This is the accessible text file for GAO report number GAO-10-968 
entitled 'Tax Gap: IRS Can Improve Efforts to Address Tax Evasion by 
Networks of Businesses and Related Entities' which was released on 
October 26, 2010. 

This text file was formatted by the U.S. Government Accountability 
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. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

Report to the Committee on Finance, U.S. Senate: 

United States Government Accountability Office:
GAO: 

September 2010: 

Tax Gap: 

IRS Can Improve Efforts to Address Tax Evasion by Networks of 
Businesses and Related Entities: 

GAO-10-968: 

GAO Highlights: 

Highlights of GAO-10-968, a report to the Committee on Finance, U.S. 
Senate. 

Why GAO Did This Study: 

A taxpayer can control a group of related entities—such as trusts, 
corporations, or partnerships—in a network. These networks can serve a 
variety of legitimate business purposes, but they also can be used in 
complex tax evasion schemes that are difficult for the Internal 
Revenue Service (IRS) to identify. 

GAO was asked to (1) describe what IRS knows about network tax evasion 
and how well IRS’s traditional enforcement programs address it and (2) 
assess IRS’s progress in addressing network tax evasion and 
opportunities, if any, for making further progress. To do this, GAO 
reviewed relevant documentation about IRS programs and interviewed 
appropriate officials about those programs and IRS’s plans for 
addressing such tax evasion. GAO also interviewed relevant experts and 
agency officials in developing criteria needed to perform the 
assessment. 

What GAO Found: 

IRS views network-based tax evasion as a problem but does not have 
estimates of the associated revenue loss in part because data do not 
exist on the population of networks. IRS does know that at least 1 
million networks existed involving partnerships and similar entities 
in tax year 2008. IRS also knows that many questionable tax shelters 
and abusive transactions rely on the links among commonly owned 
entities in a network. 

IRS generally addresses network-related tax evasion through its 
examination programs. These programs traditionally involve identifying 
a single return from a single tax year and routing the return to the 
IRS division that specializes in auditing that type of return. From a 
single return, examiners may branch out to review other entities if 
information on the original return appears suspicious. However, this 
traditional approach does not align well with how network tax evasion 
schemes work. Such schemes can cross multiple IRS divisions or require 
time and expertise that IRS may not have allocated at the start of an 
examination. A case of network tax evasion also may not be evident 
without looking at multiple tax years. 

IRS is developing programs and tools that more directly address 
network tax evasion. One, called Global High Wealth Industry, selects 
certain high-income individuals and examines their network of entities 
as a whole to look for tax evasion. Another, yK-1, is a computerized 
visualization tool that shows the links between entities in a network. 
These efforts show promise when compared to GAO’s criteria for 
assessing network analyses. They represent new analytical approaches, 
have upper-management support, and cut across divisions and database 
boundaries. However, there are opportunities for more progress. For 
example, IRS has no agencywide strategy or goals for coordinating its 
network efforts. It has not conducted assessments of its network 
tools, nor has it determined the value of incorporating more data into 
its network programs and tools or scheduled such additions. Without a 
strategy and assessments, IRS risks duplicating efforts and managers 
will not have information about the effectiveness of the new programs 
and tools that could inform resource allocation decisions. 

Network Scheme Example: Installment Sale Bogus Optional Basis 
Transaction (iBOB): 

An iBOB is an example of a network-related tax evasion scheme that 
shows how networks pose enforcement challenges for IRS. In an iBOB, a 
taxpayer uses multiple entities, all owned or controlled by the 
taxpayer, to artificially adjust the basis of an asset to evade 
capital gains taxes. The scheme can involve multiple transactions and 
take place over many tax years, making it difficult for IRS to detect. 
A short video illustrating an iBOB is available at [hyperlink, 
http://www.gao.gov/products/GAO-10-968]. 

What GAO Recommends: 

Among other items, GAO recommends that IRS establish an IRS-wide 
strategy that coordinates its network tax evasion efforts. Also, IRS 
should assess its network programs and tools and should evaluate 
adding more data to its current tools. IRS generally agreed with these 
recommendations and noted additional organizational changes the agency 
is making that will address networks. 

View [hyperlink, http://www.gao.gov/products/GAO-10-968] or key 
components. For more information, contact James White at (202) 512-
9110 or whitej@gao.gov. 

[End of section] 

Contents: 

Letter: 

Background: 

IRS Suspects Networks Pose a Growing Tax Evasion Risk but Faces 
Barriers in Addressing the Risk through Its Traditional Enforcement 
Efforts: 

IRS's Recent Efforts to Better Detect and Pursue Network Tax Evasion 
Show Promise, but Opportunities Exist for Additional Progress: 

Conclusions: 

Recommendations for Executive Action: 

Agency Comments and our Evaluation: 

Appendix I: Objectives, Scope, and Methodology: 

Appendix II: Example Diagrams of Network Relationships by Entity Type: 

Appendix III: Overview of Studies Using Formal Network Analysis to 
Examine or Detect Criminal/Illicit Activity: 

Appendix IV: Comments from the Internal Revenue Service: 

Appendix V: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Descriptions of Entities That Can Be Linked in Networks: 

Table 2: How Different Types of Entities Can Be Connected in a Network 
through Ownership or as a Beneficiary: 

Table 3: IRS Progress in Meeting Network Analysis Criteria: 

Table 4: Key Studies Using Visualization or Core Network Measures to 
Explain Criminal or Illicit Activity: 

Table 5: Key Studies Using Visualization or Core Network Measures for 
Intervention and Enforcement in Criminal Networks: 

Table 6: Key Studies Using Data-Mining and Related Approaches for 
Analyzing and Detecting Criminal/Illicit Activity: 

Table 7: Key Studies Using Multivariate and Statistical Analysis to 
Identify Associations and Causal Relationships among Network Variables 
and Key Behavioral Outcomes: 

Figures: 

Figure 1: Example of a Complex Business Network from IRS Research: 

Figure 2: Growth in Partnership Forms 1065 and 1065B and S Corporation 
Form 1120S Filings, Calendar Years 1998 through 2008: 

Figure 3: Example of yK-1 Output Graphic: 

Figure 4: Example of a Network with C Corporations: 

Figure 5: Example of a Network with S Corporations: 

Figure 6: Example of a Network with Partnerships: 

Figure 7: Example of a Network with Trusts: 

Abbreviations: 

CI: Criminal Investigation: 

EIN: Employer identification number: 

EOT: Enhancing Ownership Transparency: 

GHWI: Global High Wealth Industry: 

iBOB: Installment Sale Bogus Optional Basis Transaction: 

IRS: Internal Revenue Service: 

LDC: Lead Development Center: 

LMSB: Large and Mid-Size Business: 

NRP: National Research Program: 

OTSA: Office of Tax Shelter Analysis: 

SAT ESC: Servicewide Abusive Transaction Executive Steering Committee: 

SB/SE: Small Business/Self Employed: 

TE/GE: Tax Exempt and Government Entities: 

TIN: Taxpayer identification number: 

W&I: Wage and Investment: 

[End of section] 

United States Government Accountability Office:
Washington, DC 20548: 

September 24, 2010: 

The Honorable Max Baucus: 
Chairman: 
The Honorable Charles Grassley: 
Ranking Member: 
Committee on Finance: 
United States Senate: 

Taxpayers can use networks to carry out a wide variety of legitimate 
business functions; however, networks also can be used to evade 
federal tax obligations. A network is a collection of entities linked 
through direct ownership or through common owners, associates, or 
shareholders. For example, a network may consist of a taxpayer who 
owns a corporation that does business with a partnership in which the 
same taxpayer is a majority shareholder. 

Network-related federal tax evasion occurs when the network's 
taxpayers improperly structure complex transactions among commonly 
held entities. This allows the taxpayers to shift expenses or hide 
income, resulting in lost federal revenue. In one recent example, the 
U.S. Court of Appeals for the Ninth Circuit held that a family used 
transactions between commonly owned entities to improperly eliminate 
$200 million in capital gains and avoid $4 million in taxes.[Footnote 
1] The extent of network-related tax noncompliance has not been 
estimated, but in a 2007 tax compliance forum we sponsored with the 
Joint Committee on Taxation and the Congressional Budget Office, tax 
policy experts cited flows of income among related entities as a 
potentially large problem.[Footnote 2] Some of the participants also 
said the Internal Revenue Service (IRS) needed to be more innovative 
in auditing such flows. 

IRS recognizes the risk posed by network-related tax evasion and is 
developing new tools and programs to better identify and pursue such 
evasion. For example, one new tool helps examiners graph the 
relationship among entities in a taxpayer's network using information 
collected in an IRS database. IRS's new tools and programs are in 
various stages of development--some have yet to be fully implemented-- 
and their potential effectiveness is not known. 

Given the above, you asked for more information about network-related 
tax evasion and IRS's efforts to combat it. Specifically, you asked us 
to (1) describe what IRS knows about network tax evasion and how well 
IRS's traditional enforcement programs address network tax evasion and 
(2) assess IRS's progress in addressing network tax evasion and 
opportunities, if any, for making further progress. 

To describe what IRS knows and how well its traditional enforcement 
efforts address network-related tax evasion, we reviewed IRS planning 
documents and statistics, and interviewed relevant officials at IRS. 
We also developed a video with technical assistance from IRS's tax 
enforcement experts that illustrates a hypothetical example of a 
network tax evasion scheme. To assess IRS's progress and opportunities 
for further progress, we compiled an inventory of IRS's network 
compliance efforts by reviewing IRS documentation on auditing 
procedures for network-related cases and interviewing officials 
involved with identifying and addressing tax evasion related to 
networks. Because we could not find existing criteria that could be 
used for assessing IRS's network compliance efforts, we developed 
criteria. To do this, we conducted semistructured interviews with 
researchers who specialized in areas such as network analysis and 
computer science, interviewed relevant officials at select federal 
agencies that operate programs analyzing related entities or networks, 
and interviewed IRS examiners about their work. We selected the 
researchers based on our literature search of studies applying network 
analysis techniques to noncompliance behavior as well as from 
referrals from the researchers we had already interviewed. We 
distilled the common themes our interviews to establish criteria. 
Officials responsible for IRS's compliance programs concurred with the 
criteria. Using our work on the inventory of IRS's efforts, we then 
compared the status of each of the efforts against the criteria. 

We determined for the purposes of this review that the data used were 
reliable. (See appendix I for details on our scope and methodology.) 
We conducted this performance audit from June 2009 through September 
2010 in accordance with generally accepted government auditing 
standards. Those standards require that we plan and perform the audit 
to obtain sufficient, appropriate evidence to provide a reasonable 
basis for our findings and conclusions based on our audit objectives. 
We believe that the evidence obtained provides a reasonable basis for 
our findings and conclusions based on our audit objectives. 

Background: 

Networks of related entities are a feature of modern business 
organizations. Many legitimate reasons explain why a business owner 
(or owners) may choose to use a network of related entities to conduct 
operations. While this list is not exhaustive, a network may be used 
legitimately to: 

* isolate one line of business from the potential liabilities or risk 
of business loss of another; 

* isolate regulated industries into separate entities to manage 
ownership, reporting, or licensing requirements; 

* manage a business's financing arrangements; 

* separate ventures based in different states and countries; or: 

* separate activities for which ownership is restricted from those for 
which ownership is not restricted (such as when subchapter S 
corporation ownership restrictions apply). 

A variety of entities can be linked in networks and report taxes in 
different ways. Table 1 briefly describes some of the entities that 
will be discussed in this report.[Footnote 3] Certain entities file 
tax returns with IRS to report their taxes owed, such as subchapter C 
corporations (C corporations), which file Form 1120, U.S. Corporation 
Income Tax Return. Other entities may operate as pass-throughs. A pass-
through entity generally has the legal right to impute or pass net 
income or losses through to its partners, shareholders, and 
beneficiaries untaxed. Pass-through entities are required to provide 
each partner, shareholder, or beneficiary with a Schedule K-1[Footnote 
4] stating the individual share of net income or loss to be reported. 
The entities also provide this information to IRS. The partners, 
shareholders, or beneficiaries are responsible for reporting this 
income or loss on their individual income tax returns and paying any 
tax. Entities that may serve as pass-throughs include subchapter S 
corporations (S corporations) and partnerships. 

Table 1: Descriptions of Entities That Can Be Linked in Networks: 

Entity type: Individual; 
Description: An individual meeting certain income and age requirements 
is required to file tax returns. Individuals generally report taxes on 
Form 1040 U.S. Individual Income Tax Return. 

Entity type: Corporation; 
Description: A corporation is a separate legal entity generally 
established under state law with limited liability for shareholders. 
Corporations are treated as either C corporations or S corporations 
(named after Subchapters C and S of the Internal Revenue Code) for 
federal income tax purposes; 
* Any corporation not eligible or not electing to be treated as an S 
corporation is a C corporation. A C corporation's income is taxed at 
the corporate level; once distributed to a shareholder, it may be 
taxed again as income for the shareholder. C corporations generally 
report taxes on Form 1120 U.S. Corporation Income Tax Return; 
* An S corporation is a corporation that has elected to be treated as 
an S corporation. Only corporations meeting certain eligibility 
requirements, such as having 100 or fewer shareholders, may elect to 
be an S corporation. Only certain types of entities, including 
individuals, and certain trusts, estates, and tax exempt organizations 
may be shareholders of an S corporation. Generally, S corporations 
file Form 1120S U.S. Income Tax Return for an S Corporation, which 
provides information about the income and deductions of the 
corporation and the identity of the shareholders. In general, income 
and losses of an S corporation are not taxed at the corporate level, 
but are imputed to the shareholders. 

Entity type: Partnership; 
Description: A partnership is a for-profit business, including any 
syndicate, group, pool, joint venture, or other unincorporated 
organization, with at least two owners and which is not a trust, 
estate, or corporation. A partnership generally must file a Form 1065 
U.S. Return of Partnership Income, which provides information about 
the income and deductions of the partnership and the identity of the 
partners. Partnerships do not pay taxes; 
the income and losses of a partnership are imputed to the partners. 

Entity type: Trust; 
Description: A trust is a fiduciary relationship governed by state 
law. A trust is created when the grantor, or creator of the trust, 
transfers property to a trustee to be held for the benefit of the 
beneficiaries. A trust is sometimes said to divide ownership of the 
trust property into two parts: legal ownership, which goes to the 
trustee, and beneficial ownership, which goes to the beneficiary. 
Trusts are often used for holding the assets of beneficiaries, estate 
planning, or charitable purposes. Taxes on income earned by the trust 
generally are owed by the trust or the beneficiaries, although certain 
trusts are ignored for tax purposes and treated as if the grantor 
owned the trust property. 

Entity type: Tax exempt entity; 
Description: Tax exempt entities are those such as certain charities 
and private foundations that have applied for, and been granted, tax 
exempt status. Tax exempt entities can be various types of entities, 
such as corporations or trusts. Because a tax exempt organization may 
lose its exempt status if a substantial part of its activities 
involves a nonexempt purpose, tax exempt organizations have the 
ability to establish for-profit subsidiaries. Most tax exempt entities 
must file a Form 990 Return of Organization Exempt from Income Tax. 

Entity type: Single-owner organization; 
Description: A single owner organization, also referred to as a 
disregarded entity, is an eligible entity with a single owner that has 
elected to be disregarded for federal income tax purposes. Many 
alternative corporate forms recognized by state law such as limited 
liability companies may be treated as single owner organizations. Sole 
proprietorships, which are unincorporated businesses run by 
individuals, are commonly classified as single owner organizations. 

Source: GAO analysis of IRS guidance. 

[End of table] 

The various types of entities that make up networks can be linked in 
multiple ways. Table 2 summarizes how a select set of individuals, 
various forms of businesses, and trusts may own or control other 
entities. The ownership, beneficiary status, or family connections 
within a network may not be initially apparent. Individual entities 
connected to an owner may in turn own other entities or be the 
beneficiaries of other trusts, thus creating the potential for a 
large, complex network. For illustrative purposes, appendix II 
includes diagrams showing how certain entities can be linked and how 
the income generated by these entities can pass around a network. 

Table 2: How Different Types of Entities Can Be Connected in a Network 
through Ownership or as a Beneficiary: 

Individuals: 
Can own shares in a C corporation: [Check]; 
Can be a partner in a partnership: [Check]; 
Can own shares in an S corporation: [Check][A]; 
Can be a beneficiary of a trust: [Check]. 

C corporations: 
Can own shares in a C corporation: [Check]; 
Can be a partner in a partnership: [Check]; 
Can own shares in an S corporation: [Empty]; 
Can be a beneficiary of a trust: [Check]. 

Partnerships: 
Can own shares in a C corporation: [Check]; 
Can be a partner in a partnership: [Check]; 
Can own shares in an S corporation: [Empty]; 
Can be a beneficiary of a trust: [Check]. 

S corporations: 
Can own shares in a C corporation: [Check]; 
Can be a partner in a partnership: [Check]; 
Can own shares in an S corporation: [Check][B]; 
Can be a beneficiary of a trust: [Check]. 

Trust: 
Can own shares in a C corporation: [Check]; 
Can be a partner in a partnership: [Check]; 
Can own shares in an S corporation: [Check][C]; 
Can be a beneficiary of a trust: [Check]. 

Source: GAO analysis. 

[A] In general, individuals who are not U.S. citizens or resident 
aliens are not eligible S corporation shareholders. Nonresident aliens 
are usually not eligible to be S corporation shareholders. An 
exception does exist if a nonresident alien, who is married to a U.S. 
citizen or resident, elects to report all of his or her worldwide 
income on a joint return with the U.S. citizen or resident. 26 U.S.C. 
§ 1361(b)(1)(C); 26 C.F.R. § 1.1361-1(g). 

[B] Shares of an S corporation can only be owned by another S 
corporation if it is a qualified subchapter S subsidiary (QSub), and 
it must be wholly owned by the other S corporation. Only the parent S 
corporation owes a tax return to the federal government, as the income 
and deductions of the QSub are combined with those of the parent 
corporation. 26 U.S.C. § 1361(b)(3). 

[C] Only certain trusts may be shareholders of an S corporation. 26 
U.S.C. § 1361(c)(2). These trust types include certain grantor trusts, 
testamentary trusts, voting trusts, Qualified Subchapter S trusts, and 
electing small business trusts. 26 U.S.C. § 1361(c)(2). 

[End of table] 

When taxpayers use multiple pass-through entities, they create what 
IRS calls a "multitiered" network. In IRS's definition, a multitiered 
network exists when a pass-through entity is itself a partner, 
shareholder, or beneficiary of another pass-through entity, leading to 
a situation where income is allocated from one pass-through entity to 
another.[Footnote 5] Figure 1, adapted from an IRS study, is an 
illustration of a hypothetical, complex network. In IRS's example, the 
allocation from the observed partnership on the far left side of the 
diagram crosses nine pass-through entities along the red line before 
it reaches one of its ultimate owners on the right. 

Figure 1: Example of a Complex Business Network from IRS Research: 

[Refer to PDF for image: illustration] 

The illustration is an organizational diagram with the following 
specifically identified: 

Observed partnership: 
Partnership; 
C corporation; 
Individual; 
Trust; 
S corporation; 
High-income individual; 
A partnership owner. 

Source: GAO adaptation from IRS study. 

[End of figure] 

Owners of a network could use transactions among entities in the 
network to create tax evasion schemes in which taxpayers: 

* improperly conceal property ownership or income by diverting assets 
or income from one entity to another; 

* improperly inflate an asset's basis[Footnote 6] to reduce capital 
gains taxes by selling the asset within the network; 

* improperly generate losses or tax deductions, which are passed 
through to other entities that use the losses or tax deductions to 
offset gains; 

* inappropriately shift losses, deductions, or credits from entities 
not subject to U.S. federal income tax, such as foreign entities, to 
those who are; or: 

* inappropriately shift income from those entities subject to U.S. 
federal income tax to those entities that are not. 

One example of a network tax evasion scheme is what IRS calls an 
installment sale bogus optional basis transaction (iBOB). In an iBOB 
scheme, taxpayers use commonly owned or controlled entities to 
artificially adjust the basis of an asset and evade capital gains 
taxes. The scheme can involve many layers of ownership, can take place 
over many tax years, and can be shrouded by legitimate transactions. 
IRS understands how the scheme works and has alerted examiners to its 
existence. IRS's Web site also describes a similar abusive 
transaction.[Footnote 7] In cooperation with IRS, we developed a video 
explaining how an iBOB works and the challenges IRS faces in ensuring 
those who are engaged in such schemes are caught. In our simplified 
example, a hypothetical taxpayer, Mr. Jones, sells a hotel that has 
appreciated in value resulting in capital gains that are taxable 
income. Mr. Jones uses an iBOB to evade paying capital gains taxes. 
[Footnote 8] 

Successful tax evasion schemes exacerbate the tax gap, which is the 
difference between the tax amount--including individual income, 
corporate income, employment, estate, and excise taxes--that should 
have been paid voluntarily and on time and the amount that was 
actually paid for a specific year. IRS most recently estimated that 
the tax gap was a net of $290 billion in 2001.[Footnote 9] The tax-gap 
estimate relies on National Research Program (NRP) data. NRP compiled 
data on taxpayers' noncompliance by randomly sampling from the 
population of individual filers, intensively reviewing tax returns in 
the sample to determine the extent of noncompliance, and using the 
sample results to produce noncompliance estimates for the entire 
population. 

IRS has four operating divisions--Wage and Investment (W&I), Small 
Business/Self-Employed (SB/SE), Large and Mid-Size Business (LMSB), 
[Footnote 10] and Tax Exempt and Government Entities (TE/GE). Each 
division has its own compliance programs, such as conducting 
examinations. W&I generally addresses individual taxpayers filing Form 
1040; SB/SE addresses small businesses with assets of less than $10 
million and self-employed taxpayers; LMSB addresses C corporations, S 
corporations, and partnerships with assets of $10 million or more; and 
TE/GE addresses pension plans, exempt organizations, and government 
entities. IRS's Criminal Investigation (CI) unit also investigates 
cases of fraud that may involve networks. CI has investigative 
jurisdiction over tax, money laundering, and bank secrecy laws. 

The Office of Tax Shelter Analysis (OTSA), the Lead Development Center 
(LDC), and the Servicewide Abusive Transaction Executive Steering 
Committee (SAT ESC) are three groups within IRS with responsibilities 
that may touch on network tax evasion. OTSA is an LMSB group that 
collects and analyzes information, some of which is reported by 
taxpayers[Footnote 11] about certain tax shelters.[Footnote 12] LDC in 
SB/SE acts as the clearinghouse to receive, identify, and develop 
leads on individuals and entities that promote and/or aid in the 
promotion of abusive tax avoidance transaction schemes. IRS has 
charged SAT ESC with coordinating information about tax shelter 
schemes--including those that might involve networks--that individual 
operating divisions identify. 

IRS Suspects Networks Pose a Growing Tax Evasion Risk but Faces 
Barriers in Addressing the Risk through Its Traditional Enforcement 
Efforts: 

IRS Does Not Know the Magnitude of Network Tax Evasion, but Has 
Observed an Increase in Risk Factors for Such Evasion: 

IRS does not have an estimate of the total amount of revenue lost 
through network tax evasion because of cost and complexity 
constraints. IRS faces challenges in developing an NRP-type study to 
estimate the amount of network tax evasion because it does not know 
the population of networks. Therefore, IRS does not know what portion 
of the $290 billion net tax gap is network related. Nor does IRS 
routinely track the amount of network tax evasion it identifies 
through its enforcement programs. As will be discussed in more detail 
below, IRS's enforcement programs have traditionally focused on single 
entities as the unit of analysis, such as an individual or 
corporation, rather than networks. 

While IRS does not know the population of networks, it has estimated 
the size of a subset of that population. Based on a study of networks 
with two or more pass-through entities, IRS estimated that in tax year 
2008, more than 1 million of these networks existed, of which about 2 
percent had 11 or more different entities' returns. 

Although IRS lacks an estimate of network tax evasion, IRS officials 
said they have evidence of a problem because of their experiences with 
abusive tax shelters. Some of the tax schemes that IRS considers 
impermissible necessarily involve, or could involve, networks. IRS 
maintains a list of tax avoidance transactions on its Web site; any 
taxpayer engaging in such a listed transaction, or a transaction 
substantially similar to a listed transaction, must disclose to IRS 
certain information about that transaction.[Footnote 13] IRS's list of 
tax avoidance transactions includes examples of abusive tax shelters 
involving networks. The iBOB scheme previously described is an example 
of network tax evasion involving a tax shelter. 

IRS officials have cited trends that they said indicate an increased 
risk of network tax evasion. These officials noted the increased use 
of pass-through entities. This suggested to them that the use of 
networks is growing, that networks are becoming increasingly complex, 
and that the risk of tax evasion is growing. Figure 2 illustrates the 
extent to which partnership and S corporation tax return filings have 
increased from calendar years 1998 to 2008. 

Figure 2: Growth in Partnership Forms 1065 and 1065B and S Corporation 
Form 1120S Filings, Calendar Years 1998 through 2008: 

[Refer to PDF for image: multiple line graph] 

Fiscal year: 1998; 
Forms 1065 and 1065B: 1.9 million; 
Forms 1120S: 2.6 million. 

Fiscal year: 1999; 
Forms 1065 and 1065B: 2 million; 
Forms 1120S: 2.8 million. 

Fiscal year: 2000; 
Forms 1065 and 1065B: 2.1 million; 
Forms 1120S: 2.9 million. 

Fiscal year: 2001; 
Forms 1065 and 1065B: 2.2 million; 
Forms 1120S: 3 million. 

Fiscal year: 2002; 
Forms 1065 and 1065B: 2.3 million; 
Forms 1120S: 3.2 million. 

Fiscal year: 2003; 
Forms 1065 and 1065B: 2.4 million; 
Forms 1120S: 3.4 million. 

Fiscal year: 2004; 
Forms 1065 and 1065B: 2.5 million; 
Forms 1120S: 3.5 million. 

Fiscal year: 2005; 
Forms 1065 and 1065B: 2.7 million; 
Forms 1120S: 3.7 million. 

Fiscal year: 2006; 
Forms 1065 and 1065B: 2.9 million; 
Forms 1120S: 3.9 million. 

Fiscal year: 2007; 
Forms 1065 and 1065B: 3.1 million; 
Forms 1120S: 4.2 million. 

Fiscal year: 2008; 
Forms 1065 and 1065B: 3.3 million; 
Forms 1120S: 4.4 million. 

Source: GAO analysis of IRS data. 

[End of figure] 

Schedule K-1 filings from pass-through entities also have increased. 
From 2008 to 2009, submission of Schedules K-1 increased from 19.8 
million to 21.2 million for partnerships filing Form 1065, U.S. Return 
of Partnership Income. Meanwhile, submission of Schedule K-1 forms for 
S corporations filing Form 1120S, U.S. Income Tax Return for an S 
Corporation, stayed about the same at about 7 million from 2008 to 
2009. 

IRS examiners we spoke with who have experience in network-related 
examinations said that, anecdotally, they have noticed an increase in 
the use of disregarded entities in a network, which they said is 
another risk factor for network tax evasion. A disregarded entity can 
be part of a network, but its connection to a taxpayer may not be 
clear in the tax information IRS uses to detect network tax evasion. 
The total number of disregarded entities is unknown, but IRS estimated 
that there were at least 443,000 disregarded entities during a period 
between July 2007 and August 2008. 

IRS's Traditional Enforcement Programs Are Not Designed to Detect 
Network Tax Evasion: 

IRS's programs for addressing network-related tax evasion include its 
examinations (or audits) in which IRS examiners analyze taxpayers' 
records to ensure that the proper tax was reported. IRS's examination 
practices have made contributions to tax enforcement. In fiscal year 
2009, IRS examined 1.6 million tax returns, identifying over $49 
billion in additional recommended tax. 

IRS traditionally has conducted examinations on a return-by-return 
basis, beginning with a single tax return in a particular tax year as 
the unit of analysis and examining other tax returns connected with 
the original return, if necessary, in what can be called a bottom-up 
approach.[Footnote 14] The examination selection process generally 
involves identifying a pool of high-risk returns and from that group, 
determining which returns to examine. 

CI follows a similar approach. It starts with a taxpayer suspected of 
criminal violations of the Internal Revenue Code or related financial 
crimes and then branches out to related entities. When investigators 
want to find connections between the suspected taxpayer and other 
entities, they use Reveal, a network visualization tool. Reveal draws 
on data from multiple sources that CI uses to analyze intelligence and 
to detect patterns of criminal and terrorist activities. Data that 
Reveal uses include certain cash transactions, tax information, and 
counterterrorism information. Its outputs include a visual 
representation containing names, Social Security numbers, addresses, 
and other personal information of individuals suspected of financial 
crime or terrorist activity. Because of CI's authority to access 
sensitive information, only in rare instances do non-CI staff use 
Reveal, according to CI data security staff. 

IRS's traditional enforcement efforts are not designed to identify 
networks, select those networks that appear to be involved in tax 
evasion, or follow up with in-depth examination or investigation. 
Specifically, IRS's traditional efforts are challenged in dealing with 
network tax evasion by the combined effects of a number of factors 
such as the following. 

* A bottom up approach focusing on a single taxpayer. As with the 
businesses in the iBOB scheme previously described, an entity involved 
in a network may not raise suspicions when examined in isolation. The 
tax evasion may only be apparent in how it relates to other entities 
in the network. 

* Internal divisional boundaries. A single network may contain many 
types of entities that cross the responsibilities of IRS's operating 
divisions (i.e., W&I, SB/SE, LMSB, TE/GE). While IRS has the SAT ESC 
in place for overseeing abusive transaction issues, examiners on any 
particular audit may not have the expertise or authority to pursue the 
network connections of the taxpayer under review. For example, SB/SE 
examiners auditing a small partnership may not have the time, 
expertise, or authority to recognize or pursue a related large S 
corporation that is a member of the partnership. 

* Single tax year examinations. IRS examiners typically begin 
examinations by looking at tax return data for a single tax year, 
limiting their opportunity to notice multiyear schemes. The iBOB is an 
example of a scheme in which the transactions creating the tax evasion 
can occur in multiple tax years. 

* Competing time and resource priorities. IRS generally aims to 
conduct examinations in a manner that maximizes the amount of tax 
noncompliance found while minimizing an examiner's time commitment. 
Network examinations may be highly time-consuming for an examiner and 
the outcome is less predictable. 

Examiners' ability to follow network connections also is restricted in 
another way. The Taxpayer Browsing Protection Act[Footnote 15] 
prohibits federal employees from willfully inspecting taxpayer 
information without authorization. To comply with the law, IRS 
restricts the access examiners have to certain tax information. 
According to IRS, the law helps protect the confidentiality of 
taxpayer information, but examiners told us it also may restrict an 
examiner's flexibility to explore leads, without manager approval, 
across different tax forms that could reveal network abuse. 

IRS's Recent Efforts to Better Detect and Pursue Network Tax Evasion 
Show Promise, but Opportunities Exist for Additional Progress: 

IRS Is Developing Programs and Tools Intended to Help Address Network 
Tax Evasion: 

IRS has been creating specific programs and tools that address network 
tax evasion more directly than its traditional examination approach. 

Global High Wealth Industry (GWHI): 

Under GHWI, IRS identifies certain high-wealth individuals and then 
examines each individual's network. According to the Commissioner of 
Internal Revenue, the intent is to take a unified look at the entire 
complex web of business entities controlled by a high-wealth 
individual to assess the tax compliance of the network, rather than of 
the separate entities individually. IRS initiated GHWI in 2009. 
Although it resides in LMSB, IRS plans for GHWI to include staff with 
expertise that crosses divisional boundaries. For example, GHWI 
examiners might address small partnerships included in a network, even 
though small partnerships otherwise would be under the purview of 
SB/SE. As a result, GHWI is expected to directly examine tax issues 
that otherwise might have been missed. 

Enhancing Ownership Transparency (EOT): 

SB/SE launched the EOT project in January 2010 to gather data on the 
owners and locations of new businesses through employer identification 
number (EIN)[Footnote 16] applications. Primarily, IRS officials said 
they are interested in identifying what they refer to as the 
responsible party, which is the true beneficial owner of the business 
in this context. As of January 2010, the EIN application form requests 
additional information from business owners, such as the Social 
Security number of the responsible party and location of 
incorporation, which IRS previously did not request. The goal of the 
project is to link the new data to existing information in IRS's 
databases for identifying related businesses or a network of 
businesses. As the operating division responsible for the EIN process, 
W&I will implement the program once design is complete, which is 
tentatively scheduled for January 2012. 

yK-1: 

The network tool that is furthest along in development and most widely 
used at IRS is yK-1. yK-1 is a network visualization tool that is now 
being used by some IRS staff in doing examinations and in reviewing 
networks. Users enter a taxpayer identification number (TIN)[Footnote 
17] into the yK-1 software, which produces a picture showing how that 
TIN is connected to other entities through information filed on 
Schedule K-1. Figure 3 shows an example of yK-1 output. In this 
example, the numbers along the arrows represent the flow of money 
among the three different types of entities. 

Figure 3: Example of yK-1 Output Graphic: 

[Refer to PDF for image: illustration] 

The illustration depicts the following specific entities: 

Partnership (1); 
S corporation (2); 
Individual (4). 

Source: GAO adaptation of IRS figure. 

[End of figure] 

yK-1 diagrams can help examiners and other yK-1 users determine a 
specific entity's sources and amounts of income and whether other 
entities in the taxpayer's network need further examination. Examiners 
and users can then access other tax information about the entities in 
the network from IRS databases as well as non-IRS information from 
sources such as Accurint, a financial information database, and the 
Internet. Programmers are continuing to work on expanding yK-1's 
capabilities, such as adding estate and gift tax data and data on 
international taxpayers. 

GraphQuery: 

IRS's Research, Analysis and Statistics group is developing another 
tool related to yK-1 called GraphQuery. GraphQuery is a pattern- 
matching tool being designed to facilitate top-down identification and 
selection of those networks that have the highest risk for 
noncompliance. In this top-down approach, users would enter into 
GraphQuery a specified pattern, such as the structure of a known tax 
evasion scheme; the program would search for other networks showing 
the same pattern and list the TINs of the entities in those networks. 

NetReveal: 

IRS's Office of Performance Evaluation and Risk Analysis has 
demonstrated a program called NetReveal, which can build a diagram of 
related entities or individuals using a wider variety of data, 
including nontax return data, than are used by yK-1. NetReveal, which 
is unrelated to CI's Reveal program, remains under consideration by 
IRS and has not yet been made available to the operating divisions for 
tax compliance purposes. 

IRS's Ongoing Development of New Programs and Tools Is Generally 
Consistent with Network Analysis Criteria but Further Progress May Be 
Hindered by the Lack of a More Strategic Approach: 

Judged according to 14 network analysis criteria we developed, IRS's 
work on creating new network programs and tools already shows promise. 
We used interviews with academic experts and users of network analysis 
programs at other federal agencies to develop the 14 criteria, which 
are listed in table 3, for assessing network analysis programs and 
tools. Appendix I discusses what the criteria entail and how we 
developed them. 

While the criteria describe good management practices that could apply 
to a wide variety of programs, the experts we spoke with cited these 
criteria as directly relevant to network analysis. In particular, the 
criteria highlight the crosscutting nature of network-related 
problems. The experts we spoke with also noted that network problems, 
such as network tax evasion, include by definition multiple entities 
that could cut across databases and an oversight agency's 
organizational units. As a logical consequence, they emphasized using 
criteria that call for an agencywide strategy, access to a wide range 
of data, and good collaboration across an agency's organizational 
structures. 

For IRS, criteria focused on the crosscutting nature of network 
analysis are directly relevant to the problem of network tax evasion. 
A variety of entities could comprise a network, which could be under 
the purview of different IRS divisions. Similarly, data on the tax 
accounts for the different entities could be in different IRS 
databases. 

Our assessment of IRS's new programs and tools against the criteria is 
shown in table 3. The assessment is of IRS's progress to date--the 
programs and tools are still under development. The assessment also 
indicates areas where IRS's efforts to date do not satisfy the 
criteria. These areas present opportunities to make further progress. 

Table 3: IRS Progress in Meeting Network Analysis Criteria: 

Strategy: 

Criteria: Agencies should have clear strategies with goals for network 
analysis programs; 
IRS progress: 
* IRS recognizes the need for new approaches to network analysis and 
has various efforts underway throughout the agency; 
* Pursuing tax shelters, which may involve networks, is part of IRS's 
strategic plans, but networks are not explicitly mentioned; 
* IRS has no agencywide strategy that coordinates or sets goals for 
the various network efforts. 

Criteria: Network analysis programs should have initial and ongoing 
support from upper management; 
IRS progress: 
* Management directives to assess network risk led to GHWI; 
* Management has publicly cited networks as a compliance problem and 
supports additional network efforts; 
* Management has not developed an IRS-wide resource investment plan 
for network efforts. 

Criteria: Agencies running network analysis programs should plan for 
ongoing program development; 
IRS progress: 
* Programmers update yK-1 based on user feedback; 
* GraphQuery is being tested and developed on a time-available basis 
without a date for being operational; 
* GHWI is still developing, including acquiring additional staff, 
identifying ways to select workload, and developing measures for 
assessing their work. 

Criteria: Network analysis plans and related analytical tools should 
be flexible to adapt to emerging issues, including changes in network 
structures; 
IRS progress: 
* GraphQuery's pattern recognition capability could be applied to new 
types of tax schemes in the future, if implemented; 
* yK-1 includes data on the various types of recipients and senders of 
Schedules K-1, giving it the flexibility to trace different kinds of 
network structures; 
* IRS officials told us that legacy computer systems can make data 
inflexible, requiring adjustments before it can be used in new tools 
and programs. 

Criteria: The network analysis program should serve across internal 
organizational structures and avoid creating barriers between those 
structures; 
IRS progress: 
* GHWI crosses examination groups in three divisions but resides in 
LMSB; 
* yK-1 draws on data that can be used across divisions; 
* Typically, new network programs and tools start in one of the 
operation divisions. 

Criteria: Agencies should develop goals, measures, and methods for 
assessing network analysis program effectiveness; 
IRS progress: 
* IRS has no plans to evaluate from a network perspective existing 
tools, such as yK-1, or programs, such as GHWI, in addressing network 
tax evasion; 
* Existing examination performance measures do not take into account 
the additional time needed to examine a network's multiple entities. 

Programming and data: 

Criteria: Network analysis programs should use data from multiple 
sources; 
IRS progress: 
* yK-1 links entities just using Schedule K-1 data and IRS is 
considering other data to expand the linkages; 
however, combining external data with IRS's tax return data would 
involve significant administrative efforts, according to IRS officials; 
* IRS is piloting a program that allows state tax authorities to use 
yK-1, which may potentially reveal noncompliance at the federal level. 

Criteria: Network analysis programs should have access to the most 
current and complete data available; 
IRS progress: 
* Tax year 2008 Schedule K-1 data were updated to yK-1 in May 2010, 
several months earlier than the updates for previous tax years; 
* IRS plans to move responsibility for database administration, 
including Schedule K-1 databases, to the Modernization and Information 
Technology Services group, which may allow for earlier and more 
frequent updates; 
* According to EOT program officials, EOT intends to enhance the 
business ownership data that IRS has, which yK-1 users will be able to 
use to connect related entities in a network; 
* IRS management has decided that network analysis is not browsing. 
Therefore, GHWI and yK-1 users are not restricted from pursuing leads 
on related entities; 
* All data from paper-filed Schedules K-1 are not scanned 
electronically due to costs, which limits examiners' access to 
complete Schedule K-1 data that they told us would be helpful. 

Criteria: Network analysis programs should use historical data to test 
program's ability to identify known problems; 
IRS progress: 
* yK-1 users can use historical data to identify gaps in return 
filings once an examination is initiated, but such data are not used 
regularly to test yK-1's ability to identify known problems. 

Criteria: Network analysis programs should use a variety of research 
disciplines and analytical techniques; 
IRS progress: 
* yK-1 largely uses network visualization, and IRS is considering 
using other techniques, such as pattern matching in GraphQuery; 
* IRS is hiring specialists with expertise in pass-through entities to 
work on network cases in GHWI; 
* None of IRS's network analysis tools that are operational start with 
a network as a unit of analysis. 

Criteria: Network analysis programs should use data and analytical 
tools that match the problem to be addressed; 
IRS progress: 
* The new programs and tools are intended to move beyond IRS's 
traditional focus on single entities; 
* yK-1 helps auditors identify related entities through Schedules K-1, 
but other types of connections exist that Schedule K-1 data alone 
cannot be used to identify; 
* Although burdens and costs limit collecting all potentially useful 
data, examiners said they do not have access to data, such as some 
trust data, that they need to address the problem of identifying the 
many ways networks can be connected. 

Collaboration: 

Criteria: Network analysis programmers, analysts, and subject matter 
experts should have frequent and ongoing communication about user 
needs and program capabilities throughout development and continued 
operation of the program; 
IRS progress: 
* yK-1 developers consulted with auditors when creating yK-1; 
* yK-1 users can submit feedback to programmers directly or in an 
annual user survey; 
* IRS does not have a formal mechanism to solicit views from nonusers 
about their needs; 
* GraphQuery is primarily being developed by one staffer with limited 
input from potential users; 
* IRS has no written expectations on users and programmers sharing 
views about the development of new network programs or tools. 

Criteria: Analysts and subject matter experts should have access to 
the program tools or reports generated by program analysts without 
extensive delays or burdensome training; 
IRS progress: 
* IRS offers training for new yK-1 users; 
* Some yK-1 users we spoke with were not aware of the program's full 
range of functions; 
* IRS staff told us that some examiners are unaware that yK-1 is 
available to them. 

Criteria: Network analysis programs should have a dedicated program 
team; 
IRS progress: 
* GHWI seeks to bring together experts from across IRS because, unlike 
a dedicated program team, each IRS division alone lacks complete 
technical or analytical expertise on network tax evasion; 
* IRS does not have a dedicated team responsible for coordinating 
across IRS's multiple network analysis efforts. 

Source: GAO analysis of IRS documentation and interviews. 

[End of table] 

As table 3 indicates, IRS's efforts to focus more directly on network 
tax evasion, while still under development, are consistent with our 
criteria for judging network analyses but do not fully satisfy them. 
The efforts are supported by upper management, offer new analytical 
approaches that more directly address network tax evasion, and attempt 
to cut across IRS's divisional boundaries and databases. As a result, 
these efforts show promise at being able to detect and pursue network 
tax evasion more effectively than IRS's traditional enforcement 
programs. 

However, table 3 also shows where opportunities exist to provide more 
overall direction to IRS's efforts and perhaps hasten the development 
of specific programs and tools. For example, the table notes the lack 
of agencywide strategy and goals for IRS's various network efforts 
that are spread throughout the agency. Without agencywide strategy or 
goals to coordinate and prioritize these efforts, two risks exist. 
First, IRS could make redundant investments; second, IRS could fail to 
concentrate investments on the programs and tools with the greatest 
potential. 

IRS may need to take an incremental approach to managing these risks 
because of the uncertainties. As already discussed, the population of 
networks is not known, networks can be complex, and IRS does not know 
which programs and tools will be most effective. Further, the costs 
could be significant. IRS's current organizational structure, work 
processes, and data systems do not support using a network as a unit 
of analysis and adjusting them to do so could disrupt other important 
priorities and programs. In light of this uncertainty about potential 
benefits and the cost, IRS will need to be careful in reallocating 
resources from other compliance programs to its new network efforts. 
As IRS gathers more information, management will be better positioned 
to more fully develop its strategy. 

IRS also faces challenges in responding to the criteria for 
programming and data. As noted in table 3, adding new data, updating 
existing data, and making existing data more readily available in 
electronic form all could enhance IRS's capabilities to identify and 
pursue network tax evasion. Similarly, IRS could potentially benefit 
from more complete consideration of the potential relevance of the 
array of analytical techniques developed in the research literature 
and available in existing software applications.[Footnote 18] However, 
as also noted above, such efforts would have costs. This reinforces 
the need for a strategy that would prioritize investments in better 
data and analytical capabilities. 

IRS has not assessed the impact and effectiveness of its new network 
analysis tools, as described in table 3. For example, the benefits of 
using yK-1 relative to the additional time it takes examiners to use 
it have not been studied, but anecdotal evidence from users and 
management indicate yK-1 is a useful tool. Effectiveness assessments 
have costs which can be managed by judgments about the depth of the 
assessment needed. However, without effectiveness assessments, IRS 
managers are left without information that could help with the 
development of a strategy and with decisions about prioritizing 
investments in better data. 

Table 3 stresses the importance of regular communications and training 
among all types of staff involved in identifying, analyzing, and 
pursuing network evasion schemes. Without these, auditors could be 
missing information they need and network schemes could go undetected. 
IRS officials said that the direct costs for these actions tend to be 
minor, but that they must be mindful of how these actions might affect 
other priorities. For example, they said the initial 2-hour training 
for yK-1 imposes minor costs. However, the officials also said that 
learning yK-1 requires accessing it enough to appreciate all of its 
capabilities. 

Conclusions: 

IRS's network compliance efforts have the potential to address a 
significant part of the tax gap. However, IRS does not know the extent 
of this compliance problem or how effective its new programs and tools 
will be. Nor does IRS have a strategic approach to coordinate these 
network efforts across the agency. 

IRS needs to walk a middle course between doing too much too soon 
versus doing too little too slowly. If it does too much, IRS risks 
taking resources from other priorities without assurance that the 
investment in the network efforts will reduce network tax evasion. If 
it does too little, IRS runs the risk of not learning more about 
networks and how to detect their tax evasion schemes. 

To successfully balance these trade-offs, IRS would benefit from 
having a more strategic approach to coordinate and focus its various 
network efforts across the agency. As IRS learns more, that strategic 
approach would work toward developing a set of goals and measures to 
guide future efforts and consider ways to assess the impacts of the 
various programs and tools that are to be developed. Effectiveness 
assessments have costs, but without any assessments, managers lack 
valuable information. With such information, IRS would have a better 
sense of the pace at which it should invest its resources into 
expanding the network analysis program, including adding the 
analytical tools, data, and staff expertise that would be needed to 
address the specific compliance issues that IRS would be discovering. 
IRS's efforts to develop network programs and tools would also be 
enhanced by ensuring that staff understand the benefits of using the 
tools and are provided with a mechanism to provide feedback on the 
tools' and programs' effectiveness. 

Recommendations for Executive Action: 

We recommend that the Commissioner of Internal Revenue take the 
following three actions. 

* Establish an IRS-wide strategy with goals, which may need to be 
developed incrementally, to coordinate and plan ongoing and future 
efforts to identify and pursue network tax evasion. The strategy 
should include: 

- assessing the effectiveness of network analysis tools, such as yK-1; 

- determining the feasibility and benefits of increasing access to 
existing IRS data, such as scanning additional data from Schedule K-1, 
or collecting additional data for use in its network analysis efforts; 

- putting the development of analytical techniques and tools that 
focus on networks as the unit of analysis, such as GraphQuery, on a 
specific time schedule; and: 

- deciding how network efforts will be managed across IRS, such as 
whether a core program team or management group is needed. 

* Ensure that staff members who will be using current and additional 
network tools fully understand the tools' capabilities. 

* Establish formal mechanisms for front-line users to interact 
directly with tool programmers and program analysts to ensure future 
network analysis tools, such as GraphQuery, are easy to use and help 
achieve goals. 

Agency Comments and our Evaluation: 

In a September 8, 2010, letter responding to a draft of this report 
(appendix IV), IRS's Deputy Commissioner for Services and Enforcement 
provided comments on our findings and recommendations as well as 
information on additional agency efforts, changes, and studies to 
address network tax noncompliance. The letter also provided technical 
comments that we incorporated into our report as appropriate. 

The Deputy Commissioner said that IRS agreed with our draft on the 
challenges involving network tax compliance and the status of IRS's 
network-related efforts. IRS also generally agreed with our 
recommendations to establish an IRS-wide strategy with goals, ensure 
that staff understand the capabilities of IRS's network tools, and 
establish a more formal way for IRS staff to collaborate as new 
network tools are developed and implemented. 

In agreeing with our strategy recommendation, IRS's response noted 
that it may be more effective for IRS to consciously and appropriately 
include network issues in broader strategic plans, rather than develop 
a separate strategy for networks. We agree. Our recommendation is that 
IRS develops a strategy. We leave IRS with the discretion on how to 
articulate the strategy and point out that it may need to be developed 
incrementally. 

As agreed with your offices, unless you publicly release the contents 
earlier, we plan no further distribution of this report until 30 days 
from its date. At that time, we will send copies to interested 
congressional committees, the Secretary of the Treasury, the 
Commissioner of Internal Revenue, and other interested parties. The 
report will also be available at no charge on the GAO Web site at 
[hyperlink, http://www.gao.gov]. 

If you or your staff have any questions about this report, please 
contact me at (202) 512-9110 or whitej@gao.gov. Contact points for our 
Offices of Congressional Relations and Public Affairs may be found on 
the last page of this report. GAO staff who made major contributions 
to this report are listed in appendix V. 

Signed by: 

James R. White: 
Director, Tax Issues Strategic Issues: 

[End of section] 

Appendix I: Objectives, Scope, and Methodology: 

The objectives of this report were to (1) describe what the Internal 
Revenue Service (IRS) knows about network tax evasion and how well 
IRS's traditional enforcement efforts address network tax evasion and 
(2) assess IRS's progress in addressing network tax evasion and 
opportunities, if any, in making further progress. 

To describe what IRS knows about network tax evasion and how well the 
traditional enforcement efforts the agency has in place address 
network tax evasion, we reviewed IRS statistics, policy manuals, and 
planning documents, including strategic plans. We also interviewed 
relevant IRS officials and staff. 

We developed a video to highlight one type of network tax evasion. To 
develop this video describing the installment sale bogus optional 
basis adjustment (iBOB), we reviewed IRS technical information on iBOB 
schemes, out of which we developed a simplified, hypothetical example. 
IRS suggested the iBOB as a scheme that would make an appropriate 
example. IRS management and technical staff reviewed the video 
throughout its development, and we incorporated their technical 
comments where appropriate. 

Because no existing criteria that we could find directly applied to 
reviewing the progress of IRS's network analysis programs, we 
developed our own. We conducted two groups of interviews with network 
analysis users and experts to develop our criteria for network 
analysis program development and implementation. The first group of 
interviews was with academic researchers considered to be experts whom 
we identified through detailed literature reviews and recommendations 
from other experts. The second group of interviews was with federal 
agencies that use network analysis tools. We also interviewed IRS 
auditors about their work. 

To identify relevant academic experts, we reviewed the research 
literature using network analysis and related methods. We then created 
a literature search matrix and entered all studies obtained through 
the search that involved some quantitative/automated form of network 
analysis and an empirical application to a substantive area that had 
potentially direct applications to our review. We selected a subset of 
these studies for a more detailed review and used professional 
judgment to focus on studies of most immediate relevance. The 
literature review was the primary tool used for selecting researchers 
and experts for further follow up, which was ultimately based on 
ensuring a balance of experts with expertise across the entire array 
of substantive research topics and methodological approaches that we 
identified in our search and on determining that individual experts' 
research agendas were both broad and deep. We conducted semistructured 
interviews with these experts, during which time we asked for 
recommendations of other network analysis and data mining experts. 
Often these recommendations were for researchers or experts we had 
already identified. Not every expert we identified was available to 
speak with us. The experts we spoke with included Wayne E. Baker, 
University of Michigan; Stephen P. Borgatti, University of Kentucky; 
Kathleen M. Carley, Carnegie Mellon University; Sean Everton, Naval 
Postgraduate School; Mark Granovetter, Stanford University; David 
Jensen, University of Massachusetts Amherst; Mark Mizruchi, University 
of Michigan; Carlo Morselli, University of Montreal; Daniel M. 
Schwartz, Criminal Intelligence Service Ontario; Duncan Watts, Yahoo! 
Research; and Jennifer Xu, Bentley University. We used our interviews 
with these experts to aid only in developing our criteria; they did 
not otherwise contribute to the content of the report. 

To identify federal agencies to interview, we first reviewed academic 
literature and reports on government agencies that conduct network 
analyses, including our own reports. Through this review, we 
identified Customs and Border Patrol (CBP); Federal Bureau of 
Investigation; Financial Crimes Enforcement Network; Immigration and 
Customs Enforcement; Risk Management Agency at the United States 
Department of Agriculture; and the Securities and Exchange Commission. 
We also identified the Financial Industry Regulatory Authority 
(FINRA), which is not a federal agency but has an oversight and 
enforcement component similar to that of federal financial regulators. 
We conducted semistructured interviews with relevant program staff 
that use network analysis tools at these agencies and FINRA. During 
each interview, we asked about what works well in their network 
analysis program; what about their network analysis program needs 
improvement; what tools they feel could improve their program; what 
are best practices for developing a network analysis program; and what 
other agencies use network analysis programs. Of those other agencies 
that were mentioned that had network analysis programs, we chose not 
to meet with the Central Intelligence Agency and National Security 
Agency due to time constraints and data sensitivity issues. The Drug 
Enforcement Agency (DEA) was also recommended to us to speak with; 
while we did not directly speak with DEA due to time constraints, we 
were able to speak with a DEA liaison at CBP who briefly described 
DEA's network analysis program. 

From these two rounds of interviews, we distilled the common themes in 
those responses to establish the criteria. We first read through all 
the interviews, recording potential criteria. We then systematically 
reviewed the entire set of interviews to identify all that contained 
our initial criteria; this resulted in the rephrasing or elimination 
of some of these criteria, as well as the addition of a number of new 
ones. The themes that emerged from the interviews fell into the 
following categories: strategy, management support, program 
evaluation, data management, staffing, collaboration, methodology, and 
other. We determined that for our review of IRS, it would not be 
appropriate to set criteria for the exact methodology the agency 
should use in its network analysis program, or particular software 
packages. Therefore, we eliminated any particular research or 
methodological approaches and techniques from our final criteria list. 
[Footnote 19] We also eliminated ideas where there was a clear 
division of opinion among the experts we interviewed. For example, 
experts and users did not agree on the benefits of including narrative 
data in network analysis programs. We presented the criteria to IRS 
for its feedback. 

The final 14 criteria were categorized by theme: overall strategy; 
programming and data; and collaboration. The criteria were neither 
prioritized nor made to be specific to IRS. Each criterion was 
supported, at minimum, by five interviews; many were supported by 
eight or more interviews. The criteria with the fewest interviews for 
support generally pertain to organizational structure issues. The 
academic experts generally did not address these issues because they 
tend to use network analysis programs for research purposes compared 
to a federal agency's use for enforcement purposes. In these 
instances, we also had support from prior GAO work. 

To assess IRS's progress in identifying and addressing network tax 
noncompliance, we reviewed IRS documentation on its auditing 
procedures and interviewed officials involved with identifying and 
addressing noncompliance related to networks. We then compared the 
evidence we collected in these reviews and interviews with the 
criteria we had developed and identified specific instances where IRS 
has demonstrated progress towards meeting each of the criteria. We 
also identified opportunities for further progress in meeting the 
criteria. We determined for the purposes of this review that the data 
used were reliable. Because some of the efforts to address 
noncompliance were under development during the time of our review, we 
presented the assessment to IRS officials for their feedback and for 
related updates that might affect the assessment. 

Our review of key studies applying quantitative network analysis 
methods to areas of noncompliance or illicit activity (see appendix 
III) focused on identifying analytical approaches that individuals 
developing network analysis systems may find useful. The criteria for 
selection of these studies were similar to those used in selecting 
experts. In particular we ensured that the studies taken as a whole 
group covered a broad set of topics and methodologies and also that 
the individual studies were supported by broad and deep individual 
research agendas. We included two additional criteria to ensure more 
direct applicability of the studies to the report topic. 

* The selected studies applied network analysis directly to a specific 
set of activities most directly related to the report topic, 
particularly criminal intelligence, organized crime, fraud detection, 
public safety or security, and international trafficking. 

* Studies focused on counterterrorism applications of network analysis 
and studies in the link analysis research area, which is focused on 
algorithms for identifying relationships among items in large 
databases of textual/narrative information, were largely excluded. 

We conducted this performance audit from June 2009 through September 
2010 in accordance with generally accepted government auditing 
standards. Those standards require that we plan and perform the audit 
to obtain sufficient, appropriate evidence to provide a reasonable 
basis for our findings and conclusions based on our audit objectives. 
We believe that the evidence obtained provides a reasonable basis for 
our findings and conclusions based on our audit objectives. 

[End of section] 

Appendix II: Example Diagrams of Network Relationships by Entity Type: 

A network may comprise a variety of entities. Four types of entities 
that IRS recognizes that also can form networks are corporations, 
partnerships, trusts, and individuals. While these are not the only 
types of entities or connections that may exist in networks, they are 
entities that the Internal Revenue Service (IRS) has emphasized. The 
following examples show how networks can be connected and how income 
can flow among different entities. These examples are hypothetical and 
any resemblance with a known network is purely coincidental. 

Figure 4 shows a network that includes two C corporations and two 
partnerships. Here, income and tax attributes (such as expenses, 
deductions, and losses) earned by the partnerships would flow back to 
C corporation B. However, the extent that C corporation A might also 
have received income from C corporation B depends on their business 
arrangements. S corporation A and Individuals A and B represent other 
partners that overlap with C corporation A's network. 

Figure 4: Example of a Network with C Corporations: 

[Refer to PDF for image: illustration] 

Source: GAO analysis of IRS documents. 

[End of figure] 

S corporations are pass-through entities that pass income on to 
shareholders. In figure 5, Individual A receives income and other tax 
attributes earned by S corporations A and B and Partnership A. 
Likewise, Individuals B and C receive income and tax attributes earned 
by S corporation C, which in turn receives income from Partnership B. 
S corporation A passes income on to all three individuals in this 
example. 

Figure 5: Example of a Network with S Corporations: 

[Refer to PDF for image: illustration] 

Source: GAO analysis of IRS documents. 

[End of figure] 

Figure 6 shows how partnerships can be layered and seemingly 
unconnected individuals can be connected. Individuals A and D have no 
direct connection but both ultimately receive income and other tax 
attributes from Partnership A. Individuals A through D are also 
connected to Individual E because of the financial ties between 
Individuals D and E through Partnership X. 

Figure 6: Example of a Network with Partnerships: 

[Refer to PDF for image: illustration] 

Source: GAO analysis of IRS documents. 

[End of figure] 

Trusts can be connected to other entities in a network in several 
ways. In figure 7, Trusts A and B are partners in Partnership B and, 
along with Individual A, receive income and tax attributes from 
Partnership B. Trust A is a type of trust that is allowed to own 
shares in S corporation X, which passes income and tax attributes to 
its beneficiary, Individual B. Trust C is a partner in Partnership C 
and, along with Trust B, sends its income and tax attributes to 
Individual D. 

Figure 7: Example of a Network with Trusts: 

[Refer to PDF for image: illustration] 

Source: GAO analysis of IRS documents. 

[End of figure] 

[End of section] 

Appendix III: Overview of Studies Using Formal Network Analysis to 
Examine or Detect Criminal/Illicit Activity: 

This appendix presents examples of research that have used formal 
quantitative network analysis techniques and methods to analyze 
network noncompliance, criminal, or illicit activity. Individuals 
developing network analysis programs may find relevant analytical 
approaches and techniques from this research. The research is 
summarized under four approaches to network analysis of illicit or 
criminal behavior. Key technical terms used in this summary include 
the following. 

* Network: A set of actors and the set of ties representing some 
relationship or lack of relationship between the actors. 

* Centrality: The number of direct contacts between a given network 
member and all other network members.[Footnote 20] 

* Brokerage/network efficiency: Extent to which network members' 
connections are not to each other or are not within the same group 
(nonredundancy).[Footnote 21] 

* Cut-point: A network member that serves as the only connection among 
people or groups of people within a network. 

* Key players: The set of network members whose removal will most 
disrupt or who can most efficiently diffuse information through a 
network. 

* Density: The proportion of existing links out of all possible links 
in a network. 

* Centralization: Extent to which a network is organized around a few 
central members. 

* Clustering: Extent to which a network is subdivided into distinct, 
heavily interconnected subgroups. 

* Connectedness: Extent to which all network members can reach each 
other along unbroken paths. 

* Hierarchy: Extent to which the links in a network flow in one 
direction.[Footnote 22] 

* Chain: A network structure where a high proportion of network 
members can only reach each other via some other network member. 

I. Interpretive Approaches Using Visualization or Core Network 
Measures to Explain Criminal or Illicit Activity: 

These studies use network visualization and measurement techniques to 
develop qualitative explanations of illicit/criminal behavior. This 
research has two major implications for network tax noncompliance. 

First, these studies suggest that both individual-and network-level 
measures of position and structure may be useful diagnostics for 
identifying criminal/illicit behavior. For example, higher centrality 
of individuals in a network may be associated with higher levels of 
criminal activity. Further, high levels of brokerage/network 
efficiency may be associated with a leadership role in a criminal 
network and with success in criminal activity. At the network level, 
chain structures often characterize criminal networks, though more 
interconnected structures (measured as density) in criminal networks 
may correspond with higher-risk activities. 

Second, tracking how network measures change may provide insight into 
how criminal behavior evolves. One study suggests that network leaders 
more effectively disguise involvement in criminal networks over time 
by using indirect relationships. Changes in the structure of 
noncompliance networks may correspond to changes in strategy and 
management. For example, management crisis in a noncompliance network 
may produce new relationships between previously disconnected 
individuals or change the extent of hierarchy in the network. 

Table 4: Key Studies Using Visualization or Core Network Measures to 
Explain Criminal or Illicit Activity: 

Terrill L. Frantz and Kathleen M. Carley, "Organizational Response to 
the Turmoil of Personnel Turnover," Center for Computational Analysis 
of Social and Organizational Systems, Dynamic Networks Project (n.d.). 

Jana Diesner, Terrill L. Frantz, and Kathleen M. Carley, 
"Communication Networks from the Enron Email Corpus: It's Always about 
the People, Enron is no Different," Computational and Mathematical 
Organization Theory, vol. 11: 201-228 (2005). 
The studies use Enron Corpus data[A] from 1999-2002 to analyze the 
network response to crisis and management changes. They analyze 
changes in connectedness, hierarchy, efficiency of the networks, and 
patterns of downward, lateral, and upward communications between 
groups of individuals at different levels and functions of the 
organization. 

Jennifer Xu, Byron Marshall, Siddharth Kaza, and Hsinchun Chen, 
"Analyzing and Visualizing Criminal Network Dynamics: A Case Study" 
(paper presented at the 2nd NSF/NIJ Symposium on Intelligence and 
Security Informatics, Tucson, AZ, 2004). 
Study uses Tucson Police Department data on a large narcotics network 
over nine years to analyze the relationship between criminal behavior 
and network centrality of two key network leaders. It tracks the 
relationship of the leaders' activity levels and network position over 
time. 

Gerben Bruinsma and Wim Bernasco, "Criminal Groups and Transnational 
Illegal Markets," Crime, Law and Social Change, vol. 41: 79-94 (2004). 
Study compares the characteristics of criminal networks in three 
illegal transnational markets in Europe: heroin smuggling, trading 
stolen cars, and trafficking in women. Network characteristics 
include: cohesion, clustering (cliques), bridging, and overall chain 
structure. The relationship between network structure and the risk 
levels of the criminal activities is assessed. 

Carlo Morselli, "Structuring Mr. Nice: Entrepreneurial Opportunities 
and Brokerage Positioning in the Cannabis Trade." Crime, Law and 
Social Change, vol. 35: 203-244 (2001). 
Study examines how Howard Marks, an international drug trafficker, 
used network brokerage strategies to build his criminal career, using 
data in his autobiography. Study focuses on associations between 
brokerage (i.e. network efficiency) and key outcomes (income, shipment 
size, and arrests/judicial sentences) over his three career phases 
(Building-Attainment-Fall). 

Source: GAO research database search. 

[A] The Enron Corpus is a dataset created by the Federal Energy 
Regulatory Commission in 2002 composed of 619,449 emails from 158 
Enron employees. 

[End of table] 

II. Approaches Using Visualization or Core Network Analysis Measures 
for Intervention and Enforcement in Criminal Networks: 

Numerous studies have used core network visualization and measurement 
techniques to develop interventions into criminal networks for 
enforcement purposes. A key implication is that intervention decisions 
should arise from analyzing network structures and processes. These 
studies suggest that effectively identifying intervention strategies 
may require varied network analysis approaches, ranging from basic to 
complex. 

Basic network measures and constructs such as centrality or cut-points 
may effectively identify interventions in some contexts but not 
others. For example, in some contexts, using algorithms to find sets 
of key players may enable more efficient disruption or surveillance of 
criminal networks than approaches using centrality measures and cut- 
points. Extending the key player approach by incorporating data on 
individuals' or entities' attributes also may help. Approaches 
identifying cohesive subgroups and clusters, including the presence or 
absence of links between them, may suggest effective interventions. 
For example, if a network has disconnected or weakly connected 
subgroups that are themselves heavily connected, it may be appropriate 
to focus on the more cohesive subgroups, rather than on central 
individuals across the network. Relatedly, it may be more productive 
to disrupt decentralized criminal networks than more centralized ones. 

Table 5: Key Studies Using Visualization or Core Network Measures for 
Intervention and Enforcement in Criminal Networks: 

Carlo Morselli and Katia Petit, "Law-Enforcement Disruption of a Drug 
Importation Network," Global Crime, vol. 8: 110-130 (2006). 
Study examines how a drug importation network is decentralized and re-
ordered by intense law-enforcement. Study uses electronic surveillance 
data from a 2-year investigation of hashish and cocaine distribution 
chains. It focuses on how the network responds to the removal of a 
central leader and to the enforcement opportunities that arise as 
criminal networks decentralize. 

Jean Marie McGloin, "Policy And Intervention Considerations of a 
Network Analysis of Street Gangs," Criminology and Public Policy, vol. 
4: 607-636 (2005). 
Study examines how using cohesion analysis (cliques) and "cut-points" 
can help develop anti-gang intervention programs. Data are from 32 
group interviews at criminal justice agencies in Northern New Jersey. 
The study points to possible effective interventions given particular 
levels of cohesion in separate network sub-groups and when prominent 
cut-point nodes connect the sub-groups. 

Stephen P. Borgatti, "Identifying Sets of Key Players in a Social 
Network," Computational and Mathematical Organization Theory, vol. 12: 
21-34 (2006). 
Study develops an optimal way to find key players in a social network 
for enforcement and surveillance purposes. Two datasets are used: (1) 
data on a terrorist network; and (2) data on advice-seeking ties in a 
global consulting company. The study develops a graph fragmentation 
measure (for identifying individuals whose removal would most disrupt 
the network) and an inter-set cohesion measure (for identifying 
players who most efficiently diffuse information through a network). 

Daniel M. Schwartz and Tony (D.A.) Rouselle, "Using Social Network 
Analysis to Target Criminal Networks," Trends in Organized Crime, vol. 
12:188-207 (2009). 
Study extends the key player approach to make it more relevant to 
criminal enforcement and intelligence activity by incorporating data 
on player attributes[A] as well as on the level of uncertainty about 
their network roles. The study develops measures for maximum 
disruption and maximum reach into a network, using hypothetical data 
to illustrate them. 

Jennifer Xu and Hsinchun Chen, "Untangling Criminal Networks-A Case 
Study," Lecture Notes in Computer Science, vol. 2665: 232-248 (2003). 

Jennifer J. Xu and Hsinchun Chen, "CrimeNet Explorer: A Framework for 
Criminal Network Knowledge Discovery," ACM Transactions on Information 
Systems, vol. 23: 201-226 (2005). 
This research develops and assesses a criminal network analysis 
system, using multiple network analyses. They include centrality 
measures, shortest-path algorithms, subgroup analysis, hierarchical 
clustering, multidimensional scaling, and network visualization. The 
data come from the Tucson Police Department's narcotics and gang-
related crime incident summaries. Both studies use exercises with 
experts and students to validate the system's capabilities. 

Source: GAO research database search. 

[A] These attribute measures include the ability to corrupt public 
officials, propensity to use violence, and capacity to discipline 
members of the network. 

[End of table] 

III. Data-Mining and Related Approaches for Analyzing and Detecting 
Criminal/Illicit Activity: 

This research covers various approaches from the computational 
sciences, relying on data-mining, machine learning, and simulation 
techniques. The approaches develop methods that may help detect 
unknown aspects of noncompliance networks. Possibilities include the 
following: 

* relational machine learning approaches that identify systems of 
statistical relationships among attributes of network entities, such 
as individual roles, status and experience, and firms' locations, 
using complex relational data; 

* risk-assessment methods that use probabilistic models to identify 
firms and employees with a high risk for misconduct. For example, 
algorithms have been developed to identify individuals that are 
atypically moving together among locations or organizations, which may 
help assess non-compliance risk; 

* methods for identifying links, identities, or groups in a network. 
Link prediction, for example, uses machine-learning to identify 
unknown links in a network. An alternative method is anomalous link 
detection, which identifies links that are more likely to involve 
illicit/criminal/fraud activity. A third method is anonymous identity 
matching, which uses known relationships of unknown entities to 
predict their identities. Pattern matching or clustering algorithms 
can identify groups of entities occupying similar positions in the 
overall network; and: 

* dynamic simulation approaches that model networks' likely responses 
to varied interventions and assess the effectiveness of the 
interventions. 

Table 6: Key Studies Using Data-Mining and Related Approaches for 
Analyzing and Detecting Criminal/Illicit Activity: 

Andrew Fast, Lisa Friedland, Marc Maier, Brian Taylor, David Jensen, 
Henry G. Goldberg, and John Komoroske, "Relational Data Pre-Processing 
Techniques for Improved Securities Fraud Detection" (paper presented 
at the 13th ACM SIGKDD International Conference on Knowledge Discovery 
and Data Mining, San Jose, California, 2007). 
Study develops relational statistical models that identify firms and 
their employees at high-risk for misconduct. Data source is the 
Central Registration Depository (CRD), which has location, employment, 
and other data on federally registered firms and individuals. The 
study focuses on (1) finding "tribes" of employees who are at-risk for 
fraud due to atypical joint movements between branch offices, and (2) 
developing risk scores and models based on the finding tribes analysis 
and data on disciplinary actions against individuals. 

Matthew J. Rattigan and David Jensen, "The Case for Anomalous Link 
Discovery," SIGKDD Explorations: The Newsletter of the ACM Special 
Interest Group on Knowledge Discovery and Data Mining, vol. 7: 41-47 
(2005); 
Jennifer Neville and David Jensen, "Relational Dependency Networks," 
Journal of Machine Learning Research 8: 653-692 (2007). 
Both studies use probabilistic modeling and relational machine 
learning methods to identify data anomalies and statistical 
dependencies in complex relational data.[A] The first study explores 
how algorithms commonly used in computationally problematic link 
prediction tasks may be more effectively used in detecting anomalous 
links in relational data. The second study uses an iterative 
estimation method to identify statistical dependencies between 
variables in several relational databases. 

Shawndra Hill, "Social Network Relational Vectors for Anonymous 
Identity Matching" (paper presented at Workshop on Learning 
Statistical Models from Relational Data, International Joint 
Conference on Artificial Intelligence, Acapulco, Mexico, 2003). 
The study indicates that anonymous identity matching methods using 
relational data may help identify unknown actors in networks. Testing 
relies on the CiteSeer database, a scientific literature library. 

Brett W. Bader, Richard A. Harshman, and Tamara G. Kolda, "Temporal 
Analysis of Social Networks using Three-way DEDICOM." Sandia National 
Laboratories, SAND2006-2161 (2006). 
Using the Enron corpus data, the study develops a factor analysis-
based pattern-matching approach to analyze how employees' 
organizational roles and attributes are related. Models are produced 
that identify distinct clusters of employees and their relations to 
each other. 

Jeffrey Baumes, Mark Goldberg, and Malik Magdon-Ismail, "Efficient 
Identification of Overlapping Communities," Lecture Notes in Computer 
Science, vol. 3495: 27-36 (2005). 
The study develops an algorithm for identifying communities or 
clusters based on density within a network to discover groups of 
actors that hide their communications, possibly for malicious reasons. 
Unlike other network clustering approaches, these algorithms allow for 
overlap so that one node can belong in multiple clusters. The 
algorithms iteratively scan network members and their neighborhoods 
until an optimal collection of clusters is reached. The algorithm's 
efficiency is tested on random graphs as well as several datasets. 

Kathleen M. Carley and Daniel T. Maxwell, "Understanding Taxpayer 
Behavior and Assessing Potential IRS Interventions Using Multiagent 
Dynamic-Network Simulation," In J. Dalton and B. Kilss (Eds.), Recent 
Research on Tax Administration and Compliance: Selected Papers Given 
at the 2006 IRS Research Conference (2006); 
Kathleen M. Carley, "Destabilization of Covert Networks." 
Computational and Mathematical Organization Theory, vol. 12: 51-66 
(2006). 
These studies use dynamic-network analysis and multi-agent simulations 
to assess the effectiveness of interventions into illicit activity. 
The first study examines how variation in network structures 
conditions the effects of IRS ad campaigns on diffusion of 
information, change in beliefs, and participation in tax schemes. The 
second study examines how changes in network personnel affect 
information diffusion, depending on the structure of the network, the 
logic of interaction, and the presence of technology. 

Source: GAO research database search. 

[A] In a relational database, instances record the characteristics of 
heterogeneous objects and the relations among those objects. Examples 
of relational data include citation graphs, fraud detection data, and 
data on interrelated people, places, and events extracted from text 
documents. 

[End of table] 

IV. Multivariate and Statistical Analyses of Associations and Causal 
Relationships among Network Variables and Key Behavioral Outcomes: 

These studies use regression analysis or other multivariate 
statistical methods to examine collusion, unethical behavior, and 
adoption of illegal innovations. They emphasize statistically 
significant associations and causal relationships between measures of 
network position, overall structure, behaviors, and outcomes, while 
controlling for an array of individual and organizational attributes. 
Key implications include: 

* Comparing different types of networks by their structural 
characteristics can help identify illicit activity. For example, 
illegitimate networks may be more hierarchical than legitimate 
networks. A network's need for secrecy, typically associated with 
illicit activity, as well as its information-processing demands, may 
determine the degree of centralization exhibited by the network. 

* Important relationships between network and non-network variables 
can influence illicit behavior. For example, common codes of conduct 
may mitigate the likelihood that hierarchical or asymmetric 
relationships produce unethical behavior. While lower-status middle 
managers usually are the most vulnerable network participants, higher-
status upper-level managers may be more vulnerable in centralized 
networks than in decentralized networks. 

Table 7: Key Studies Using Multivariate and Statistical Analysis to 
Identify Associations and Causal Relationships among Network Variables 
and Key Behavioral Outcomes: 

D.J. Brass, K.D. Butterfield, and B.C. Skaggs, "Relationships and 
Unethical Behavior: A Social Network Perspective," Academy of 
Management Review, vol. 23: 14-31 (1998). 
This study examines the associations between key network measures, 
individual and organization attributes, and their combined effects on 
unethical behavior. The study generates propositions about the 
likelihood of unethical behavior given particular types of individual 
relationships as well as overall network structure[A] and 
actor/organizational attributes. It emphasizes the interaction effects 
between relationships and attributes (e.g., empathy, values, or the 
cost of losing a strong relationship) in influencing unethical 
behavior. 

Brandy Aven, "The Network Structure of Corrupt Innovation: The Case of 
Enron," Unpublished (2009). 
Study examines differences in structures (using measures such as 
connectedness and hierarchy) in legitimate and corrupt networks, using 
descriptive statistics and correlation analysis. It also uses 
regression analysis to compare how actor centrality and constraint 
affect employees' adoption of innovations in legitimate and corrupt 
networks. The data come from the Enron Email corpus from 1998 to 2002. 

Wayne E. Baker and Robert R. Faulkner, "The Social Organization of 
Conspiracy: Illegal Networks in the Heavy Electrical Equipment 
Industry," American Sociological Review, vol. 58: 837-860 (1993). 
This study examines how core network measures[B] affect enforcement 
and judicial outcomes (i.e. verdict, sentence, and fine) in three 
conspiracies in the heavy electrical equipment industry. It also 
examines how information-processing and secrecy needs determine the 
structure of corrupt networks. The data source is sworn testimony 
before the U.S. Senate Committee on the Judiciary. 

Source: GAO research database search. 

[A] Measures examined include tie strength, multiplexity, asymmetry, 
structural holes, closeness centrality, density, and cohesion. An 
example of one proposition at the level of the entire network is: "The 
effects of the constraints on unethical behavior of the density of 
relationships within a group will increase as the constraints of group 
norms, social consensus, and codes of conduct increase." 

[B] These include degree, betweenness, and closeness centrality, 
density, and centralization. 

[End of table] 

[End of section] 

Appendix IV: Comments from the Internal Revenue Service: 

Department Of The Treasury: 
Deputy Commissioner: 
Internal Revenue Service: 
Washington, D.C. 20224: 
	
September 8, 2010: 

Mr. James R. White: 
Director, Tax Issues: 
Strategic Issues Team: 
U.S. Government Accountability Office: 
441 G Street, N.W. 
Washington, DC 20548: 

Dear Mr. White: 

Thank you for providing your draft report, Tax Gap: IRS Can Further 
Develop Its Efforts to Address Tax Evasion by Networks of Businesses 
and Other Related Entities (GAO-10-968), for our review and comments. 
We appreciate the time you and your team spent reviewing IRS programs 
and tools we use to identify business network tax evasion and the 
criteria that GAO developed for assessing network analysis processes. 

The draft report does a good job presenting the challenges for tax 
administration created by business networks and, in general, 
accurately represents the current state of some of the steps taken by 
the Internal Revenue Service to address these challenges. As you state 
in your report, we need to walk a "middle course between doing too much
too soon versus doing too little too slowly." We agree with the 
direction the report recommends: more focused strategic attention to 
network compliance, continued improvement and integration of tools 
used to analyze networks, and additional collaboration between IRS 
personnel with differing roles in the compliance process. We offer the 
following additional comments: 

* Because of the broad impact of networks and the complex 
organizational responses required, developing a separate strategy to 
address network compliance will be useful, but ultimately may not be 
as effective as ensuring that network issues are consciously and 
appropriately included in broader strategic plans. 

* The organizational changes we are undertaking (e.g., establishing 
the Global High Wealth Industry, piloting the Compliance Management 
Operations (CMO) approach, centralizing International compliance 
resources, and developing a Servicewide High Income Strategy) will 
help in this area, and have a significant impact on our capabilities 
to address network compliance. 

* In the end, the IRS will always be challenged to find technological, 
administrative, or auditing approaches to address the tax problems 
associated with the ever-increasing complexity and variability of both 
legitimate and abusive entity structures that use tiered flow through 
tax reporting. We are in the process of studying potential legislative 
and guidance changes to reduce the tax risks inherent in network 
structures. 

The IRS response addressing your recommendations is enclosed. In 
addition, we are providing technical comments (Enclosure 2) on several 
specific statements in the draft report that we think may need 
clarification. However, these statements do not affect the overall 
conclusions of the report. 

If you have any questions, please contact me, or a member of your 
staff may contact Don McPartland, Director, Research and Workload 
Identification, at (510) 637-2191. 

Sincerely, 

Signed by: 

Steven T. Miller: 

[End of letter] 

Enclosure 1: 

GAO recommends that the Commissioner of the Internal Revenue Service 
take the following actions: 

Recommendation 1: 

Establish an IRS-wide strategy with goals, which may need to be 
developed incrementally, to coordinate and plan ongoing and future 
efforts to identify and pursue network tax evasion. The strategy 
should include: 

1. Assessing the effectiveness of network analysis tools, such as yK-1. 

2. Determining the feasibility and benefits of increasing access to 
existing IRS data, such as scanning additional data from Schedule K-1, 
or collecting additional data for use in its network analysis efforts. 

3. Putting the development of analytical techniques and tools that 
focus on networks as the unit of analysis, such as GraphQuery, on a 
specific time schedule. 

4. Deciding how network efforts will be managed across IRS, such as 
whether a core program team or management group is needed. 

Response: 

Because of the broad impact of networks and the complex organizational 
responses required, developing a separate strategy to address network 
compliance will be useful, but may not be as effective as ensuring 
that network issues are consciously and appropriately included in 
broader strategic plans. In any event, a better articulated
strategy for deploying, maintaining, and improving tools for network 
analysis is needed. 

Item 1: 

The IRS agrees that it is useful to assess the effectiveness of 
analysis tools, but it would be necessary to balance the costs of such 
an assessment. 

Item 2: 

The IRS agrees with this recommendation. 

Item 3: 

The IRS agrees that it would be useful to better structure and support 
the development of analytical tools. 

Item 4: 

The IRS will look at this issue. It may not be possible nor 
appropriate to manage network compliance activity centrally. However, 
at a minimum, we will consider how to manage better the analytic tools 
and whether a core program team would be useful in this regard. 

Recommendation 2: 

Ensure that staff members who will be using current and additional 
network tools fully understand the tools' capabilities. 

Response: 

The IRS agrees that training of compliance employees in the use of 
analytic tools can be improved. 

Recommendation 3: 

Establish formal mechanisms for front-line users to interact directly 
with tool programmers and program analysts to ensure future network 
analysis tools, such as GraphQuery, are easy to use and help achieve 
goals. 

Response: 

Currently, the system developers have access to appropriate field 
employees, but we will consider improving the ability of field 
employees to have direct input and feedback to systems and risk 
assessment activities. 

[End of section] 

Appendix V: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

James R. White, 202-512-9110 or whitej@gao.gov: 

Acknowledgments: 

In addition to the contact named above, Tom Short, Assistant Director; 
Lydia Araya; Katie Arredondo; Russ Burnett; David Dornisch; Robert 
Gebhart; Eric Gorman; Sherrice L. Kerns; Melissa L. King; Adam Miles; 
Danielle Novak; Melanie Papasian; Ernest L. Powell Jr.; and A.J. 
Stephens made key contributions to this report. 

[End of section] 

Footnotes: 

[1] Stobie Creek Investments LLC v. U.S., 608 F.3d 1366 (9th Cir. 
2010). 

[2] GAO, Highlights of the Joint Forum on Tax Compliance, [hyperlink, 
http://www.gao.gov/products/GAO-08-703SP] (Washington, D.C.: June 
2008). 

[3] Under the Department of the Treasury's check-the-box rules, 
eligible business entities may choose their entity type for federal 
income tax purposes. Business entities with two or more members can be 
a corporation or a partnership. Business entities with only one member 
can be a corporation or a single-owner organization. Certain 
corporations are not eligible to choose their entity type and are 
always taxed as corporations. 26 C.F.R. §§ 301.7701-1 to -4. 

[4] Schedule K-1 is an information return that certain entities file 
to report income, credits, deductions, and other items that are 
distributed to other parties. Different Schedules K-1 exist for 
different entities, such as partnerships, S corporations, and trusts. 

[5] IRS's definition of tiering is more restrictive than our term, 
network. For purposes of this report, tiered networks should be 
considered a subset of networks. 

[6] Basis is generally the amount invested in a property for tax 
purposes. Basis is used to figure depreciation, amortization, 
depletion, casualty losses, and any gain or loss on the sale, 
exchange, or other disposition of the property. 

[7] See the redemption bogus optional basis transaction at [hyperlink, 
http://www.irs.gov/businesses/corporations/article/0,,id=154246,00.html]
, accessed Sept. 23, 2010. 

[8] To view the video, follow the link [hyperlink, 
http://www.gao.gov/products/GAO-10-968]. 

[9] The net estimate takes into account taxes that IRS expects to 
recover. The gross estimated tax gap was $345 billion for tax year 
2001. 

[10] As part of an organizational shift, LMSB will be known as the 
Large Business and International division beginning October 1, 2010. 

[11] OTSA collects Form 8886, Reportable Transaction Disclosure 
Statements, which is filed by investors who participate in listed 
transactions or other reportable transactions. 

[12] A tax shelter generally refers to a strategy or promotion that 
"shelters" income from taxation. Depending on the facts and legal 
analysis, a specific transaction/promotion may represent either lawful 
tax avoidance or unlawful tax evasion. Tax shelters resulting in 
evasion are said to be “abusive” tax shelters. The Internal Revenue 
Code defines tax shelter in specific circumstances such as when 
applying certain penalties or for certain tax accounting rules. 26 
U.S.C. §§ 461(i)(3), 6662(d)(2)(C). 

[13] 26 C.F.R. § 1.6011-4(b)(2). For IRS’s list of transactions, see 
[hyperlink, 
http://www.irs.gov/businesses/corporations/article/0,,id=120633,00.html]
, accessed Sept. 23, 2010. 

[14] Additionally, under the Tax Equity and Fiscal Responsibility Act 
of 1982 (TEFRA), the tax treatment of a partnership, including any 
adjustment or penalty, is generally determined at the partnership 
level. Pub. L. No. 97-248, §§ 401-406, 96 Stat. 324, 648-671 (Sept. 3, 
1982), codified at 26 U.S.C. § 6221. 

[15] 26 U.S.C. § 7213A. 

[16] An EIN is a taxpayer identification number that IRS uses to 
identify a business entity. 

[17] A TIN is a number that IRS uses in the administration of tax 
laws. It is either a Social Security number issued by the Social 
Security Administration or a taxpayer identification number issued by 
IRS, such as an EIN. 

[18] While we determined that it would not be appropriate to set 
criteria for the exact methodology the agency should use in their 
network analysis program, we performed a review of studies using 
approaches that may be of use to those attempting to develop such 
programs; see appendix III. 

[19] An overview of the information we gathered on methodological 
approaches to network analysis can be found in appendix III. 

[20] This definition applies to the “degree” centrality measure, which 
is the most basic measure of centrality. There are a number of other 
measures, such as closeness and betweenness centrality, which take 
into account indirect as well as direct contacts of network members. 

[21] Network constraint is a closely related measure that can be 
thought of as the opposite of brokerage as it gauges the extent to 
which a network member’s connections are to others who are also 
connected to one another. 

[22] Hierarchy can also be understood as reflecting the degree of 
reciprocity in the network, or the extent to which there is mutual 
exchange between actors. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation and 
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 GAO's Web site [hyperlink, http://www.gao.gov]. Each 
weekday, GAO posts newly released reports, testimony, and 
correspondence on its Web site. To have GAO e-mail you a list of newly 
posted products every afternoon, go to [hyperlink, http://www.gao.gov] 
and select "E-mail Updates." 

Order by Phone: 

The price of each GAO publication reflects GAO’s actual cost of
production and distribution and depends on the number of pages in the
publication and whether the publication is printed in color or black and
white. Pricing and ordering information is posted on GAO’s Web site, 
[hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card,
MasterCard, Visa, check, or money order. Call for additional 
information. 

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: 

Congressional Relations: 

Ralph Dawn, Managing Director, dawnr@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, D.C. 20548: 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, D.C. 20548: