Cheating Ourselves: The Economics of
Tax Evasion

Joel Slemrod

No government can announce a tax system and then rely on taxpayers’sense of duty to remit what is owed. Some dutiful people will undoubtedlypay what they owe, but many others will not. Over time the ranks of the
dutiful will shrink, as they see how they are being taken advantage of by the others.
Thus, paying taxes must be made a legal responsibility of citizens, with penalties
attendant on noncompliance. But even in the face of those penalties, substantial
tax evasion exists—and always has. The history of taxation is replete with episodes
of evasion, often notable for their inventiveness. During the third century, many
wealthy Romans buried their jewelry or stocks of gold coin to evade the luxury tax,
and homeowners in eighteenth-century England temporarily bricked up their
fireplaces to escape notice of the hearth tax collector (Webber and Wildavsky, 1986,
p. 141).

This essay reviews what is known about the magnitude, nature, and determi-
nants of tax evasion, with an emphasis on the U.S. income tax. Alm (1999),
Andreoni, Erard, and Feinstein (1998), and Slemrod and Yitzhaki (2002) offer
more comprehensive recent reviews of the literature. It then places this informa-
tion into a conceptual and policy context.

Tax Evasion in the United States

Determining the extent of evasion is not straightforward for obvious reasons.
(Would you answer survey questions about tax evasion honestly?) Because tax

y Joel Slemrod is the Paul W. McCracken Collegiate Professor of Business Economics and
Public Policy, Professor of Economics, and Director of the Office of Tax Policy Research at the
University of Michigan, Ann Arbor, Michigan. His e-mail address is �jslemrod@umich.edu�.

Journal of Economic Perspectives—Volume 21, Number 1—Winter 2007—Pages 25– 48



evasion is both personally sensitive and potentially incriminating, self-reports are
vulnerable to substantial underreporting (Baumeister, 1982). Moreover, the divid-
ing line between illegal tax evasion and legal tax avoidance is blurry. Under U.S.
law, tax evasion refers to a case in which a person, through commission of fraud,
unlawfully pays less tax than the law mandates. Tax evasion is a criminal offense
under federal and state statutes, subjecting a person convicted to a prison sentence,
a fine, or both. An overt act is necessary to give rise to the crime of income tax
evasion; therefore, the government must show willfulness and an affirmative act
intended to mislead. Some tax understatement is, however, inadvertent error, due
to ignorance of or confusion about the tax law (as is some overpayment of taxes).
Although the theoretical models of this issue generally refer to willful understate-
ment of tax liability, empirical analyses cannot precisely identify the taxpayers’
intent and therefore cannot precisely separate the willful from the inadvertent. Nor
can they, in complicated areas of the tax law, precisely distinguish the illegal from
the legal. Although this review is intended to address willful tax noncompliance,
the difficulty of identifying this behavior is reflected in the varying terms to which
the analyses refer, such as “evasion,” “noncompliance,” “misreporting,” and “tax
gap.” In what follows, when discussing empirical estimates I generally use the term
that generated the estimates employed, and use the term “evasion” in discussing
theoretical treatments of willful noncompliance.

Tax Evasion by Sources of Income
The most careful and comprehensive estimates of the extent and nature of tax

noncompliance anywhere in the world have been made for the federal taxes that
the U.S. Internal Revenue Service (IRS) oversees. The IRS has, beginning in 1979,
periodically estimated what it calls the “tax gap,” meaning how much tax should be
paid, but is not paid voluntarily in a timely way (U.S. Department of the Treasury,
Internal Revenue Service, 2005b, 2005d, 2005e, 2006). These studies provide
separate estimates of the failure to pay the proper amount of tax due to nonfiling;
underreporting of tax due on tax returns; and nonpayment or late payment of taxes
owed. The IRS comes up with its estimates by combining information from a
program of random intensive audits, originally known as the Taxpayer Compliance
Measurement Program (TCMP), with information obtained from ongoing enforce-
ment activities and special studies about particular sources of income—for in-
stance, the tips and cash earnings of informal suppliers like nannies and house-
painters—that can be difficult to uncover even in an intensive audit.

The Taxpayer Compliance Measurement Program data consist of line-by-line
information about what the taxpayer reported and what the examiner concluded
was correct. The primary purpose of the TCMP was to improve the process for
selecting returns for operational audits, rather than to estimate the tax gap. The
program of random audits began in 1963 and lasted until 1988; it was cancelled in
1995 in part to save money, but more importantly because of complaints in
Congress that it was unfair to subject one group of taxpayers to detailed audits as
a research tool to improve operations, and because of objections within the IRS to

26 Journal of Economic Perspectives



diverting resources from more productive operational audits to audits of taxpayers
who are less likely to be noncompliant. Until recently, IRS estimates of the tax gap
have been fairly simple extrapolations based on the 1988 data and the assumption
that rates of noncompliance for each tax source have not changed since 1988.

A modified version of the Taxpayer Compliance Measurement Program, called
the National Research Program (NRP), was recently implemented to examine
individual income tax returns from the 2001 tax year. For the NRP, the IRS
randomly selected about 46,000 returns for review and examination and, as with
previous TCMP studies, it oversampled high-income returns as well as individual
taxpayers who reported (Schedule C) sole proprietorship income. Specifically, of
the 46,000 in the overall sample, 21,000 had filed a Schedule C reporting sole
proprietorship income, comprising 46 percent of the sample, even though only
about 6 percent of individual taxpayers actually file a Schedule C. All of these
returns were given a manual review by experienced auditors, who classified the
returns into three categories of action: 1) accept the return as filed based on
completely corroborating information reported by third parties; 2) contact the
taxpayer by correspondence to investigate up to three items on the return that
could not be corroborated; or 3) conduct an in-person audit of the return. For
those returns selected for in-person audits, the classifiers also determined which
items were to be reviewed, with auditors granted discretion to follow up on
additional items that might arouse suspicion. In contrast to the NRP, the TCMP
performed line-by-line audits on all returns in the sample. To correct for the errors
potentially introduced by variability in auditor judgment, the tax gap estimates
based on the NRP data employed a modified version of the statistical correction
procedure developed by Feinstein (1991). Because the line-by-line audits can fail to
uncover substantial amounts of noncompliance, the tax gap estimates based on
both the TCMP and NRP studies make significant adjustments for undetected
noncompliance that rely on special studies of particular sources of income and
deductions.

Analyses of the National Research Program data form the basis for the first
substantial updates since 1988 of the individual income tax underreporting gap.
The IRS released preliminary estimates in March 2005 and updated these estimates
in February 2006 (U.S. Department of the Treasury, Internal Revenue Service,
2005c, 2006). Table 1 presents the tax gap figures for 2001 based on the NRP study
for the individual income tax filers and on estimates extrapolated from earlier
studies for other taxes. The overall gross tax gap estimate is $345 billion, which
amounts to 16.3 percent of estimated actual (paid plus unpaid) tax liability. This
percentage is not much different than earlier estimates based on extrapolations
from the 1988 tax gap studies (for example, U.S. Department of the Treasury,
Internal Revenue Service, 1996). However, taking into account changes in meth-
odology and the uncertainty of the estimating procedures, one cannot conclude
that the noncompliance rate has remained steady, as opposed to trending up or
down. Of the $345 billion estimate, the IRS expects to recover $55 billion, resulting

Joel Slemrod 27



Table 1
Components of the 2001 U.S. Federal Tax Underreporting Gap

Tax gap
($billion)

Percentage of amount
that should have

been reported/filed/
paid/collected

Gross tax gap 345 16.3
Underreporting 285 13*

Individual Income Tax 197 18

Underreported nonbusiness income 56 4
Wages and salaries 10 1
Net capital gains 11 12
Taxable pension annuities, IRA distributions 4 4
Taxable interest and dividends 3 4
Other 28 n.a.

Underreported business income 109 43

Nonfarm proprietor income 68 57
Partnership, S corporation, estate, and net trust income 22 18
Rent and royalty net income 13 51
Farm net income 6 72

Overreported Offsets to Income 15 4

Deductions 14 5
Exemptions 4 5
Other adjustments �3 �21

Overreported Credits 17 26

Employment Tax 54 7

Self-employment tax 39 52*
FICA and unemployment taxes 15 2*

Corporation Income Tax 30 17*

Large (�$10 million assets) corporations 25 14*
Small (�$10 million assets) corporations 5 29*

Estate and Excise Taxes 4 4*

Nonfiling 27 1*
Individual income tax 25 2*
Other 2 2*

Underpayment 34 2*
Individual income tax 23 2*
Corporation income tax 2 1*
Other 9 1*

Enforced and other late payments 55 3*
Net tax gap (tax not collected) 290 13.7*

Source: U.S. Department of the Treasury, Internal Revenue Service (2006).
Note: Only the figures for the individual income tax and the self-employment tax are based on the IRS’
National Research Program results; the rest are IRS extrapolations from earlier TCMP studies.
* Calculated by the author.

28 Journal of Economic Perspectives



in a “net tax gap” (the amount of tax not collected) for tax year 2001 of $290
billion, which is 13.7 percent of the tax that should have been paid.

Table 1 breaks down the aggregate estimate of the 2001 tax gap into its
components. About two-thirds of all underreporting of income happens on the
individual income tax. For the individual income tax, understated income—as
opposed to overstating of exemptions, deductions, adjustments, and credits—
accounts for over 80 percent of individual underreporting of tax. Underreported
business income is nearly twice as large as underreported nonbusiness income.
Taxpayers who were required to file an individual tax return, but did not file in a
timely manner, accounted for slightly less than 10 percent of the gap. While the
individual income tax comprises about two-thirds of the estimated underreporting,
the corporation income tax makes up slightly more than 10 percent, and the
employment-tax tax gap makes up about one-fifth of total underreporting.

The most striking and important aspect of Table 1 is the huge variation in the
rate of misreporting as a percentage of actual income by type of income (or offset).
Only 1 percent of wages and salaries are underreported, and 4 percent of taxable
interest and dividends are misreported. (These percentages exclude underreport-
ing associated with nonfiling.) Of course, wages and salaries, interest, and dividends
must all be reported to the IRS by those who pay them; in addition, wages and
salaries are subject to employer withholding. Self-employment business income is
not subject to information reports, and its estimated noncompliance rate is sharply
higher. An estimated 57 percent of nonfarm proprietor income is not reported, a
total of $68 billion, which by itself accounts for more than a third of the total
estimated underreporting for the individual income tax. As with prior estimates of
the individual income tax underreporting gap, over half is attributable to the
underreporting of business income, of which nonfarm proprietor income is the
largest component.

The particularly high noncompliance rate associated with self-employed in-
come has been corroborated through an indirect approach pioneered by Pissarides
and Weber (1989) in the United Kingdom. They show that, conditional on house-
hold characteristics and recorded incomes, the self-employed spend a higher
proportion of their reported income on food, and they argue that this reflects an
underreporting of income, not a higher propensity to consume food. After adjust-
ment for the differing variances of self-employment and employee incomes, Pissar-
ides and Weber estimate that self-employed people in the United Kingdom on
average underreported their income by about one-third. Feldman and Slemrod
(forthcoming) applied this methodology to U.S. individual tax return data by
examining whether the relationship between charitable contributions (rather than
food expenditure) and income depends on the source of income. They find that,
other things equal, reported positive self-employment income of $1.54 is associated
with the same level of contributions as $1.00 of wage and salary income, which
implies—assuming a negligible wage and salary noncompliance rate and that the
self-employed are not inherently more charitable than others—a self-employment
noncompliance rate of 35 percent; for positive farm net income, the implied

Cheating Ourselves: The Economics of Tax Evasion 29



noncompliance rate is 74 percent. Strikingly, negative reported values for self-
employment income are also associated with more contributions than reported by
taxpayers with no self-employment income, suggesting that on average these re-
ported losses are associated with higher true incomes.

All in all, there is substantial evidence that the extent of evasion for sole
proprietor income is high compared to such income sources as wages, salaries,
interest, and dividends, and may be more than half of true income. Other compo-
nents of taxable income for which information reports are nonexistent or of
limited value, such as other nonwage income and tax credits, also have relatively
high estimated misreporting rates. The IRS reports that the net misreporting rate
is 53.9, 8.5, and 4.5 percent for income types subject to “little or no,” “some,” and
“substantial” information reporting, respectively, and is just 1.2 percent for those
amounts subject to both withholding and substantial information reporting (U.S.
Department of the Treasury, Internal Revenue Service, 2006).

Who Evades?
One intriguing question is how the level of noncompliance, and its proportion

to income, varies by income class. Somewhat surprisingly, little is known about this
from the IRS tax gap studies. Christian (1994) did report, based on the 1988 TCMP
study, that higher-income people evade less than those with lower incomes, relative
to the size of their true income; indeed, according to this study those with adjusted
gross income above $500,000 on average reported 97.1 percent of their true
incomes to the IRS, compared to just 78.7 percent for those with adjusted gross
income between $5,000 and $10,000. This pattern appears consistent with the old
saying among tax professionals that “the poor evade and the rich avoid,” meaning
that the rich tend to reduce their taxes through legal “avoidance” measures such as
tax shelters, while those with lower incomes attempt more outright evasion. How-
ever, the Christian study is not conclusive. These figures do not adjust for the
noncompliance the TCMP auditors did not detect. TCMP audits of personal tax
returns do not generally investigate corporate or partnership tax returns, so any
evasion at those business levels is generally not accounted for; because high-income
individuals have proportionately more business income, the relatively high rates
shown in this table may overestimate the voluntary compliance of this group. The
TCMP audits may not detect sophisticated tax shelters pursued mainly by high-
income taxpayers, some of which are legal avoidance but others of which are
probably illegal. Finally, many of those categorized as low-income in this study may
have reported business losses, so that they are not people with low permanent
income and may even, by dint of noncompliance, have placed themselves in the
(reported) low-income category.

Tax noncompliance seems related to some other observable characteristics of
taxpayers. According to Andreoni, Erard, and Feinstein (1998, pp. 821–22), mar-
ried filers and taxpayers younger than 65 have significantly higher average levels of
noncompliance than others, and econometric studies by Clotfelter (1983) and
Feinstein (1991) that control for income and marginal tax rates come to similar

30 Journal of Economic Perspectives



conclusions. Baldry (1987) found evidence, in an experimental setting, that men
evade more than women.

There also seems to be substantial heterogeneity in tax evasion. The TCMP
studies concluded that, within any group defined by income, age, or other demo-
graphic category, there are some who evade, some who do not, and even some who
overstate tax liability. For example, for taxpayers with reported income between
$50,000 and $100,000 in 1988, 60 percent understated tax, 26 percent reported
correctly, and 14 percent overstated tax (Christian, 1994, p. 39). These studies do
not explore to what extent this heterogeneity is explained by different “tastes” for
evasion as opposed to different opportunities to evade. I return to this issue later.

Big Business and Tax Evasion
Businesses play a central role in the tax system. For taxes that by statute fall on

employees (and independent contractors), businesses “withhold” tax; remit the tax
on behalf of the employees; and provide the IRS with information reports of these
activities that can be matched against the individuals’ reports to the IRS. In
addition, a prominent tax is levied on corporate business entities—the corporation
income tax. As Table 1 shows, the IRS estimates noncompliance with the corpora-
tion income tax in 2001 to be $30 billion, which corresponds to a noncompliance
rate of 17 percent. Of this $30 billion, noncompliance by corporations with over
$10 million in assets make up $25 billion. But the estimated noncompliance rate of
the larger companies is lower: 14 percent, compared to 29 percent for corporations
with less than $10 million of assets.

This estimate has not been updated by the National Research Program.
Instead, the estimates for the corporation income tax gap come from three sources,
according to the U.S. General Accounting Office (1988): For small corporations
the IRS used Taxpayer Compliance Measurement Program data, adjusted for
underreporting unlikely to be detected by that method. For medium-sized corpo-
rations, the gap was calculated by estimating, based on operational (that is, non-
TCMP) audits from the mid-1980s, how much tax revenue would have been
generated if the IRS examined all these corporations’ tax returns. Finally, for large
corporations, because the IRS routinely examines a high percentage of these
companies, examination results were used as the basis of estimates of the tax gap.

The Bureau of Economic Analysis also calculates an annual measure of cor-
porate tax misreporting. In 2000, its 13.8 percent estimate of the ratio of misre-
porting to actual liability was close to, but slightly lower than, the extrapolated IRS
estimate. Petrick (2002, p. 7) discusses the Bureau of Economic Analysis method-
ology and Slemrod (2004, fn. 5) compares it to the IRS tax gap studies.

Because these estimates are largely based on operational audits, and because
most big corporations are routinely audited, some caveats apply to the tax gap
estimates they generate. The deficiencies proposed by the examination team are
not a perfect measure of actual noncompliance. Due to the complexity of the tax
law, exactly what is actual tax liability—and therefore what is actual tax noncom-
pliance—is often not clear. Second, any given examination is not perfect. Some

Joel Slemrod 31



noncompliance may be missed, and there will also be mistakes in characterizing as
noncompliance what is legitimate tax planning. For this reason, the data reflect not
only the reporting behavior of the companies but also the enforcement behavior of
the tax authority. Knowing that the resolution of the ultimate tax liability is often
a long process of negotiation that may or may not involve the judicial system, the
tax liability per the originally filed return, as well as the initial deficiency assessed
by the examination team, may be partly a tactical “opening bid” that is neither
party’s best estimate of the “true” tax liability. Indeed, IRS Commissioner Mark
Everson (2005) testified to the President’s Advisory Panel on Tax Reform that the
IRS had “a reputation for trading [penalties] away,” so that it was “always in the interest
of the noncompliant taxpayer to take an aggressive position with the Service.”

The proposed deficiency also does not necessarily capture the long-term effect
of tax noncompliance on the present value of revenues collected, both because
some of the proposed tax deficiency may involve temporary adjustments in tax
liability—such as immediate expensing versus depreciation—that will have offset-
ting consequences in future years, and because upward adjustments made to the
taxable income of corporations with negative taxable income would not increase
that year’s tax liability, but would in general increase the present value of tax
liability to the extent that it reduces the expected amount of loss carryforwards.

Based on an examination of previously undisclosed IRS operational audits and
appeals data merged with confidential tax return data for corporations, Hanlon,
Mills, and Slemrod (forthcoming) calculated that tax noncompliance of large
corporations, as measured by tax deficiencies proposed by IRS auditors, amounted
in the period 1983 to 1998 to approximately 13 percent of “true” tax liability, slightly
lower than the IRS and Bureau of Economic Analysis estimates. All in all, 60 percent of
the proposed deficiency was either agreed to by the taxpayer or upheld at a later stage.
This 60 percent sustention rate is almost certainly an upper bound estimate of the rate
for all companies, however, because it excludes the (generally more contested) tax
return filings that had not been settled when the data set was compiled.

They also found that the largest companies (those with assets greater than $5
billion) had the greatest percentage of firms with a tax deficiency (74 percent) and
the highest proposed deficiency rate (14.6 percent, versus a range of 9.9 percent to
13.4 percent for the other six groups). This finding is consistent with the larger
firms with more complex operations having more opportunities for tax noncom-
pliance. There was also some evidence suggesting that the noncompliance rate for
corporations relative to their size is “U-shaped,” with medium-sized businesses
among the set of large companies having the lowest rate of noncompliance.1

1 In a study of much smaller companies, Rice (1992) did not find a similar association between firm size and
tax compliance, although he concluded that managers of corporations whose profit performance is below its
industry norm may utilize tax noncompliance as a strategy to cut costs. In contrast, high-profit companies
may take advantage of their greater ability to underreport income without being audited.

32 Journal of Economic Perspectives



On average, Hanlon, Mills, and Slemrod (forthcoming) found that private
companies have higher proposed deficiency rates than public companies (17.1
percent versus 12.5 percent), which is similar to the survey results analyzed in Cloyd
(1995) and Cloyd, Pratt, and Stock (1996). Privately held firms may be more
aggressive in angling for lower taxes because they have fewer capital market
pressures and thus can sacrifice reporting high financial accounting earnings in an
attempt to reduce taxes owed. We also found a positive relationship between the
amount of intangible assets a firm holds (as proxied by research and development
expenses and market-to-book ratio) and its tax deficiency rate, which is consistent
with the idea that these firms have greater tax planning opportunities. Finally, the
percentage of annual executive compensation in the form of bonuses and the level
of equity incentives from exercisable stock options are positively related to the
proposed tax deficiency.

How U.S. Tax Evasion Compares to Other High-Income Countries
How does the magnitude and nature of tax evasion in the United States stack

up against other high-income countries? The answer to that question is elusive. No
other country has undertaken a broad-based analysis of tax evasion like the Tax-
payer Compliance Measurement Program or the National Research Program.
Australia, Canada, Sweden, and the United Kingdom have in place small-scale
random audit programs for selected taxes and taxpayer groups, but detailed results
of these programs are generally not published (although they presumably inform
their tax agencies’ risk management assessments and resource allocation deci-
sions). The Swedish Tax Agency (2004) estimated the total gap as a percentage of
taxes to be 9 percent in 1997 and 8 percent in 2000. Although no official U.K.
estimate has been released, an official document does speculate that “it is likely that
the United Kingdom has a tax gap of a similar magnitude” to that of Sweden and
the United States (O’Donnell, 2004).

It is also instructive to compare estimates of U.S. income tax evasion with
noncompliance rates estimated for the value-added tax, a consumption tax that,
while a staple of foreign revenue systems, is not in the U.S. tax arsenal. According
to a confidential study made in 2005 by the Forum on Tax Administration, a
subsidiary body of the OECD’s Committee on Fiscal Affairs, only a few countries
were prepared to make public their estimates of overall noncompliance, which
ranged from 4.0 to 17.5 percent.2 These estimates are generally based on “top-
down” exercises, which compare actual revenues from the value-added tax to a
theoretical tax base and tax amount derived by examining consumption expendi-
ture adjusted to account for factors that affect the base for the value-added tax

2 The 4.0 to 17.5 percent range was cited in the report of the President’s Advisory Panel on Federal Tax
Reform (2005, p. 202), but the citation suggests that the numbers present an OECD-wide picture, when
they in fact represent a small number of countries; nor was the OECD responsible for any of the research
that led to these figures. I am grateful to Richard Highfield of the OECD for providing this information.

Cheating Ourselves: The Economics of Tax Evasion 33



(for example, tax policy choices concerning goods that are exempt from the
value-added tax, or which pay different rates). The United Kingdom is the most
transparent with this approach. Its most recent report suggests a gap for its
value-added tax in 2003–2004 and 2004 –2005 of 13.5 percent, down from an
average of 15.7 percent in 2000 –2003 (HM Revenues and Customs, 2005). The
U.K. reports claim to have validated their top-down estimates with “bottom-up”
estimates, but the details of this procedure are not published. Earlier studies across
Europe found a wide range of estimated noncompliance rates for the value-added
tax (Agha and Haughton, 1996): 2 to 4 percent for revenue foregone for the
United Kingdom in 1986 (Tait, 1988); 40 percent of revenue uncollected in Italy
(Pedone, 1981); 6 percent for the Netherlands in 1976; and 8 percent for Belgium
in 1980 (Oldman and Woods, 1983). Silvani and Brondolo (1993) report calcula-
tions of the net evasion rate for the value-added tax in 19 mostly developing
countries and report a median evasion rate of 31.5 percent, with New Zealand the
lowest at 5.1 percent and Peru the highest at 68.2 percent. In this study, the evasion
rates were “prepared by the national authorities following a standard methodology
supplied by the authors, which essentially compares VAT revenue collected to the
potential VAT base (times the average VAT tax rate)” (p. 1).

Finally, there are voluminous cross-country estimates of the extent of what is
referred to as the shadow, irregular, underground, informal, or black economy.
Although the definition of the shadow economy is often not precise, it typically
entails the production and distribution of services that are themselves not illegal,
but become unlawful either by tax evasion or the bypassing of regulations. The
underground economy is generally measured as a fraction of GDP, not tax liability,
and would include business activity that is not subject to regulations, but which does
not generate taxable income. It also excludes some forms of tax evasion such as the
overstatement of deductions and credits, and the use of sophisticated tax shelters.
In spite of these definitional differences, the underreporting of business income,
particularly proprietor income that is the focus of estimates of the underground
economy, comprises a significant fraction of the estimated tax gap even in the
United States.

The methodology for estimating the underground economy generally relies
on inferring the level or trends in the underground economy from data on
measurable quantities, such as currency demand, electricity consumption, or na-
tional income accounts. For example, Feige (1989) estimates the size of the
underground economy by assuming that most unreported economic activity takes
place in cash, and that there is a “base year” when the underground economy did
not exist; under these assumptions, increases in cash holdings over the base year
indicate a growing underground economy. Similarly, Tanzi (1980, 1983) pioneered
a methodology based on regressions explaining the ratio of currency to M2, and he
interpreted the portion of this ratio explained by changes in the tax level as an
indication of changes in the size of the underground economy. It is very difficult to
verify the accuracy of these estimates. Even Tanzi (1999, p. F340) himself has
recently said that “as long as the estimates remain as divergent as they have been,

34 Journal of Economic Perspectives



they cannot provide much of a guidance for policy.” Breusch (2005) has recently
presented a compelling critique of one widely used methodology for generating
estimates of the underground economy. Such estimates may say more about the
relative extent of the shadow economy, across time and countries, than about its
absolute magnitude. According to the calculations of Mummert and Schneider
(2002, Table 1) based on the currency demand approach, in 2001–2002 the United
States had the smallest shadow economy (relative to official GDP) among 21 OECD
countries, at 8.7 percent, only just more than half the average of 16.7 percent.
Switzerland (9.4 percent) and Austria (10.6 percent) were the next lowest, while
Italy (27.0 percent) and Greece (28.5 percent) were the highest.

Another useful international perspective is the system of tax administration
and enforcement. According to OECD (2004), of the 24 OECD countries that
reported the data, in 2001 the United States had the second-lowest ratio of
administrative costs to net-of-refund revenue collections, at 0.52 percent, compared
to a 1.11 percent unweighted average, and the lowest ratio of full-time staff of the
national revenue body to the labor force. This does not necessarily signal an
inappropriately low attention to tax enforcement; it could instead reflect a more
efficient use of staff and computers, as well as economies of scale.

Like all but two OECD countries, the United States relies on withholding-at-source
arrangements for the collection of the bulk of personal tax revenue on wage and salary
income. Unlike most OECD countries, the United States does not use withholding-at-
source for the collection of personal income tax on dividends or interest. Neither does
it use withholding-at-source for independent personal services or royalties and rents,
though many OECD countries do. The United States does maintain the most substantial
program of information reporting of any OECD country. An extremely wide variety of
transactions must be reported to the IRS, including interest, dividends, real estate
transactions, rents, sales of securities, and wages. In 2002–2003, some 1.3 billion such
reports were received (96 percent electronically) and computer-matched with taxpayer
records; the program entailed some 4.3 million taxpayer contacts and resulted in
additional assessments amounting to almost $5 billion.

Finally, the United States is in the minority of OECD countries (nine out of 30)
in requiring that individuals file tax returns. Half of the OECD countries operate
a system in which most wage earners need not file due to exact, cumulative
employer withholding, and in other countries the tax authority sends taxpayers a
“pre-populated” tax return based on the information they receive from third
parties, requiring the taxpayer only to verify that the information is correct.

The Positive Theory of Tax Evasion: Models and Tests

The Deterrence Model of Tax Evasion
The standard framework for considering an individual’s choice of whether and

how much to evade taxes is a deterrence model first formulated by Allingham and
Sandmo (1972), who adapted Becker’s (1968) model of the economics of crime. In
this model, taxpayers decide whether and how much to evade taxes in the same way

Joel Slemrod 35



they would approach any risky decision or gamble— by maximizing expected
utility—and are influenced by possible legal penalties in just the same way they are
influenced by any other contingent cost. Optimal tax evasion depends on the
chance of getting caught and penalized, the size of the penalty for evasion, and the
individual’s degree of risk aversion. In an intriguing extension of the theory,
Yitzhaki (1974) pointed out that if the penalty (and any associated nonpecuniary
costs) for discovered evasion is proportional to the tax understated (rather than, as
Allingham and Sandmo assumed, the income understated), then the tax rate has
no effect on the terms of the tax evasion gamble. As the tax rate rises, the cost of
a detected understatement of taxes rises in exact proportion to the reward from a
successful understatement of taxes, so the reward-to-risk ratio is unchanged. In this
situation, a higher tax rate has only an income effect and, for example, if a
taxpayer’s level of (absolute) risk aversion increases as after-tax income falls, a
higher tax rate will decrease tax evasion.

Thirty-plus years of subsequent analysis has extended this model in a number
of dimensions, including allowing an endogenous probability of detection, analyz-
ing evasion jointly with the labor supply decision (thus directly addressing the
shadow economy), incorporating sources of uncertainty other than the chance of
audit, and addressing general equilibrium considerations. Andreoni, Erard, and
Feinstein (1998) offer a comprehensive survey of the theory, and Sandmo (2005)
provides a nice retrospective on tax evasion modeling.

A recent literature adapts the theory of tax evasion, which for the most part
concerns individual decision makers, to the tax compliance decisions made by
businesses. Arguably, we would expect large public companies to act in a risk-
neutral manner, rather than like the risk-averse individuals in the Allingham and
Sandmo (1972) model. Furthermore, in large, publicly held corporations, deci-
sions about tax compliance are not made by the shareholders directly but by their
agents, like the chief financial officer or a vice president responsible for tax matters.
To align the incentives of the decision makers and the shareholders, the corpora-
tion should tie the agent’s compensation to observable outcomes that affect after-
tax corporate profitability. Crocker and Slemrod (2005) show that this implies that,
if penalties for evasion were to apply to the agent, the principal can alter its
compensation contract with the agent, possibly offsetting the intended conse-
quences of the IRS policy (see also Chen and Chu, 2005). The conclusion that
enforcement strategies directed at the tax director and those directed at the
corporation itself may impact corporate behavior differently is especially pertinent
in light of the Sarbanes-Oxley rules, which create important changes in the respon-
sibility for misreporting: for instance, they require that the chief executive officer
sign the company’s federal income tax return. Slemrod (2004) expands on the
issues related to tax noncompliance by businesses.

Attempts to verify the predictions of the Allingham and Sandmo (1972) model
empirically have focused on how evasion is affected by enforcement intensity and
the level of tax rates. However, empirical tests have been plagued by the same
measurement issues that arise in assessing the magnitude of tax noncompliance.

36 Journal of Economic Perspectives



Clotfelter (1983) examines the micro data from the Taxpayer Compliance Mea-
surement Program studies and finds that noncompliance is strongly positively
related to the marginal tax rate; however, Feinstein (1991) finds a negative impact.
As with any cross-sectional study of the impact of taxes on behavior, their approach
is made difficult by the fact that the marginal tax rate is a (complicated, nonlinear)
function of income, making it difficult to identify the tax rate and income effects
separately without making strong functional form assumptions. (Having data from
two separate years in which the tax rate as a function of income differs, as Feinstein
did, mitigates this problem to some degree.) Beron, Tauchen, and Witte (1992)
examined TCMP data aggregated to the 3-digit zip code level and concluded that
increasing the odds of an audit significantly increased reported adjusted gross
income and tax liability for some income groups, but not all. However, the
district-level audit rate is not exogenous, perhaps reflecting something about the
compliance characteristics of the population. The authors use the level of IRS
resources relative to the number of returns as an instrument for the audit rate,
arguing that the IRS has not been able to allocate its resources so as to achieve its
goals, but this approach is invalid to the extent that the IRS succeeds in targeting
its resources toward areas believed to be particularly noncompliant.

Field experiments offer another source of evidence. Slemrod, Blumenthal, and
Christian (2001) analyze the results of a randomized controlled experiment con-
ducted by the State of Minnesota Department of Revenue. They found that low- and
middle-income taxpayers who received a letter promising an audit reported slightly
more, but statistically significantly more, income than those who did not receive
such a letter, and the difference was larger for those with greater opportunities to
evade. However, high-income taxpayers receiving an audit threat on average re-
ported lower income. The authors speculate that sophisticated, high-income tax-
payers view an audit as a negotiation, and view reported taxable income as the
opening (low) bid in a negotiation that does not necessarily result in the determi-
nation and penalization of all noncompliance.

Perhaps the most compelling empirical support for the Allingham and
Sandmo (1972) deterrence model is the cross-sectional variation in noncompliance
rates across types of income and deductions, as indicated in Table 1. Line item by
line item, there is a clear positive correlation between the rate of compliance and
the presence of enforcement mechanisms such as information reports and em-
ployer withholding. Klepper and Nagin (1989) showed compellingly that, across
line items, noncompliance rates are related to proxies for the traceability, deni-
ability, and ambiguity of items, which are in turn related to the probability that
evasion will be detected and punished. They also find evidence of a substitution-like
effect across line items, such that greater noncompliance on one item lowers the
attractiveness of noncompliance on others, because increasing the latter jeopar-
dizes the expected return to the former by increasing the probability of detection.
Another example of the link from a lack of deterrence to tax compliance involves
state use taxes, which are due on sales purchased from out-of-state vendors but
consumed in the state of residence. These taxes are largely unenforceable (except

Cheating Ourselves: The Economics of Tax Evasion 37



perhaps for some expensive items like cars), and noncompliance rates are in the
range of 90 percent (Bruce and Fox, 2000, pg. 1380).

However, there has been no compelling empirical evidence addressing how
noncompliance is affected by the penalty for detected evasion, as distinct from the
probability that a given act of noncompliance will be subject to punishment.

The demonstrated deterrent effect of detecting (and penalizing) noncompli-
ance is important because IRS enforcement activities have declined sharply in
recent years. Between 1996 and 2004, the share of nonbusiness individual returns
audited dropped from 1.67 percent to 0.74 percent. The share of corporate returns
subject to face-to-face audits dropped from 2.62 percent to 0.71 percent, and the
coverage ratio for corporations with assets over $250 million fell from 51.7 percent
in fiscal year 1997 to 29.7 percent in fiscal year 2003 (U.S. Department of the
Treasury, Internal Revenue Service, 2005a, Table 10). In part, the reduction in
face-to-face audits may reflect a more efficient use of computerized checks as a
substitute. However, the number of criminal tax cases recommended for prosecu-
tion by the IRS declined by 50 percent between 1992 and 2002 (IRS data cited in
Johnston, 2003). The use by the IRS of its three main weapons for collecting tax
debts from recalcitrant tax evaders—levies (garnishing wages or seizing money
from bank accounts), liens (taking ownership of the taxpayer’s property until a tax
debt is paid), and outright seizures of property—also declined sharply. Between
1996 and 2004, the number of levies fell from 3.1 million to 2.2 million, liens fell
from 750,000 to 534,000, and seizures fell from 10,449 to 440 (U.S. Department of
the Treasury, Internal Revenue Service, 2005a, Table 16).

Some of the decline in IRS enforcement is a temporary phenomenon—in
response to the IRS Restructuring and Reform Act of 1998, the IRS had to divert
resources temporarily towards its reorganization efforts—and many of the enforce-
ment indicators mentioned above have started to rebound a bit. In November 2005,
the IRS Commissioner announced plans to increase the number of audits it
conducts in 2006, focusing more of its resources on taxpayers with incomes over
$100,000 and self-employed workers who deal largely in cash (Herman and Silver-
man, 2005). But some of the decline is due to a longer-term trend. For instance,
between 1992 and 2002, the number of tax returns grew by 12 percent, but the
number of IRS tax auditors fell by a quarter from 16,000 to 12,000 (Johnston, 2003,
based on IRS data). Real operating costs actually fell between fiscal year 1993 and
2000, but have risen by slightly more than 7 percent from 2000 to 2005. During this
time, both the complexity of the tax law and the sophistication of abusive tax
shelters and other means of evading taxes increased notably. The decline in
enforcement poses the danger that tax evasion will increase.

Behavioral Models
Although the Allingham and Sandmo (1972) approach has dominated the

economics literature, some have argued that it misses important elements of the tax
evasion decision in such a way that the model predicts a compliance rate much
lower than what we actually observe. For example, Feld and Frey (2002, p. 5) assert

38 Journal of Economic Perspectives



that it is “impossible to account for tax compliance in terms of expected punish-
ment.” The dismissive argument goes as follows: given the average probability of
audit (less than 1 percent for individual returns with no business income), the
penalties typically assessed for noncompliance (typically 10 percent of the amount
underpaid), and what we know about the degree of risk aversion from other
contexts, noncompliance should be much, much higher than it apparently is.

But this dismissive argument is not persuasive, because the low average audit
coverage rate vastly understates the chances that the average dollar of unreported
net income would be detected. A wage or salary earner whose employer submits the
employee’s taxable income and Social Security number electronically to the Inter-
nal Revenue Service, but who does not report that income on his own personal
return, will be flagged for further scrutiny with a probability much closer to
100 percent than to 1 percent. Thus, the low rates of noncompliance for labor
income reported in Table 1 by no means patently contradict the deterrence theory.
Whether the 57 percent noncompliance rate of nonfarm sole proprietors is lower
than deterrence theory predicts is less clear; Andreoni, Erard, and Feinstein (1998,
pp. 821– 822) argue that it is lower.

Nonetheless, considerable experimental (and anecdotal) evidence suggests
that the story of tax evasion involves more than amoral cost– benefit calculation.
Frey (1997) argues for the importance of differentiating between intrinsic motiva-
tion, under which taxpayers comply with tax liabilities because of “civic virtue,” and
extrinsic motivation, in which they pay because of threat of punishment. He
suggests that increasing extrinsic motivation—say with more punitive enforcement
policies—may “crowd out” intrinsic motivation by making people feel that they pay
taxes because they have to, rather than because they want to. In an experimental
setting, Scholz and Lubell (2001) find that the level of cooperation in certain
settings declines significantly when penalties are introduced, suggesting that the
increased deterrence motivation did not compensate for the changes higher pen-
alties bring about in how people frame their decisions.

Some laboratory experiments have found that subjects respond not only to the
probabilities and stakes of a tax evasion game, but also to the context provided to
them (Spicer and Becker, 1980; Alm, Jackson, and McKee, 1992). In particular, tax
evasion decisions may depend on perceptions of the fairness of the tax system. If,
the argument goes, perceived tax equity strengthens the social norm against
evasion, then evasion becomes more costly in terms of bad conscience (if not
caught) or bad reputation (if caught). Falkinger (1995) elaborates on this argu-
ment, while Cowell (1990, p. 219) reports on experiments that fail to find links
between perceived inequities in the tax system and noncompliance. An individual
can also find unfairness in what the government uses tax revenues for—a person
with some of the spirit of Henry David Thoreau may avoid taxes because that
person thinks government (nontax) policy wrong (Andreoni, Erard, and Feinstein,
1998). Such individual judgments can be complex; for example, expenditures on
warfare might be tolerated in a patriotic period, but rejected during another period
characterized by antimilitarism (Daunton, 1998).

Joel Slemrod 39



These patterns suggest that a form of reciprocal altruism may be at work—a
form in which the taxpayer’s behavior depends on the behavior, motivations, and
intentions not of any subset of other individuals, but of the government itself.3 Levi
(1998, p. 91) argues that if citizens believe that the government will act in their
interests, that its procedures are fair, and that their trust of the state and others is
reciprocated, then people are more likely to become “contingent consenters” who
cooperate in paying taxes even when their short-term material interest would make
free-riding the individual’s best option.

Some survey evidence supports this view. Torgler (2003) and Slemrod (2003)
show a positive relationship across countries between survey-based attitudes toward
tax evasion on the one hand and professed trust in government, and Slemrod
(2003) finds that the same relationship holds across individuals within the United
States and Germany. A 2002 poll in the Czech Republic indicated that a person
would be more likely to evade taxes if that person believed government services
were substandard (Hanousek and Palda, 2004). Of course, such survey responses
may also reflect after-the-fact rationalization of noncompliant behavior.

If perceptions matter for tax compliance, a natural question is to what extent
tax compliance behavior can be manipulated by the government to lower the cost
of raising resources. Appeals to conscience go back at least to Hammurabi’s reign
in ancient Babylon, when the tax collector sent the following notice when payments
were late: “Why have you not sent to Babylon the 30 lambs as your tax? Are you not
ashamed of such behavior?” (as cited in Webber and Wildavsky, 1986, p. 58).
Appeals to patriotism to induce citizens to pay their taxes (and, often, buy war
bonds) are common in recent times; the U.S. Secretary of Treasury during World
War I, William Gibbs McAdoo, referred to these campaigns as “capitalizing patri-
otism.” Kang and Rockoff (2006) discuss the World War I experience, while Jones
(1996) discusses fiscal propaganda during World War II.

That such campaigns are successful during ordinary (nonwar) times in swaying
taxpayers from their otherwise optimal compliance strategy has not been compel-
lingly demonstrated. In a randomized field experiment with Minnesota taxpayers in
a peacetime setting, Blumenthal, Christian, and Slemrod (2001) find no evidence
that either of two written appeals to taxpayers’ consciences had a significant effect
on compliance. One letter stressed the beneficial effects of tax-funded projects,
while the other conveyed the message that most taxpayers were compliant. Torgler
(2004), using a controlled field experiment in Switzerland, also found that moral
suasion has hardly any effect on taxpayers’ compliance behavior.

Survey evidence also suggests that attitudes about the acceptability of tax

3 Some have proposed substituting the expected-utility-maximization framework with an alternative
framework, in the spirit of the prospect theory developed by Kahneman and Tversky (1979). For
example, Dhami and al-Nowaihi (2004) argue that, compared to an Allingham and Sandmo (1972)
model, such a framework (with the addition of a stigma cost for discovered evasion) can more
satisfactorily explain the level of observed evasion, the non-ubiquity of evasion, and the fact that tax rates
negatively affect evasion.

40 Journal of Economic Perspectives



evasion vary considerably across countries. In the World Values Surveys done
between 1999 and 2002, respondents were asked whether, given the chance, tax
evasion is never, sometimes, or always justified, where a value of 1 corresponds to
“never justifiable” and a value of 10 corresponds to “always justifiable.” The U.S.
average was 2.28, just slightly below the OECD average of 2.34. These attitude
measures of the World Values Survey across countries are associated, holding other
factors constant, with already-discussed measures of the shadow economy and
widely used survey measures of actual evasion (Torgler, 2004).

The difficulties of separating out whether people pay their taxes because they
feel they “ought to,” or whether they fear the penalties attendant to not doing so,
is well illustrated by some evidence from a recent survey sponsored by the IRS
Oversight Board (U.S. Department of the Treasury, Internal Revenue Service
Oversight Board, 2006). While 96 percent of those surveyed in 2005 mostly or
completely agreed that “It is every American’s civic duty to pay their fair share of
taxes,” 62 percent also said that “fear of an audit” had a great deal or somewhat of
an influence on whether they report and pay their taxes “honestly.”

The Normative Theory of Taxation and Its Policy Implications

Fairness and Incidence Concerns
The normative questions raised by tax evasion are often complex, involving

issues of fairness, efficiency, and how to measure social costs and benefits, so that
how to apply economic reasoning to the range of relevant policy instruments is still
being formulated. Tax evasion affects the distribution of the tax burden as well as
the resource cost of raising taxes— bread-and-butter concerns of public econom-
ics.4 If the estimated $290 billion net income tax gap could somehow be costlessly
eliminated, that money could be used to finance worthy government projects, or
used to finance an across-the-board cut in tax rates that would benefit compliant
taxpayers. But expanding government programs could be financed in a number of
other ways, such as raising tax rates or broadening the income tax base, and a tax
reduction could be financed by cuts in overall spending; the real question is
whether curbing evasion would improve the equity and efficiency of the public
finances. Moreover, curtailing tax evasion is in fact not costless, and these costs
must be considered in developing optimal policy.

Good tax policy should be designed with the realities of evasion in mind. If all
Americans were genetically predisposed to underpay their legal tax liability by
20 percent, at no cost to them, tax evasion wouldn’t matter at all. Government
would simply readjust everyone’s “sticker price” tax liability upward so that the
desired amount of tax would be collected, even after the 20 percent “discount” was
taken. Each taxpayer might think that he or she is beating the system, but in fact no

4 This section draws from the lengthier treatment in Slemrod and Bakija (2004).

Cheating Ourselves: The Economics of Tax Evasion 41



one gains in either relative or absolute terms compared to a world with no
evasion—we are, in effect, “cheating ourselves.” Similarly, if opportunities or pre-
dilections for evasion were systematically related to income, then the tax rate
schedule could just be adjusted to achieve whatever degree of progressivity is
deemed optimal. Of course, not everyone evades taxes by the same proportionate
amount or by an amount strictly related to income, both because of differences in
personal characteristics—like attitudes toward risk, the tax system, and honesty—
and because of different opportunities and potential rewards for evasion. As a
result, tax evasion can cause serious inequities and inefficiencies.

Evasion creates horizontal inequity because equally well-off people end up with
different tax burdens. Attempts to reduce tax evasion can create vertical equity
concerns, as when the IRS is criticized for spending resources to reduce fraud
related to the Earned Income Tax Credit, whose recipients are low-income house-
holds, instead of devoting those enforcement resources to tax shelters that are
pursued by high-income households.

Tax evasion also imposes efficiency costs. The most obvious are the resources
taxpayers expend to implement and camouflage noncompliance, and the resources
the tax authority expends to address it. In addition, when the tax system is
otherwise close to optimal it provides a socially inefficient incentive to engage in
those activities for which it is relatively easy to evade taxes. For example, because the
income from house painting can be done on a cash basis and is therefore harder
for the IRS to detect, this occupation is more attractive than otherwise. Although a
supply of eager and cheap housepainters undoubtedly is greeted warmly by pro-
spective buyers of that service, the work of the extra people drawn to house
painting, or any activity that facilitates tax evasion, would have higher value in some
alternative occupation.

The same argument applies to self-employment generally, as the enhanced
opportunity for noncompliance inefficiently attracts people who would otherwise
be employees. The opportunity for noncompliance can distort resource allocation
in a variety of other ways, such as causing companies that otherwise would not find
it attractive to set up a financial subsidiary, or set up operations in a tax haven, to
facilitate or camouflage abusive avoidance or evasion.

A tax incidence story also lurks here. The supply of eager housepainters bids
down the market price of a house painting job. Thus, the amount of taxes evaded
overstates the benefit of being a tax-evading housepainter. The biggest loser in this
game is the scrupulously honest (or risk-averse) housepainter, who sees his or her
wages bid down by the unscrupulous competition, but who dutifully pays taxes.

Similarly, a tax policy instrument that facilitates evasion for all corporations (as
opposed to noncorporate businesses) might attract entry, so that its effects are
shifted to corporations’ customers through lower prices. The windfall gains to those
companies that successfully play the tax lottery by acting aggressively probably
accrue to the shareholders in their role as residual claimants, perhaps shared to
some extent with the tax managers through their compensation contracts. If there
are particular characteristics of corporations in certain sectors that facilitate evasion

42 Journal of Economic Perspectives



or abusive avoidance, such as the presence of corporate intangibles, the apparent
gains that accrue to firms in these sectors via a lower effective tax rate will be
partially eroded to the extent that competitors have similar characteristics, and thus
the apparent tax gains will partly benefit some other constituency, including this
sector’s customers.

Optimal Tax Enforcement Policy
The mere presence of tax evasion does not imply a failure of policy. Just as it

is not optimal to station a police officer at each street corner to eliminate robbery
and jaywalking completely, it is not optimal to eliminate tax evasion (for a formal
demonstration of this point, see Baldry, 1984). The recognition of tax evasion
introduces a new set of policy instruments whose optimal setting is at issue; for
instance, what should be the extent of audit coverage, the strategy for choosing
audit targets, and the penalty imposed on detected evasion? The reality of tax
evasion also invites a rethinking of standard taxation problems.

With respect to penalties, it has been well known since Becker (1968) that a
government concerned with maximizing the expected utility of a representative
citizen will want to set the penalty for detected crimes as high as possible, so that
even with a low resource cost of enforcement, the overall expected deterrent effect
will be large. But this argument ignores, inter alia, the possibility of a corrupt tax
administrator who abuses the system or, alternatively, harshly punishes someone
who makes an honest mistake. The harsher the penalty, the more damage a corrupt
administrator could inflict and, in the case of an honest mistake, the more capri-
cious the system. Hence the harsher the penalty, the more detailed and cautious
the prosecution process must be. In addition, with harsher penalties, courts may be
more reluctant to find the taxpayer guilty of evasion, so that one practical conse-
quence may be fewer penalties imposed. This argument also flies in the face of the
common notion that the level of punishment should in some sense “fit” the crime.
In the absence of explicitly modeling the interaction between the penalty rate and
administrative costs, analytical models usually assume a ceiling on the penalty rate.

Regarding how many resources to devote to enforcing the tax laws, Slemrod
and Yitzhaki (1987) show that one superficially intuitive rule—increase the prob-
ability of detection until the marginal increase of revenue thus generated equals
the marginal resource cost of so doing—is incorrect. Although the cost of hiring
more auditors, buying better computers, and the like, is a true resource cost, the
revenue brought in does not represent a net gain to the economy, but rather a
transfer from private (noncompliant) citizens to the government. The correct rule
equates the marginal social benefit of reduced evasion (which is not well measured
by the increased revenue) to the marginal resource cost. The distinction suggests
that unregulated privatization of tax enforcement, in which profit-maximizing
firms would maximize revenue collection net of costs, would lead to socially
inefficient overspending on enforcement. Greater enforcement might also entail
nonpecuniary costs, such as invasion of privacy (Slemrod, 2006). In Slemrod and
Yitzhaki (1987), the social benefit is related to the reduced risk-bearing that comes

Joel Slemrod 43



with reduced tax evasion; more generally, social benefits can accrue via increased
efficiency; it would also tend to mitigate the inefficiencies discussed earlier. Cowell
(1990, p. 136) suggests another complication: perhaps a specific social welfare
discount should apply to the utility of those who are found to be guilty of tax
evasion and thus “are known to be antisocial.” In sum, no one has yet compellingly
translated the theoretically correct characterization of optimal enforcement into a
statement about how much evasion should be tolerated. But just as an important
difference exists between oil reserves and “economically recoverable” oil reserves,
a parallel difference exists between tax evasion and economically (read optimally)
recoverable tax evasion.

The tools of efficiency analysis and optimal taxation can be extended to cover
enforcement policy instruments. Ignoring distributional concerns, these tools
reveal that all tax policy instruments—not just the standard instruments such as tax
rates—should be utilized so as to equalize the marginal efficiency cost per dollar of
revenue raised, which should in turn equal the marginal social benefit of raising
revenue (Mayshar, 1991; Slemrod and Yitzhaki, 1996, 2002). Distributional consid-
erations can be introduced into this framework, as well. Unfortunately, the empir-
ical analysis needed to flesh out the policy implications of these rules is not far
along. However, their logic suggests that, given the costs and difficulties of detect-
ing noncompliance of the self-employed, the higher rates of noncompliance make
at least some degree of economic sense.5

The normative theory has not yet made much progress in providing concrete
policy advice regarding the key tools of tax administration, especially the role of
information reporting by arms-length parties. The ability of the IRS to rely on
reports by firms about wages and salaries paid to employees explains why the
(optimal) noncompliance rate of labor income is so much lower than for self-
employment income, for which no such information reports exist. The ability to
match firm-to-firm sales is touted by advocates as a major administrative advantage
of value-added taxes, and the difficulty of monitoring firm-to-consumer sales and to
distinguish them from firm-to-firm sales has been noted as the Achilles heel of
administering a retail sales tax. Overall, when relatively disinterested third parties
can be required to provide information, as with wages and salaries, high compli-
ance rates can be achieved at fairly low cost. But when only interested parties are
involved, an alternative mechanism—such as in a credit-invoice value-added tax
(where taxes on input purchases can be deducted only if the seller produces an
invoice for taxes paid)—must be found or else compliance will be low absent costly

5 The recent theoretical and empirical attention to the elasticity of taxable income is welcome, because
it recognizes that all behavioral responses— of which tax evasion is one—are symptoms of the efficiency
cost of price distortions due to taxation. For useful starting points on this literature, see Feldstein (1999)
on the theory, Gruber and Saez (2002) on the empirical work, and Slemrod (1998) for a critical review
of both. What distinguishes traditional, “real” responses to taxation, such as labor supply, on the one
hand, from evasion responses is that while, with some exceptions, economists generally accept that
inalterable preferences guide the former, the evasion response is mediated by the set of enforcement
policies that government chooses.

44 Journal of Economic Perspectives



auditing. Formal modeling of these differences is in their infancy. Kopczuk and
Slemrod (2006) have suggested a modeling strategy based on the importance of
monitorable arms-length transactions among firms and between firms and employ-
ees, and Gordon and Li (2005) have argued that the financial sector is a key to tax
administration in developing countries.

The ubiquity and importance of evasion call into question one of the canons
of undergraduate public finance textbooks—that the incidence and efficiency of
taxes does not in the long run depend on which side of the market the tax is levied.
Once the reality of tax evasion is recognized, the incidence and efficiency of a tax
system may depend critically on which side of the market remits the tax to the
government and which side must report its transactions to the government. A
uniform value-added tax and a uniform national retail sales tax may look identical
in a world of no evasion or administrative costs, but they may have very different
effects in the real world, a point noted by the President’s Advisory Panel on Federal
Tax Reform (2005, pp. 191–222).

Conclusions

Tax evasion is widespread, always has been, and probably always will be.
Variations in dutifulness and honesty can explain some of the across-individual and,
perhaps, across-country heterogeneity of evasion. But the stark differences in
compliance rates across taxable items that line up closely with detection rates
suggest strongly that deterrence is a powerful factor in evasion decisions.

The overall net noncompliance rate for all U.S. federal taxes and the individ-
ual income tax seems to stand at about 14 percent. But given the current state of
theory and evidence on tax evasion, it isn’t clear in what way or how much
enforcement might most efficiently be increased. Although the normative theory of
taxation has been extended to tax system instruments such as the intensity of
enforcement, the empirical knowledge to put these rules into operation is sparse.
Furthermore, theory is only beginning to address core issues such as the role of
third-party reporting of information that facilitates enforcement of the taxation of
wages and salaries, but helps little for self-employment income. Modeling these
information flows—and the critical role played by firms—is an important item in
the future research agenda.

y I am grateful for insights gained from comments on a earlier draft received from Jim Alm,
Brian Erard, Eric Toder, David Ulph, Shlomo Yitzhaki, the JEP editorial team, and especially
for communications with Richard Highfield of the OECD and Alan Plumley of the IRS. I also
am grateful for research assistance provided by Katherine Blauvelt.

Cheating Ourselves: The Economics of Tax Evasion 45



References

Agha, Ali, and Jonathan Haughton. 1996. “De-
signing VAT Systems: Some Efficiency Consider-
ations.” The Review of Economics and Statistics,
May, 78(2): 303– 8.

Allingham, Maurice, and Agnar Sandmo.
1972. “Income Tax Evasion: A Theoretical Anal-
ysis.” Journal of Public Economics, November, 1(3–
4): 323–38.

Alm, James. 1999. “Tax Compliance and Ad-
ministration.” In Handbook on Taxation, ed. W.
Bartley Hildreth and James A. Richardson, 741–
68. New York: Mercel Dekker.

Alm, James, Betty R. Jackson, and Michael
McKee. 1992. “Estimating the Determinants of
Taxpayer Compliance with Experimental Data.”
Economic Development and Cultural Change, March,
39(4): 107–14.

Andreoni, James, Brian Erard, and Jonathan
Feinstein. 1998. “Tax Compliance.” Journal of
Economic Literature, June, 36(2): 818 – 60.

Baldry, Jonathan. 1984. “The Enforcement of
Tax Laws: Efficiency Implications.” Economic
Record, June, 60(169), 156 –59.

Baldry, Jonathan. 1987. “Income Tax Evasion
and the Tax Schedule: Some Experimental Re-
sults.” Public Finance, 42(3), 357– 83.

Baumeister, Roy F. 1982. “A Self-Presenta-
tional View of Social Phenomena.” Psychological
Bulletin, 91(1): 3–26.

Becker, Gary S. 1968. “Crime and Punish-
ment: An Economic Approach.” Journal of Politi-
cal Economy, March–April, 76(2), 169 –217.

Beron, Kurt J., Helen V. Tauchen, and Ann
Dryden Witte. 1992. “The Effect of Audits and
Socioeconomic Variables on Compliance.” In
Why People Pay Taxes, ed. Joel Slemrod, 67– 89.
Ann Arbor: University of Michigan Press.

Blumenthal, Marsha, Charles Christian, and
Joel Slemrod. 2001. “Do Normative Appeals
Affect Tax Compliance? Evidence from a Con-
trolled Experiment in Minnesota.” National Tax
Journal, March, 54(1): 125–38.

Breusch, Trevor. 2005. “The Canadian Under-
ground Economy: An Examination of Giles and
Tedds.” Canadian Tax Journal, 53(2): 367–91.

Bruce, Donald, and William F. Fox. 2000.
“E-Commerce in the Context of Declining State
Sales Tax Bases.” National Tax Journal, Decem-
ber, 53(4): 1373–90.

Chen, Kong-Ping, and C. Y. Cyrus Chu. 2005.
“Internal Control vs. External Manipulation: A
Model of Corporate Income Tax Evasion,”
RAND Journal of Economics, Spring, 36(1): 151–
64.

Christian, Charles W. 1994. “Voluntary Com-
pliance with the Individual Income Tax: Results
from the 1988 TCMP Study.” In The IRS Research
Bulletin, 1993/1994, Publication 1500 (Rev.
9 –94), 35– 42. Washington, D.C.: Internal Reve-
nue Service.

Clotfelter, Charles. 1983. “Tax Evasion and
Tax Rates: An Analysis of Individual Returns.”
The Review of Economics and Statistics, August,
65(3): 363–73.

Cloyd, Bryan. 1995. “The Effects of Financial
Accounting Conformity on Recommendations
of Tax Preparers.” Journal of the American Taxa-
tion Association. Fall, 17(2): 50 –70.

Cloyd, Bryan, Jamie Pratt, and Toby Stock.
1996. “The Use of Financial Accounting Choice
to Support Aggressive Tax Positions: Public and
Private Firms.” Journal of Accounting Research,
Spring, 34(1): 23– 43.

Cowell, Frank. 1990. Cheating the Government:
The Economics of Evasion. Cambridge, MA: MIT
Press.

Crocker, Keith, and Joel Slemrod. 2005. “Cor-
porate Tax Evasion with Agency Costs.” Journal of
Public Economics, September, 89(9 –10): 1593–1610.

Daunton, Martin. 1998. “Trusting Leviathan:
British Fiscal Administration from the Napole-
onic Wars to the Second World War.” In Trust
and Governance, ed. Valerie Braithwaite and Mar-
garet Levi, 102–34. New York: Russell Sage Foun-
dation.

Dhami, Sanjit, and Ali al-Nowaihi. 2004. “Why Do
People Pay Taxes? Prospect Theory versus Expected
Utility Theory.” http://www.le.ac.uk/economics/
research/RePEc/lec/leecon/dp05-23.pdf.

Everson, Mark. 2005. Testimony given to the
President’s Advisory Panel on Federal Tax Re-
form. March 3.

Falkinger, Josef. 1995. “Tax Evasion, Con-
sumption of Public Goods, and Fairness,” Journal
of Economic Psychology, March, 16(1): 63–72.

Feige, Edgar L. 1989. “Monetary Methods of
Estimating Informal Activity in Developing Na-
tions.” In Demographic Change and Economic Devel-
opment, ed. Alois Wenig and Klaus F. Zimmer-
man, 211–32. Berlin: Springer Verlag.

Feinstein, Jonathan. 1991. “An Econometric
Analysis of Income Tax Evasion and its Detec-
tion.” RAND Journal of Economics, Spring, 22(1):
14 –35.

Feld, Lars P., and Bruno S. Frey. 2002. “Trust
Breeds Trust: How Taxpayers Are Treated.”
Economics of Governance, July, 3(2): 87–99.

Feldman, Naomi, and Joel Slemrod. Forth-

46 Journal of Economic Perspectives

http://www.ingentaconnect.com/content/external-references?article=0033-2909()91:1L.3[aid=5054122]
http://www.ingentaconnect.com/content/external-references?article=0033-2909()91:1L.3[aid=5054122]
http://www.ingentaconnect.com/content/external-references?article=1435-6104()3:8L.7[aid=7708450]
http://www.le.ac.uk/economics/research/RePEc/lec/leecon/dp05-23.pdf
http://www.le.ac.uk/economics/research/RePEc/lec/leecon/dp05-23.pdf


coming. “Estimating Tax Noncompliance with
Evidence from Unaudited Tax Returns.” The Eco-
nomic Journal.

Feldstein, Martin. 1999. “Tax Avoidance and
the Deadweight Loss of the Income Tax.” The
Review of Economics and Statistics, November,
81(4): 674 – 80.

Frey, Bruno. 1997. “A Constitution for Knaves
Crowds Out Civic Virtues.” The Economic Journal,
July, 107(443): 1043–53.

Gordon, Roger H., and Wei Li. 2005. “Tax
Structure in Developing Countries: Many Puz-
zles and a Possible Explanation.” NBER Working
Paper 11267.

Gruber, Jonathan, and Emmanuel Saez. 2002.
“The Elasticity of Taxable Income: Evidence and
Implications.” Journal of Public Economics, April,
84(1): 1–32.

Hanlon, Michelle, Lillian Mills, and Joel Slem-
rod. Forthcoming. “An Empirical Examination
of Corporate Tax Noncompliance.” In Taxing
Corporate Income in the 21st Century, ed. Alan Auer-
bach, James R. Hines Jr., and Joel Slemrod. Cam-
bridge: Cambridge University Press.

Hanousek, Jan, and Filip Palda. 2004. “Quality
of Government Services and the Civic Duty to
Pay Taxes in the Czech and Slovak Republics,
and other Transition Countries.” Kyklos, 57(2):
237–52.

HM Revenues and Customs. 2005. Measuring
Indirect Tax Losses—2005. London, December.

Herman, Tom, and Rachel Emma Silverman.
2005. “IRS to Increase Audits Next Year.” Wall
Street Journal, November 23, p. D1.

Johnston, David Cay. 2003. “Tax Inquiries Fall
as Cheating Increases.” New York Times, April 14,
p. A16.

Jones, Carolyn C. 1996. “Mass-based Income
Taxation: Creating a Taxpayer Culture, 1940 –52.”
In Funding the American State: The Rise and Fall of the
Era of Easy Finance, ed. W. Elliot Brownlee. New
York: Woodrow Wilson Center Press and Cam-
bridge University Press.

Kahnemann, Daniel, and Amos Tversky. 1979.
“Prospect Theory: An Analysis of Decision under
Risk.” Econometrica, March, 47(2): 263–91.

Kang, Sun Won, and Hugh Rockoff. 2006.
“Capitalizing Patriotism: The Liberty Loans of
World War I.” NBER Working Paper 11919.

Klepper, Steven, and Daniel Nagin. 1989.
“The Anatomy of Tax Evasion.” Journal of Law,
Economics, and Organization, Spring, 5(1): 1–24.

Kopczuk, Wojciech, and Joel Slemrod. 2006.
“Putting Firms into Optimal Tax Theory.” Amer-
ican Economic Review Papers and Proceedings, May,
96(2): 130 –34.

Levi, Margaret. 1998. “A State of Trust.” In

Trust and Governance, ed. Valerie Braithwaite and
Margaret Levi, 77–101. New York: Russell Sage
Foundation.

Mayshar, Joram. 1991. “Taxation with Costly
Administration.” Scandinavian Journal of Econom-
ics, 93(1): 75– 88.

Mummert, Annette, and Friedrich Schneider.
2002. “The German Shadow Economy: Parted in
a United Germany?” Finanzarchiv: Public Finance
Analysis, July, 58(3): 286 –316.

O’Donnell, Gus. 2004. Financing Britain’s Fu-
ture: Review of the Revenue Departments. Presented
to Parliament by the Chancellor of the Ex-
chequer by Command of Her Majesty, HM
Treasury. March 17. CM 6163.http://www.
hmtreasury.gov.uk/budget/budget_04/associated_
documents/bud_bud.

Oldman, Oliver, and LaVerne Woods. 1983.
“Would a Value-Added System Relieve Tax Com-
pliance Problems?” In Income Tax Compliance: A
Report of the ABA Section on Taxation Invitational
Conference on Income Tax Compliance. Chicago:
American Bar Association.

Organisation for Economic Co-operation and
Development. 2004. Tax Administration in OECD
Countries: Comparative Information Series. Centre
for Tax Policy Administration. Paris, France.

Pedone, Antonio. 1981. “Italy.” In The Value
Added Tax: Lessons from Europe, ed. Henry J.
Aaron, 31– 42. Washington D.C.: The Brookings
Institution.

Petrick, Kenneth A. 2002. Corporate Profits: Prof-
its before Tax, Profits Tax Liability, and Dividends.
Methodology paper, U.S. Department of Com-
merce, Economics and Statistics Administration,
Bureau of Economic Analysis, September.
http://bea.gov/bea/ARTICLES/NATIONAL/
NIPA/Methpap/methpap2.pdf.

Pissarides, Christopher A., and Guglielmo We-
ber. 1989. “An Expenditure-Based Estimate of
Britain’s Black Economy.” Journal of Public
Economics, June, 39(1): 17–32.

President’s Advisory Panel on Federal Tax Re-
form. 2005. Simple, Fair, and Pro-Growth: Proposals
to Fix America’s Tax System. Washington, D.C.:
U.S. Government Printing Office.

Rice, Eric. 1992. “The Corporate Tax Gap:
Evidence on Tax Compliance by Small Corpora-
tions.” In Why People Pay Taxes, ed. Joel Slemrod,
125– 61. Ann Arbor: University of Michigan.

Sandmo, Agnar. 2005. “The Theory of Tax
Evasion: A Retrospective View.” National Tax
Journal, December, 58(4), pp. 643– 63.

Scholz, John T., and Mark Lubell. 2001. “Co-
operation, Reciprocity, and the Collective Ac-
tion Heuristic.” American Journal of Political Sci-
ence, January, 45(1): 160 –78.

Joel Slemrod 47

http://www.ingentaconnect.com/content/external-references?article=0023-5962()57:2L.237[aid=7708463]
http://www.ingentaconnect.com/content/external-references?article=0023-5962()57:2L.237[aid=7708463]
http://bea.gov/bea/ARTICLES/NATIONAL/NIPA/Methpap/methpap2.pdf
http://bea.gov/bea/ARTICLES/NATIONAL/NIPA/Methpap/methpap2.pdf
http://www.hmtreasury.gov.uk/budget/budget_04/associated_ documents/bud_bud
http://www.hmtreasury.gov.uk/budget/budget_04/associated_ documents/bud_bud
http://www.hmtreasury.gov.uk/budget/budget_04/associated_ documents/bud_bud


Silvani, Carlos and John Brondolo. 1993. “An
Analysis of VAT Compliance.” International
Monetary Fund Fiscal Affairs Department
mimeo, November.

Slemrod, Joel. 1998. “Methodological Issues
in Measuring and Interpreting Taxable Income
Elasticities.” National Tax Journal, December, 51(4):
773– 88.

Slemrod, Joel. 2003. “Trust in Public Fi-
nance.” In Public Finance and Public Policy in the
New Century, ed. Sijbren Cnossen and Hans-
Werner Sinn, 49 – 88. Cambridge, MA: MIT
Press.

Slemrod, Joel. 2004. “The Economics of Cor-
porate Tax Selfishness.” National Tax Journal,
December, 57(4): 877–99.

Slemrod, Joel. 2006. “Taxation and Big Brother:
Information, Personalization, and Privacy in 21st

Century Tax Policy.” Fiscal Studies, March, 27(1):
1–15.

Slemrod, Joel, and Jon Bakija. 2004. Taxing
Ourselves: A Citizen’s Guide to the Debate over Taxes.
3rd ed. Cambridge, MA: MIT Press.

Slemrod, Joel, Marsha Blumenthal, and
Charles Christian. 2001. “Taxpayer Response to
an Increased Probability of Audit: Evidence
from a Controlled Experiment in Minnesota.”
Journal of Public Economics, March, 79(3): 455– 83.

Slemrod, Joel, and Shlomo Yitzhaki. 1987.
“The Optimal Size of a Tax Collection Agency.”
Scandinavian Journal of Economics, 89(2): 25–34.

Slemrod, Joel, and Shlomo Yitzhaki. 1996.
“The Cost of Taxation and the Marginal Effi-
ciency Cost of Funds.” International Monetary
Fund Staff Papers, March, 43(1): 172–98.

Slemrod, Joel, and Shlomo Yitzhaki. 2002.
“Tax Avoidance, Evasion, and Administration.”
In Handbook of Public Economics, ed. Alan J. Auer-
bach and Martin S. Feldstein. Volume 3, pp.
1423–70. Amsterdam: Elsevier B.V.

Spicer, Michael W., and Lee A. Becker. 1980.
“Fiscal Inequity and Tax Evasion: An Experimen-
tal Approach.” National Tax Journal, June, 33(2):
171–75.

Swedish Tax Agency. 2004. Taxes in Sweden.
Solna, Sweden.

Tait, Alan. 1988. Value Added Tax: International
Practice and Problems. Washington, D.C.: Interna-
tional Monetary Fund

Tanzi, Vito. 1980. “The Underground Econ-
omy in the United States: Estimates and Impli-
cations.” Banco Nazionale del Lavoro Quarterly Re-
view. December, 135(2): 427–53.

Tanzi, Vito. 1983. “The Underground Econ-
omy in the United States: Annual Estimates,
1930-80.” IMF Staff Papers. June, 30(2): 283–305.

Tanzi, Vito. 1999. “Uses and Abuses of Esti-
mates of the Underground Economy.” The Eco-
nomic Journal, June, 109(456): F338 –F347.

Torgler, Benno. 2003. “Tax Morale, Rule-Gov-
erned Behaviour, and Trust.” Constitutional Polit-
ical Economy, June, 14(2): 119 –140.

Torgler, Benno. 2004. “Moral Suasion: An Al-
ternative Tax Policy Strategy? Evidence from a
Controlled Field Experiment.” Economics of Gov-
ernance, November, 5(3): 235–53.

U.S. Department of the Treasury. Internal
Revenue Service. 1996. Federal Tax Compliance
Research: Individual Income Tax Gap Estimates for
1985, 1988, and 1992. Pub. 1415 (rev. 4-96),
April, Washington, D.C.

U.S. Department of the Treasury. Internal
Revenue Service. 2005a. IRS Data Book: 2004.
Publication 55b, Washington, D.C. From http://
www.irs.gov/taxstats/compliancestats/article/
0,,id�97177,00.html.

U.S. Department of the Treasury. Internal
Revenue Service. 2005b. New IRS Study Provides
Preliminary Tax Gap Measure. IR-2005-38, March
29. Washington, D.C.

U.S. Department of the Treasury. Internal Rev-
enue Service. 2005c. Preliminary Update of the Tax
Year 2001 Individual Income Tax Underreporting Gap
Estimates, 8 –16. Washington, D.C.: Office of
Research, Analysis, and Statistics. From http://
www.irs.gov/taxstats/compliancestats/article/
0,,id�97177,00.html.

U.S. Department of the Treasury. Internal
Revenue Service. 2005d. Tax Gap Facts and Fig-
ures. Washington, D.C.

U.S. Department of the Treasury. Internal
Revenue Service. 2005e. Understanding the Tax
Gap. FS-2005-14, March. Washington, D.C.

U.S. Department of the Treasury. Internal Rev-
enue Service. 2006. Updated Estimates of the TY 2001
Individual Income Tax Underreporting Gap. Overview.
February 22. Washington, D.C.: Office of Re-
search, Analysis, and Statistics.

U.S. Department of the Treasury. Internal
Revenue Service Oversight Board. 2006. 2005
Taxpayer Attitude Survey. http://www.ustreas.
gov/irsob/releases/2006/02212006.pdf.

U.S. General Accounting Office. 1988. IRS’
Tax Gap Studies. GAO/GGD-88-66BR. Washing-
ton, D.C.

Webber, Carolyn, and Aaron B. Wildavsky.
1986. History of Taxation and Expenditure in the
Western World. New York: Simon & Schuster.

Yitzhaki, Shlomo. 1974. “A Note on ‘Income
Tax Evasion: A Theoretical Analysis.’” Journal of
Public Economics, May, 3(2): 201–2.

48 Journal of Economic Perspectives

http://www.ingentaconnect.com/content/external-references?article=1435-6104()5:23L.5[aid=7708467]
http://www.ingentaconnect.com/content/external-references?article=1435-6104()5:23L.5[aid=7708467]
http://www.ustreas.gov/irsob/releases/2006/02212006.pdf
http://www.ustreas.gov/irsob/releases/2006/02212006.pdf
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html
http://www.irs.gov/taxstats/compliancestats/article/0,,id=97177,00.html