error (revised) contacts are the same as the 
number of math error contacts for fiscal years 1993 through 1996 
because data for the number of masterfile notices does not exist for
these years. 

[D] The math error (revised) contact rate is the same as the math error 
contact rate for fiscal years 1993 through 1996 because data for the 
number of masterfile notices does not exist for these years. 

[End of table] 

Figure 4: Individual Audit and Other Enforcement Contacts, Fiscal Years 
1993 through 2002: 

This figure is a multiple line graph depicting Individual Audit and 
Other Enforcement Contacts, Fiscal Years 1993 through 2002. The 
vertical axis of the graph represents Contacts from 0 to 9,000. The 
horizontal axis of the graph represents years from 1993 to 2002. Five 
contact rates are depicted on the graph: Math error; Math error 
(revised); Nonfiler; Document matching; and Audits. 

Source: GAO analysis of IRS data. 

The absolute number of contacts with taxpayers under the four 
enforcement programs follows the same general year-to-year and overall
pattern as for contact rates. Similarly, as with the contact rates, the
number of audits and the number of document matching contacts were
lower (30 and 38 percent, respectively) in fiscal year 2002 than in 
1993. The number of nonfiler contacts also was somewhat higher (17 
percent) in 2002 than in 1993 and the number of math error contacts—not 
counting math errors identified from masterfile comparisons—was 
significantly higher (53 percent) in 2002 than in 1993. 

Table 7: Number and Rates of Individual Audit and Document Matching 
Contacts by Income Level, Fiscal Years 1993 through 2002: 

Audit contacts: $100,000 and over; 
1993: 204,079; 
1994: 172,483; 
1995: 179,871; 
1996: 210,032; 
1997: 200,070; 
1998: 164,314; 
1999: 128,398; 
2000: 99,547; 
2001: 91,550; 
2002: 112,266. 

Audit contacts: $25,000 to under $100,000; 
1993: 361,787; 
1994: 347,200; 
1995: 510,764; 
1996: 552,011; 
1997: 423,548; 
1998: 349,378; 
1999: 226,261; 
2000: 148,306; 
2001: 157,296; 
2002: 183,847. 

Audit contacts: Under $25,000; 
1993: 493,100; 
1994: 706,024; 
1995: 1,228,802; 
1996: 1,179,503; 
1997: 895,625; 
1998: 679,088; 
1999: 745,614; 
2000: 369,912; 
2001: 482,910; 
2002: 447,800. 

Document matching contacts: $100,000 and over; 
1993: 213,070; 
1994: 263,287; 
1995: 285,767; 
1996: 238,330; 
1997: 131,348; 
1998: 228,934; 
1999: 231,482; 
2000: 208,839; 
2001: 218,981; 
2002: 275,088. 

Document matching contacts: $25,000 to under $100,000; 
1993: 1,337,067; 
1994: 1,378,983; 
1995: 1,339,480; 
1996: 914,540; 
1997: 469,761; 
1998: 863,408; 
1999: 891,237; 
2000: 674,154; 
2001: 597,034; 
2002: 786,172. 

Document matching contacts: Under $25,000; 
1993: 1,173,693; 
1994: 1,002,805; 
1995: 1,085,839; 
1996: 777,456; 
1997: 330,449; 
1998: 633,756; 
1999: 647,976; 
2000: 470,552; 
2001: 345,886; 
2002: 429,879. 

Returns filed: $100,000 and over; 
1993: 5,240,200; 
1994: 5,635,300; 
1995: 6,058,100; 
1996: 6,546,700; 
1997: 7,301,900; 
1998: 8,156,600; 
1999: 9,178,000; 
2000: 10,368,600; 
2001: 11,610,500; 
2002: 13,020,183. 

Returns filed: $25,000 to under $100,000; 
1993: 45,333,900; 
1994: 45,640,000; 
1995: 46,506,400; 
1996: 47,865,000; 
1997: 49,805,300; 
1998: 51,389,100; 
1999: 53,389,200; 
2000: 55,729,700; 
2001: 57,268,000; 
2002: 59,216,431. 

Returns filed: Under $25,000; 
1993: 64,144,800; 
1994: 62,479,100; 
1995: 62,118,900; 
1996: 61,648,000; 
1997: 61,255,400; 
1998: 60,796,700; 
1999: 59,979,700; 
2000: 58,788,800; 
2001: 58,218,900; 
2002: 57,208,333. 

Audit rate: $100,000 and over; 
1993: 3.89; 
1994: 3.06; 
1995: 2.97; 
1996: 3.21; 
1997: 2.74; 
1998: 2.01; 
1999: 1.40; 
2000: 0.96; 
2001: 0.79; 
2002: 0.86. 

Audit rate: $25,000 to under $100,000; 
1993: 0.80; 
1994: 0.76; 
1995: 1.10; 
1996: 1.15; 
1997: 0.85; 
1998: 0.68; 
1999: 0.42; 
2000: 0.27; 
2001: 0.27; 
2002: 0.31. 

Audit rate: Under $25,000; 
1993: 0.77; 
1994: 1.13; 
1995: 1.98; 
1996: 1.91; 
1997: 1.46; 
1998: 1.21; 
1999: 1.24; 
2000: 0.63; 
2001: 0.83; 
2002: 0.78. 

Document matching rate: $100,000 and over; 
1993: 4.07; 
1994: 4.67; 
1995: 4.72; 
1996: 3.64; 
1997: 1.80; 
1998: 2.81; 
1999: 2.52; 
2000: 2.01; 
2001: 1.89; 
2002: 2.11. 

Document matching rate: $25,000 to under $100,000; 
1993: 2.95; 
1994: 3.02; 
1995: 2.88; 
1996: 1.91; 
1997: 0.94; 
1998: 1.68; 
1999: 1.67; 
2000: 1.21; 
2001: 1.04; 
2002: 1.33. 

Document matching rate: Under $25,000; 
1993: 1.83; 
1994: 1.61; 
1995: 1.75; 
1996: 1.26; 
1997: 0.54; 
1998: 1.04; 
1999: 1.08; 
2000: 0.80; 
2001: 0.59; 
2002: 0.75. 

Source: GAO analysis of IRS data. 

Notes: Returns filed consists of previous calendar year data. Number of 
returns filed in 2002 (calendar year 2001 data) is estimated. 

[End of table] 

Figure 5: Number of Individual Audit and Document Matching Contacts by 
Income Level, Fiscal Years 1993 through 2002: 

[See PDF for image] 

This figure contains two multiple line graphs depicting the number of 
individual audit and document matching contacts by income level, fiscal 
years 1993 through 2002. The vertical axis of the graphs represent 
Audits from 0 to 1,400,000 (first graph) and Document matching from 0 
to 1,600,000 (second graph). The horizontal axis of both graphs 
represents years from 1993 to 2002. Lines depicting the following three 
income levels are represented on each graph: $200,000 and over; $25,000 
to under $100,000; and Under $25,000. 

Source: GAO analysis of IRS data. 

Note: Fiscal year 2002 data is estimated. 

[End of figure] 

The absolute number of audits by income group generally follows the 
same year-to-year and overall pattern, as do the contact rates for the 
income groups. However, the change in the number of audits conducted in
fiscal year 1993 compared to 2002 is not as dramatic for the upper and
middle-income groups as was the change in their audit rates. The number
of audits for the higher income group declined by 45 percent between 
1993 and 2002 while the rate at which this group was audited declined 
78 percent. The number of audits of the middle-income group declined 49 
percent, while the rate at which this group was audited declined 61 
percent. The rate at which these groups were audited fell more than did 
the absolute number of audits because the number of taxpayers in each 
group expanded over the 10-year period. Higher income taxpayers grew in
numbers by 148 percent between 1993 and 2002 and middle-income 
taxpayers grew by 31 percent. 

As with the absolute number of audits by income group, the number of
document matching contacts by income group generally follows the same 
year-to-year and overall pattern as do the document matching contact
rates. However, comparisons of the number of document matching contacts 
in fiscal year 1993 to those in 2002 differ substantially from 
comparisons of document matching contact rates for those years for one
income group—the higher income taxpayers. The number of document 
matching contacts with the higher income taxpayers increased by 29 
percent between 1993 and 2002, whereas the contact rate for this group
fell by 48 percent. The percentage changes in numbers and rates of 
contacts for the other two groups were more similar. Middle-income
document matching contacts fell 41 percent between fiscal year 1993 and
2001 while their document matching contact rate declined 55 percent.
Lowest income taxpayer document matching contacts fell 63 percent,
while their contact rate declined 59 percent. 

Table 8: Number and Rates of Individual Audits by Type of Audit and 
Income Level, Fiscal Years 1993 through 2002: 

Field audits: Under $25,000; 
1993: 80,881; 
1994: 189,748; 
1995: 161,800; 
1996: 85,153; 
1997: 49,053; 
1998: 35,891; 
1999: 26,228; 
2000: 21,433; 
2001: 16,784; 
2002: 15,480. 

Field audits: $25,000 to under $100,000; 
1993: 82,731; 
1994: 98,393; 
1995: 100,501; 
1996: 80,509; 
1997: 66,558; 
1998: 49,891; 
1999: 35,540; 
2000: 28,666; 
2001: 26,159; 
2002: 33,093. 

Field audits: $100,000 and over; 
1993: 87,100; 
1994: 75,875; 
1995: 76,304; 
1996: 86,768; 
1997: 94,170; 
1998: 82,272; 
1999: 62,750; 
2000: 41,487; 
2001: 35,007; 
2002: 40,323. 

Office audits: Under $25,000; 
1993: 248,704; 
1994: 244,952; 
1995: 250,656; 
1996: 238,561; 
1997: 231,944; 
1998: 171,918; 
1999: 95,308; 
2000: 57,017; 
2001: 42,016; 
2002: 31,192. 

Office audits: $25,000 to under $100,000; 
1993: 201,924; 
1994: 167,594; 
1995: 171,438; 
1996: 221,207; 
1997: 223,646; 
1998: 170,341; 
1999: 109,696; 
2000: 68,740; 
2001: 54,526; 
2002: 56,509. 

Office audits: $100,000 and over; 
1993: 54,911; 
1994: 43,670; 
1995: 36,786; 
1996: 49,652; 
1997: 50,244; 
1998: 41,107; 
1999: 30,621; 
2000: 20,218; 
2001: 19,429; 
2002: 23,994. 

Correspondence audits: Under $25,000; 
1993: 163,515; 
1994: 271,324; 
1995: 816,346; 
1996: 855,789; 
1997: 614,628; 
1998: 460,795; 
1999: 608,154; 
2000: 284,981; 
2001: 420,346; 
2002: 399,175. 

Correspondence audits: $25,000 to under $100,000; 
1993: 77,132; 
1994: 81,213; 
1995: 238,825; 
1996: 250,295; 
1997: 133,344; 
1998: 124,451; 
1999: 73,996; 
2000: 45,346; 
2001: 73,189; 
2002: 92,293. 

Correspondence audits: $100,000 and over; 
1993: 62,068; 
1994: 52,938; 
1995: 66,781; 
1996: 73,612; 
1997: 55,656; 
1998: 39,775; 
1999: 33,639; 
2000: 36,330; 
2001: 35,706; 
2002: 47,311. 

Returns filed: Under $25,000; 
1993: 64,144,800; 
1994: 62,479,100; 
1995: 62,118,900; 
1996: 61,648,000; 
1997: 61,255,400; 
1998: 60,796,700; 
1999: 59,979,700; 
2000: 58,788,800; 
2001: 58,218,900; 
2002: 57,208,333. 

Returns filed: $25,000 to under $100,000; 
1993: 45,333,900; 
1994: 45,640,000; 
1995: 46,506,400; 
1996: 47,865,000; 
1997: 49,805,300; 
1998: 51,389,100; 
1999: 53,389,200; 
2000: 55,729,700; 
2001: 57,268,000; 
2002: 59,216,431. 

Returns filed: $100,000 and over; 
1993: 5,240,200; 
1994: 5,635,300; 
1995: 6,058,100; 
1996: 6,546,700; 
1997: 7,301,900; 
1998: 8,156,600; 
1999: 9,178,000; 
2000: 10,368,600; 
2001: 11,610,500; 
2002: 13,020,183. 

Field audit rate: Under $25,000; 
1993: 0.13; 
1994: 0.30; 
1995: 0.26; 
1996: 0.14; 
1997: 0.08; 
1998: 0.06; 
1999: 0.04; 
2000: 0.04; 
2001: 0.03; 
2002: 0.03. 

Field audit rate: $25,000 to < $100,000: 
1993: 0.18; 
1994: 0.22; 
1995: 0.22; 
1996: 0.17; 
1997: 0.13; 
1998: 0.10; 
1999: 0.07; 
2000: 0.05; 
2001: 0.05; 
2002: 0.06. 

Field audit rate: $100,000 and over; 
1993: 1.66; 
1994: 1.35; 
1995: 1.26; 
1996: 1.33; 
1997: 1.29; 
1998: 1.01; 
1999: 0.68; 
2000: 0.40; 
2001: 0.30; 
2002: 0.31. 

Office audit rate: Under $25,000; 
1993: 0.39; 
1994: 0.39; 
1995: 0.40; 
1996: 0.39; 
1997: 0.38; 
1998: 0.28; 
1999: 0.16; 
2000: 0.10; 
2001: 0.07; 
2002: 0.05. 

Office audit rate: $25,000 to < $100,000; 
1993: 0.45; 
1994: 0.37; 
1995: 0.37; 
1996: 0.46; 
1997: 0.45; 
1998: 0.33; 
1999: 0.21; 
2000: 0.12; 
2001: 0.10; 
2002: 0.10. 

Office audit rate: $100,000 and over: 
1993: 1.05; 
1994: 0.77; 
1995: 0.61; 
1996: 0.76; 
1997: 0.69; 
1998: 0.50; 
1999: 0.33; 
2000: 0.19; 
2001: 0.17; 
2002: 0.18. 

Correspondence audit rate: Under $25,000; 
1993: 0.25; 
1994: 0.43; 
1995: 1.31; 
1996: 1.39; 
1997: 1.00; 
1998: 0.76; 
1999: 1.01; 
2000: 0.48; 
2001: 0.72; 
2002: 0.70. 

Correspondence audit rate: $25,000 to < $100,000: 
1993: 0.17; 
1994: 0.18; 
1995: 0.51; 
1996: 0.52; 
1997: 0.27; 
1998: 0.24; 
1999: 0.14; 
2000: 0.08; 
2001: 0.13; 
2002: 0.16. 

Correspondence audit rate: $100,000 and over; 
1993: 1.18; 
1994: 0.94; 
1995: 1.10; 
1996: 1.12; 
1997: 0.76; 
1998: 0.49; 
1999: 0.37; 
2000: 0.35; 
2001: 0.31; 
2002: 0.36. 

Source: GAO analysis of IRS data. 

Note: Revenue officer examiner audits are not included. See table 6. 

[End of table] 

Table 9: Average Direct Staff Hours by Type of Audit and For Document 
Matching Cases, Fiscal Years 1993 through 2002: 

Program: Field audit; 
1993: 17.81; 
1994: 12.69; 
1995: 13.99; 
1996: 20.21; 
1997: 21.84; 
1998: 22.08; 
1999: 24.84; 
2000: 27.64; 
2001: 30.83; 
2002: 28.87. 

Program: Office audit; 
1993: 4.47; 
1994: 4.51; 
1995: 4.27; 
1996: 4.56; 
1997: 4.34; 
1998: 4.49; 
1999: 5.66; 
2000: 7.09; 
2001: 8.91; 
2002: 9.37. 

Program: Correspondence audit; 
1993: 1.43; 
1994: 1.15; 
1995: 0.74; 
1996: 0.73; 
1997: 0.83; 
1998: 0.91; 
1999: 1.08; 
2000: 1.80; 
2001: 1.79; 
2002: 1.71. 

Program: Document matching; 
1993: 0.80; 
1994: 0.81; 
1995: 1.09; 
1996: 0.98; 
1997: 0.53; 
1998: 0.52; 
1999: 0.57; 
2000: 0.61; 
2001: 0.57; 
2002: 0.61. 

Source: GAO analysis of IRS data. 

Note: Revenue officer examiner audits are not included. See table 6. 

[End of table] 

[End of section] 

Appendix IV: Soft Notices and Voluntary Compliance Agreements: 

In addition to the four major enforcement programs, IRS started two 
programs in the mid-1990s to help ensure that taxpayers file timely and
accurate returns, and to minimize the need for enforcement. 

Through the soft notice program, IRS has been sending notices for 
apparent errors on two tax issues—duplicate claims for one allowable
dependent exemption and unfiled self-employment tax returns. IRS uses
soft notices when it has information to indicate that some taxpayer made
an error but not enough information to know for sure, such as which
taxpayer overclaimed a dependent. Soft notices are intended to stimulate
taxpayers to correct the error without IRS having to invest audit time. 

In addition, IRS uses the voluntary compliance agreements program to
address known compliance problems in reporting tip income. To improve
compliance of employees in industries where tip income is a part of
wages, IRS had been auditing the tax returns of tipped employees, which
burdened the employees and employers as well as IRS. To minimize these
burdens while also addressing the compliance problems, IRS began to
explore new methods to achieve voluntary compliance by tipped 
employees, such as voluntary compliance agreements. IRS has negotiated
three types of agreements with certain employers (e.g., restaurants) to
improve compliance by their individual employees in reporting tip 
income. These three types of agreements follow. 

* The Tip Rate Determination Agreement (TRDA) requires that IRS and the
business agree upon a tip rate for various occupations in the business 
and that at least 75 percent of employees in the business agree to 
report at that rate on their income tax return. 

* The Tip Reporting Alternative Commitment (TRAC) does not require a tip
rate to be determined, but does require that the business create written
statements to record employee tips and send the statements to IRS. This
agreement covers all employees and requires that the business educate
employees about their obligation to report their tip income. 

* The Employer-designed Tip Reporting Alternative Commitment Agreement 
(EmTRAC) requires that businesses establish tip reporting procedures 
and prepare a statement on a regular basis (no less than monthly) to 
reflect all tips for each employee. The business must establish an 
education program to train employees about their obligation to report
tip income. 

In general, these two programs are similar to the four major enforcement
programs in that they attempt to correct noncompliance. They differ
because, rather than enforcing the tax laws, both attempt to reduce the
need for enforcement. In sum, their differences tend to outnumber their
similarities, as discussed below. 

Similar to the four enforcement programs, IRS sends soft notices to 
inform taxpayers of potential errors. However, the soft notice does not 
require taxpayers to take any action, and IRS takes no action to verify 
the error or assess tax. Instead, the notice asks taxpayers to examine 
the potential error and file an amended return if they confirm the 
error. Also, the notice informs taxpayers that IRS will monitor these 
types of errors and might contact them if they do not alter their 
reporting in the future. 

The similarity between the voluntary compliance agreements and the
other enforcement programs is that they attempt to correct
noncompliance. Unlike the other programs, these agreements occur
before a return is filed and do not involve sending any notices to
taxpayers. IRS believes that these agreements enhance voluntary
compliance so that IRS can avoid the need to take enforcement action and
assess additional taxes after a return is filed. IRS assures the 
businesses that IRS will not audit their books and records as long as 
they abide by the agreement. However, IRS may still audit the books and 
records of a tipped employee and report any changes to the business. 
IRS officials said that current procedures require follow-up to check 
adherence to these agreements, but the officials were not sure about 
the extent to which this has been occurring. 

IRS has limited data for the soft notice and voluntary compliance
agreement programs, as follows. 

* In 2002, IRS sent 1.2 million soft notices to taxpayers on duplicate
dependent claims on 2001 tax returns; in 1998, IRS sent 1.6 million soft
notices on these duplicate claims and on self-employment tax for 1996 
and 1997 returns.[Footnote 34] This involved 329,000 notices sent to 
taxpayers who reported self-employment income but had not filed a 
schedule SE or paid self-employment tax. IRS did not provide data on 
these notices for any later years. 

* Through 2001, TRDAs and TRACs covered 48,348 establishments in the
casino, beauty, and food and beverage industries. IRS did not have data 
on the number of individual taxpayers covered by these agreements 
because the agreements are made with employers rather than directly 
with the individual taxpayers. 

[End of section] 

Appendix V: Comments from the Commissioner of Internal Revenue: 

[As part of its comments, IRS included an enclosure that provided 
additional data on nonaudit contacts. We did not include this enclosure 
as part of IRS's written comments because the data provided did not 
materially affect our conclusions and recommendations]. 

Commissioner: 
Department Of The Treasury: 
Internal Revenue Service: 
Washington, DC 20224: 

January 27, 2003: 

Mr. Michael Brostek: 
Director, Tax Issues: 
United States General Accounting Office: 
Washington, D.C. 20548: 

Dear Mr. Brostek: 

Thank you for the opportunity to review and comment on your draft 
report entitled "Tax Administration: IRS Should Continue to Expand 
Reporting on Its Enforcement Efforts." I agree with both 
recommendations to the extent we can cost effectively provide 
informative and meaningful data. I believe that expanded reporting will 
clarify the nature of IRS enforcement efforts, which consist of a wide 
variety of actions, ranging from correction of simple errors on a tax 
return to extensive corporate examinations. 

We will address your recommendations as follows: 

* Determine whether additional data on each non-audit program can be 
cost effectively extracted and collected to make future annual 
reporting on enforcement program more complete and comparable. We will 
look for a cost effective manner to capture data for non-audit programs 
in the manner described in the report, which would be similar to that 
traditionally published as audit information. 

* When publishing data on IRS' math error program, provide information 
on all types of math error contact. 

We will include all math error contacts when reporting math error 
program statistics and have adjusted our reporting criteria accordingly. 

I plan to assign the responsibility for implementing the 
recommendations to the Small Business/Self-Employed Division, Wage and 
Investment Division, and the Office of Tax Administration Coordination. 

We have already taken steps to expand reporting of non-audit program 
information that is currently available. Enclosed is additional data 
for 2002, which will be published on the IRS Website and in the 2002 
IRS Data Book. Please note we have separated the Automated 
Underreporter Program into two categories. The first is when IRS 
personnel screen tax returns, cannot resolve discrepancies, and must 
contact the taxpayer. The second is when IRS personnel screen tax 
returns and resolve discrepancies. The screening process eliminated 
many taxpayer contacts and reduced burden on 1.1 million taxpayers in 
2002. 

If you have any questions, your staff may contact Gary Doniger, Acting 
Deputy Director, Office of Tax Administration Coordination at (202) 927-
4800 or Joseph R. Brimacombe, Director, Compliance Policy, Small 
Business/Self-Employed Division at (202) 283-2200. 

Sincerely, 

Signed by: 

Bob Wenzel: 
Acting Commissioner: 

Enclosure: 

[End of section] 

Appendix VI: GAO Contacts and Staff Acknowledgments: 

GAO Contacts: 

Michael Brostek, (202) 512-9110: 

Tom Short, (202) 512-9110: 

Staff Acknowledgments: 

In addition to those named above Susan Baker, Grace Coleman, Susan 
Conlon, Brendan Culley, Michele Fejfar, Leon Green, Marshall Hamlett, 
Shirley Jones, and Jay Pelkofer made key contributions to this product. 

[End of section] 

Related GAO Products: 

U.S. General Accounting Office. Tax Administration: Advance Tax
Refund Program Was a Major Accomplishment, but Not Problem Free.
GAO-02-827. Washington, D.C.: August 2, 2002. 

U.S. General Accounting Office. Tax Administration: New Compliance
Research Effort Is on Track, but Important Work Remains. GAO-02-769.
Washington, D.C.: June 27, 2002. 

U.S. General Accounting Office. Tax Administration: Impact of
Compliance and Collection Program Declines on Taxpayers. GAO-02-674.
Washington, D.C.: May 22, 2002. 

U.S. General Accounting Office. IRS Audit Rates: Rate for Individual
Taxpayers Has Declined But Effect on Compliance Unknown. GAO-01-
484. Washington, D.C.: April 25, 2001. 

U.S. General Accounting Office. Tax Administration: Information on
Selected IRS Tax Enforcement and Collection Efforts. GAO-01-589T.
Washington, D.C.: April 5, 2001. 

U.S. General Accounting Office. Tax Administration: IRS’ Use of
Nonaudit Contacts. GAO/GGD-00-7. Washington, D.C.: March 16, 2000. 

U.S. General Accounting Office. Tax Administration: IRS Is Working to
Improve Its Estimates of Compliance Burden. GAO/GGD-00-11. Washington, 
D.C.: May 22, 2000. 

U.S. General Accounting Office. IRS Audits: Weaknesses in Selecting and
Conducting Correspondence Audits. GGD-99-48. Washington, D.C.: March
31, 1999. 

U.S. General Accounting Office. Tax Administration: IRS’ Audit and
Criminal Enforcement Rates for Individual Taxpayers Across the Country. 
GAO/GGD-99-19. Washington, D.C.: December 23, 1998. 

U.S. General Accounting Office. Internal Revenue Service: Results of
Nonfiler Strategy and Opportunities to Improve Future Efforts. GAO/GGD-
96-72. Washington, D.C.: May 13, 1996. 

U.S. General Accounting Office. Tax Administration: Audit Trends and
Results for Individual Taxpayers. GAO/GGD-96-91. Washington, D.C.: 
April 26, 1996. 

Statement of Johnny C. Finch, Senior Associate Director, General
Government Division, GAO, Before the Subcommittee on Commerce, 
Consumer, and Monetary Affairs, Committee on Government Operations,
House of Representatives, on IRS’ Information Returns Matching Program,
April 29, 1986. 

[End of section] 

Footnotes: 

[1] The fiscal 2002 data on number of tax returns filed by individuals 
had not been finalized before we issued the report. 

[2] This report refers to the portion of tax returns audited by IRS as 
the audit contact rate. Similarly, the portion of tax returns for which 
IRS contacts taxpayers about possible noncompliance is the contact rate 
for the math error, document matching, and nonfiler programs. 

[3] U.S. General Accounting Office, Tax Administration: Audit Trends 
and Results for Individual Taxpayers, GAO/GGD-96-91 (Washington, D.C.: 
Apr. 26, 1996). 

[4] U.S. General Accounting Office, IRS Audit Rates: Rate for 
Individual Taxpayers Has Declined But Effect on Compliance Unknown, GAO-
01-484 (Washington, D.C.: Apr. 25, 2001). 

[5] U.S. General Accounting Office, Tax Administration: IRS’ Use of 
Nonaudit Contacts, GAO/GGD-00-7 (Washington, D.C.: Mar. 16, 2001). 

[6] U.S. General Accounting Office, Tax Administration: IRS Is Working 
to Improve Its Estimates of Compliance Burden, GAO/GGD-00-11 
(Washington, D.C.: May 22, 2000). 

[7] For example, an audit might address unreported income, which is the 
focus of document matching, because document matching can only verify 
individual income reported on information returns. According to IRS, 
information returns only report 80 percent of all individual income 
reported on tax returns. 

[8] IRC sec. 6213 grants math error authority for issues such as 
calculation errors, entries that are inconsistent with or exceed 
statutory limits, various omissions of information, or incorrect use of 
an IRS table. 

[9] IRC sec. 6041, 6044, 6045, 6049, 6050, and 6051, among others, 
authorize information reporting to help identify a discrepancy on 
individual income such as wages, interest, dividends, pension 
distributions, and gross proceeds from stock sales. 

[10] IRC sec. 6020 grants this authority to IRS and allows IRS to 
prepare a return for a taxpayer who did not file as required and 
process that return to assess taxes owed. 

[11] IRC sections 7602 through 7606 and 7609 grant these powers. 

[12] IRC sections 7602, 7605, 7609, and 7491 address these 
restrictions. 

[13] The Restructuring Act also restricted the audit technique that can 
be used to identify unreported income; requires IRS to explain taxpayer 
rights including the right to be represented during audits; requires 
IRS to disclose the general criteria for why the return was selected 
for audit; and restricts the ability to summon a third party for an 
audit. 

[14] Publication 1 (Your Rights as a Taxpayer) explains the rights that 
taxpayers have in contacts with IRS. 

[15] Taxpayers are not always required to file a tax return such as 
when their income is too low. 

[16] Document matching staff time includes the time to resolve 
discrepancies before and after contacting taxpayers. 

[17] The masterfile is IRS’s historical record of transactions 
involving each taxpayer’s account. 

[18] IRS letter to respond to questions from the Senate Finance 
Committee, March 26, 2001. 

[19] IRS had not collected data on these masterfile math errors prior 
to fiscal year 1997. 

[20] U.S. General Accounting Office, Tax Administration: Advance Tax 
Refund Program was a Major Accomplishment, but Not Problem Free, GAO-02-
827 (Washington, D.C.: Aug. 2, 2002), U.S. General Accounting Office, 
IRS’s 2002 Tax Filing Season: Returns and Refunds Processed Smoothly; 
Quality of Assistance Improved GAO-03-314 (Washington, D.C.: Dec. 20, 
2002), and Internal Revenue Service, Tax Compliance Activities Report,
June 24, 2002, prepared in response to a directive in the House Report 
accompanying the legislation (P.L. 107-67). 

[21] Also see GAO-01-484. 

[22] We did not analyze the portion of potentially noncompliant tax 
returns that IRS could not check due to resource limitations. In its 
September 2002 progress report to the IRS Oversight Board, IRS 
presented data showing that it is checking compliance on a decreasing 
portion of potentially noncompliant returns. 

[23] To ensure efficient returns processing, returns with small dollar 
value discrepancies are accepted as filed and taxpayers are not sent 
math error notices. 

[24] IRS had measured the voluntary compliance of individual taxpayers 
periodically, last doing so for tax year 1988. IRS stopped because of 
various congressional and other concerns about the measurement program. 

[25] See U.S. General Accounting Office, Tax Administration: New 
Compliance Research Effort Is On Track, but Important Work Remains, GAO-
02-769 (Washington, D.C.: June 27, 2002). 

[26] IRS annually compiles data about its enforcement activities in its 
Data Book. IRS also publishes that data on its public Web site and may 
use some of that data in other publications such as budget documents. 

[27] IRS’s letter dated March 26, 2001. 

[28] IRS does not track how many taxpayers are contacted by more than 
one program for a tax return. 

[29] IRS’s letter dated March 26, 2001. 

[30] In general, IRS sends taxpayers a written notice, called a 
statutory notice of deficiency, which states that additional tax will 
be assessed and provides 90 days for them to respond. The proposed tax 
is automatically assessed if the taxpayer does not respond or does not
file an appeal. 

[31] These payers file the information returns on income paid with IRS 
as well as the taxpayers receiving the income to induce their voluntary 
compliance in reporting the income on their tax returns. 

[32] Among other reasons, we used 1983 because it was the first year 
after a major expansion of the information reporting requirements. 

[33] IRS had done very limited K-1 matching, relying on electronically 
filed schedules K-1, but stopped this matching in the mid-1990s due to 
resource and other constraints. 

[34] GAO/GGD-00-7. 

[End of section] 

GAO's Mission: 

The General Accounting Office, the investigative arm of Congress, 
exists to support Congress in meeting its constitutional 
responsibilities and to help improve the performance and accountability 
of the federal government for the American people. GAO examines the use 
of public funds; evaluates federal programs and policies; and provides 
analyses, recommendations, and other assistance to help Congress make 
informed oversight, policy, and funding decisions. GAO's commitment to 
good government is reflected in its core values of accountability, 
integrity, and reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through the Internet. GAO's Web site [hyperlink, 
http://www.gao.gov] contains abstracts and full-text files of current 
reports and testimony and an expanding archive of older products. The 
Web site features a search engine to help you locate documents using 
key words and phrases. You can print these documents in their entirety, 
including charts and other graphics. 

Each day, GAO issues a list of newly released reports, testimony, and 
correspondence. GAO posts this list, known as "Today's Reports," on its 
Web site daily. The list contains links to the full-text document 
files. To have GAO e-mail this list to you every afternoon, go to 
[hyperlink, http://www.gao.gov] and select "Subscribe to e-mail alerts" 
under the "Order GAO Products" heading. 

Order by Mail or Phone: 

The first copy of each printed report is free. Additional copies are $2 
each. A check or money order should be made out to the Superintendent 
of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or 
more copies mailed to a single address are discounted 25 percent. 
Orders should be sent to: 

U.S. General Accounting Office: 
441 G Street NW: 
Room LM: 
Washington D.C. 20548: 

To order by Phone: 

Voice: (202) 512-6000:
TDD: (202) 512-2537:
Fax: (202) 512-6061: 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 

Web site: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]: 
E-mail: fraudnet@gao.gov: 

Automated answering system: (800) 424-5454 or (202) 512-7470: 

Public Affairs: 

Jeff Nelligan, managing director, NelliganJ@gao.gov: 
(202) 512-4800: 
U.S. General Accounting Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: