Vous êtes sur la page 1sur 44

Tax Compliance and Administration

James Alm 1

I. Introduction
Although it is commonly said that the only things certain in life are death and
taxes, it is unmistakable that taxes are in fact far from inevitable. Individuals do
not like paying taxes, and they take a variety of actions to reduce their tax
liabilities. Some of these actions can be classified as tax avoidance, or the legal
reduction in tax liabilities by practices that take full advantage of the tax code,
such as income splitting, postponement of taxes, and tax arbitrage across income
that faces different tax treatment. Tax evasion consists of illegal and intentional
actions taken by individuals to reduce their legally due tax obligations.
Individuals and firms can evade taxes by underreporting incomes, sales, or
wealth; by overstating deductions, exemptions, or credits; or by failing to file
appropriate tax returns. For its part, government must take actions to ensure
compliance with the tax laws.
As discussed in more detail later, tax evasion is difficult to measure. Still, there is
widespread evidence that tax evasion is extensive and commonplace in nearly all
countries. For the United States, the most reliable estimates suggest that the
amount of unpaid federal individual and corporate income taxes totaled $127
billion for 1992, with an annual growth rate of 10 percent since 1973 (Internal
Revenue Service, 1988, 1990, 1996). Other taxes at other levels of government
are also subject to nonpayment. Evidence from other countries clearly indicates
that the American experience is not an isolated one.

James Alm is Professor of Economics and Chairman of the Economics Department at the
Andrew Young School of Policy Studies, Georgia State University. This paper was originally
published as chapter in the Handbook on Taxation (New York: Marcel Dekker, 1999).

Tax evasion is important for many reasons. The most obvious is that its presence
reduces tax collections, thereby affecting taxes that compliant taxpayers face and
public services that citizens receive. Evasion creates misallocations in resource
use when individuals alter their behavior to cheat on their taxes, such as in their
choices of hours to work, occupations to enter, and investments to undertake. Its
presence requires that government expend resources to deter noncompliance, to
detect its magnitude, and to penalize its practitioners. Noncompliance alters the
distribution of income in unpredictable ways. Evasion may contribute to feelings
of unfair treatment and disrespect for the law. It affects the accuracy of
macroeconomic statistics. More broadly, it is not possible to understand the true
impact of taxation without recognizing the existence of tax evasion.
In this chapter I discuss the current research on tax compliance and
administration. This literature has grown enormously in only the last 25 years, and
it has generated numerous and important insights. The vast bulk of this research
has examined compliance and administration of the individual income tax in the
United States, and it is this area that I examine. I focus in particular on several
key issues in this research. How extensive is tax evasion? What factors motivate
individuals to cheat on their taxes? What are appropriate government policies
toward evasion? What is the evidence -- empirical and experimental -on
individual behavioral responses? What has been learned, and also what remains
to be learned from this research? Each issue is discussed in turn.

II. Measuring Noncompliance


A major difficulty in analyzing evasion is its measurement. After all, individuals
have incentives to conceal their cheating. Several methods have been developed
to measure evasion. All are subject to much imprecision and controversy.
One method relies on information generated by the tax authority as part of its
audit process. In the United States, the Internal Revenue Service (IRS) has since
1965 conducted detailed line-by-line audits of a stratified random sample of
roughly 50,000 individual tax returns on a roughly 3-year cycle for its Taxpayer
Compliance Measurement Program (TCMP); the most recent TCMP was
performed in 1988. These audits yield an IRS estimate of the taxpayer's "true"
income, which allows measures of individual and aggregate income tax evasion to
be calculated. Such estimates are probably the most accurate that are available.
However, TCMP data have some serious and well-recognized deficiencies: the
audits do not detect all underreported income, nonfilers are not often captured,
honest errors are not identified, and final audit adjustments are not included.

Another direct method involves surveys. These surveys are typically designed to
elicit taxpayer attitudes about the roles that such factors as perceptions of the
probability of detection, the fairness of taxation, and the responsiveness of
government play in their reporting decision. The surveys can also be used to
estimate noncompliance. However, the accuracy of surveys is uncertain:
individuals may not remember their reporting decisions, they may not respond
truthfully or at all, and the respondents may not be representative of all taxpayers.
Surveys are also unable to control for many relevant determinants of
noncompliance. Finally, surveys cannot determine the direction of causality
between evasion and its determinants; that is, statements about, say, the inequity
of the income tax may result from an ex post rationalization for noncompliance
rather than be the ex ante cause of noncompliance.
A variety of indirect methods have attempted to infer the magnitude of unreported
income from its traces in other, observable areas. These methods have typically
been used to measure the amount of activities that take place outside formal
markets, in what has commonly been called the underground, shadow, irregular,
subterranean, or black economy. One approach looks at the discrepancy between
income and expenditure, either in budget surveys or in national income accounts.
Another looks at the discrepancy between "official" labor force participation rates
and estimates of "true" participation rates. A related approach assumes that there
is a fixed and predictable relationship between some observable variable and the
amount of unreported income; the most common application here looks for traces
of unreported income in monetary aggregates, but other variables (e.g., electricity)
have also been used. All of these methods are subject to serious criticisms. They
may simply compound measurement errors, they attribute all discrepancies to
unreported income, and they are often able only to estimate the change in
unreported income over some period, not its absolute level.
Despite these measurement difficulties, tax evasion appears to be a widespread
and growing problem in the United States. As noted earlier, the most reliable
estimates for the United States project the amount of unpaid federal individual
and corporate income taxes at $127 billion for 1992. Of this total, $94 billion is
estimated as unpaid individual income taxes, consisting of $72.4 billion from
underreporting of income, $10.2 billion from nonfiling of returns, and $11.4
billion from underpaying of taxes. Overall, roughly 85 percent of individual
income taxes that are due are actually collected. Estimates from a variety of
methods for other countries, such as Argentina (Herschel, 1978), the Netherlands
(Hessing, et al., 1987), the Philippines (Manasan, 1988), Jamaica (Alm, Bahl, and
Murray, 1990, 1993), and Spain (de Juan, Lasheras, and Mayo, 1994), indicate
that tax evasion is a pervasive and extensive phenomenon.

III. Theoretical Foundations


Compliance Decision

of

the

Individual

Tax

Since Allingham and Sandmo (1972) and Srinivasan (1973), the standard
approach to the analysis of tax compliance has relied upon the economics-ofcrime methodology pioneered by Becker (1968). Here a rational individual is
assumed to maximize the expected utility of the evasion gamble, balancing the
benefits of successful evasion with the risky prospect of detection and
punishment. Although their work has been extended in a variety of dimensions,
nearly all models continue to use their basic approach.
In this section I first review the basic model of individual compliance behavior.
There have been numerous extensions to the basic model, and I then discuss these
extensions. This overall literature is then assessed.
A. The Basic Model of Individual Choice
The standard economics-of-crime model of compliance is based upon the work of
Allingham and Sandmo (1972) and Srinivasan (1973). In its simplest form, an
individual is assumed to receive a fixed amount of income I, and must choose
how much of this income to declare to the tax authorities and how much to
underreport. The individual pays taxes at rate t on every dollar D of income that
is declared, while no taxes are paid on underreported income. However, the
individual may be audited with a fixed, random probability p, however; if audited,
then all underreported income is discovered, and the individual must pay a
penalty at rate f on each dollar that he or she was supposed to pay in taxes but did
not pay. The individual's income IC if caught underreporting equals
I=I-tD-f[t(I-D)], while if underreporting is not caught income IN is IN=I - tD. The
individual chooses declared income to maximize the expected utility E U(I) of the
evasion gamble, or E U(I)=pU(IC)+(1-p)U(IN), where E is the expectation
operator and utility U(I)is a function only of income. This optimization generates
a standard first-order condition for an interior solution; given concavity of the
utility function, the second-order condition will be satisfied.
Note that the probability of detection is assumed here to be fixed and random, so
that the audit agency is not allowed to use information from the taxpayers' returns
in determining whom to select for audit. It seems obvious that the tax agency can
do better in identifying tax evaders if it uses this initial transmission of
information from taxpayers than if it simply ignores the information and audits all
taxpayers with equal frequency. Various audit schemes that allow the tax agency
to adjust its audit selection in light of information provided by the taxpayer are
discussed later, when the behavior of the tax authority is examined.

Comparative statics results are easily derived. It is straightforward to show that


an increase in the probability of detection p and the penalty rate f unambiguously
increase declared income. An increase in income has an ambiguous effect on
declared income, an effect that depends upon the individual's attitude toward risk.
Surprisingly, an increase in the tax rate t has an ambiguous effect on declared
income. A higher tax rate increases the return to cheating, which reduces the
amount of declared income. However, a higher tax rate also reduces income; if,
as is usually assumed, the individual exhibits decreasing absolute risk aversion,
then the lower income makes the evasion gamble less attractive and declared
income increases accordingly. In fact, Yitzhaki (1974) has shown that a higher
tax rate will increase declared income when the penalty is imposed at a
proportional rate on evaded taxes.
This economics-of-crime approach gives the sensible result that compliance
depends upon enforcement. However, it is essential to recognize that this
approach also concludes that an individual pays taxes because -- and only because
-- of the fear of detection and punishment. Again, this is a plausible and
productive insight, with the obvious implication that the government can
encourage greater tax compliance by increasing the audit and the penalty rates.
This theme is discussed in detail later when government policies are considered.
However, it is clear to many observers that compliance cannot be explained
entirely by the financial incentives generated by the level of enforcement (Graetz
and Wilde, 1985; Smith and Kinsey, 1987; Elffers, 1991). Consider, for example,
enforcement policies in the United States. The percentage of individual income
tax returns that are subject to a thorough tax audit is quite small and has fallen in
recent years to roughly 1 percent. Similarly, the penalty on even fraudulent
evasion is only 75 percent of unpaid taxes, and these penalties are infrequently
imposed; civil penalties on non-fraudulent evasion are even less (e.g., 20 percent
of unpaid taxes). A purely economic analysis of the evasion gamble suggests that
most rational individuals should either underreport income not subject to source
withholding or overclaim deductions not subject to independent verification
because it is extremely unlikely that such cheating will be caught and penalized.
However, most individuals pay most of their taxes most of the time, and there are
substantial numbers of individuals who apparently pay all of their taxes regardless
of the financial incentives they face from the enforcement regime.

This dilemma can be illustrated more precisely. Suppose that the utility function
1 e
of the individual is I i /(1 e) , where the subscript i refers to the state of the world

(i=C,N) and e is a measure of the individual's constant relative risk aversion.


Using the definitions of IC and IN, the expected utility maximization can then be
solved for the optimum amount of declared income D*. Now suppose that D* is
calculated for specific, realistic values of the various parameters. For example, if
t=0.4, f=2, p=0.02, and e=1, then the individual will optimally declare no
income. Very large values for relative risk aversion are required to generate
compliance consistent with actual U.S. experience. When e=3, declared income
is only 14 percent of true income; when e=5, it is still only 44 percent; when
e=10, it is 71 percent. Risk aversion must exceed 30 for compliance to exceed 90
percent. However, existing field evidence on the coefficient of relative risk
aversion suggests that e ranges between 1 and 2. Risk aversion must be
abnormally large for behavior to be even roughly comparable to actual observed
choices.
As a further illustration of the difficulties of the basic model, consider a slight
reinterpretation. Denoting the taxes paid by the individual on declared income as
T and the fine on unreported income as F, the expected utility of the individual if
he or she declares no income is pU(I-F)+(1-p)U(I), while the certain utility if he
or she declares all income is U(I-T). The maximum amount of taxes that the
individual will voluntarily pay can be found by equating these two expressions
and solving for T; that is; the individual will voluntarily pay taxes only until
utility with full declaration equals expected utility with no declaration. Using a
linear approximation for the utility function, which implies that the individual is
risk neutral, the solution for T is pF, so that taxes equal the expected value of the
penalty. An individual who paid more than the expected value of the penalty is
worse off than if he or she took a chance in the audit lottery. Even if the fine is as
high as half of income, a low value for the probability of detection suggests that
voluntary compliance will also be low.
In short, the basic model of individual compliance behavior implies that rational
individuals should report virtually no income. However, compliance with the
individual income tax remains relatively high. It seems implausible that
government enforcement activities alone can account for these levels of
compliance; the basic model, in its reliance on expected utility theory, is certainly
unable to explain this behavior. Indeed, the puzzle of tax compliance behavior is
why people pay taxes, not why they evade them (Alm, McClelland, and Schulze,
1992). This observation suggests that the compliance decision must be affected
by other factors not mentioned by the basic model, or it must be affected in ways
not captured by the theory. These considerations are discussed next.

B. Extensions of the Basic Model


The basic model of individual choice has undergone numerous refinements and
extensions since first presented by Allingham and Sandmo (1972) and Srinivasan
(1973). Nearly all models of evasion retain its basic features, especially its
reliance on expected utility theory. Other models attempt to incorporate
noneconomic factors in the compliance decision. Both areas are discussed.
1. Economic Models
The vast bulk of the theoretical work has attempted to introduce formally other
factors thought to be relevant to the individual compliance decision. One obvious
extension is to allow the individual to choose declared income jointly with
additional variables, such as labor supply (Pencavel, 1979; Cowell, 1981),
occupational choice (Pestieau and Possen, 1991), and tax avoidance schemes
(Cross and Shaw, 1982; Alm, 1988a). Alternative penalty, tax, and tax
withholding functions have been considered (Pencavel, 1979; Kesselman, 1989;
Yaniv, 1988). The impact of complexity and uncertainty about the relevant fiscal
parameters has been analyzed (Alm, 1988b; Beck and Jung, 1989a; Scotchmer
and Slemrod, 1989; Cronshaw and Alm, 1995). An increasing number of
individuals use paid practitioners in the preparation of their tax return, and the
effects of such usage on compliance has been examined (Klepper and Nagin,
1989b; Scotchmer, 1989; Reinganum and Wilde, 1991; Erard, 1993). Individuals
receive something from government for their tax payments, and the receipt of
government services has been shown to affect the compliance decision (Cowell
and Gordon, 1988); that is, individuals pay taxes because they value what they get
for their taxes, and they pay more in taxes the more responsive is government in
providing what they value. A related aspect is that individuals may be responsive
to positive rewards given to them if they are found in the audit process to be
honest (Falkinger and Walther, 1991). As discussed later, individuals do not
typically face a fixed and random probability of audit. Instead, the tax authority
often uses information from the tax returns to determine strategically whom to
audit, so that the audit probability is endogenous and dependent in part on the
behavior of the taxpayer and the tax authority (Reinganum and Wilde, 1985,
1986; Graetz, Reinganum, and Wilde, 1986; Beck and Jung, 1989b; Cronshaw
and Alm, 1995). A similar element that generates an endogenous audit selection
rule is that individuals pay taxes repeatedly over time, and the audit agency can
utilize this intertemporal information in the strategic selection of tax returns
(Landsberger and Meilijson, 1982; Rickard, Russell, and Howroyd, 1982;
Greenberg, 1984). Other factors are surely relevant. To date, no single theory has
been able to incorporate more than a few of these factors in a meaningful way.

Unfortunately, the numerous refinements and extensions considerably complicate


the theoretical analyses, and generally render clear-cut analytical results
impossible. Consider, for example, the addition of labor supply choice to the
standard model. Even in the basic model with the assumption of fixed income,
any change in the marginal tax rate has an ambiguous effect on compliance. With
labor supply endogenous, a change in the marginal tax rate has the standard
substitution and income effects on labor supply, thereby making the level of
earned income endogenous. It should not be surprising that the effect of tax
changes on declared income becomes even more complicated than in the basic
model. Similar comments apply to other variations.
2. Noneconomic Factors
There has also been some work to expand the basic model of individual choice by
introducing some aspects of behavior or motivation considered explicitly by other
social sciences. Many of these aspects can be incorporated in "prospect theory,"
as developed by Kahneman and Tversky (1979). Other approaches that consider
such factors as deviancy, personal and situational characteristics, social contexts,
and attribution theory have also been usefully applied.
A first factor is the way in which individuals perceive probabilities. There is
overwhelming evidence from psychology that individuals "overweight" the low
probabilities that they face in tax compliance (Kahneman and Tversky, 1979); that
is, even when fully informed, individuals will systematically act as if the
probability of audit that they face is much higher than its actual probability.
Overweighting may therefore provide an additional explanation for tax
compliance. If taxpayers give more weight to the probability of an audit than they
ought to (at least relative to an expected utility model), then compliance will be
greater than the level predicted by the standard economics approach. Similar
conclusions can be generated by generalized expected utility theory (Quiggin,
1993).
A related factor suggested by Kahneman and Tversky (1979) is that many
individuals apparently adapt to an unchanged environment and perceive stimuli
relative to the environment. Many individuals react much differently to gains than
to (equal-but-opposite valued) losses. Kahneman and Tversky (1979) therefore
suggest that individuals act on the basis of a "value function" (rather than the
utility function in economic models). The value function is assumed to depend
upon changes in income from some reference point, rather than the level of
income itself. It is also assumed to be steeper for losses than for gains because a
loss in income is disliked much more than an equal gain, and it is concave for
gains (risk aversion) but convex for losses (risk seeking), so that individuals may

exhibit risk-averse behavior when confronted with risky but positive gambles,
while the same individuals may become risk-lovers when faced with gambles that
involve possible losses. The relevance of these assumptions for tax compliance is
subtle yet powerful. Since some individuals frame any payment of taxes as a loss,
these individuals will be likely to engage in risk-seeking behavior; that is, these
individuals will declare less income than predicted by the basic model of expected
utility theory.
A third factor is that there is much evidence of what may be termed a "social
norm" of tax compliance. Although difficult to define precisely, a social norm
can be distinguished by the feature that it is process-oriented, unlike the outcomeorientation of individual rationality (Elster, 1989). A social norm therefore
represents a pattern of behavior that is judged in a similar way by others and that
therefore is sustained in part by social approval or disapproval. Consequently if
others behave according to some socially accepted mode of behavior, then the
individual will behave appropriately; if others do not so behave, then the
individual will respond in kind. The presence of a social norm is also consistent
with the framework suggested by Kahneman and Tversky (1979). It is also
consistent with other approaches, such as those that rely upon social customs or
upon individual feelings of morality, guilt, and alienation.
Overall, then, this last factor suggests that an individual will comply as long as he
or she believes that compliance is the social norm. Conversely, if noncompliance
becomes pervasive, then the social norm of compliance disappears. This
perspective also suggests that, if government can affect the social norm of
compliance, then such government policies represent another, potentially
significant tool in government's battle with tax evaders. Of course, policies to
change the social norm of compliance are difficult to determine in theory. Some
possible policies are discussed later when the behavior of the tax authority is
considered.
There is considerable intuitive appeal to the potential importance of social norms
in tax compliance behavior. There is overwhelming evidence that many countries
with roughly the same fiscal system exhibit far different patterns of compliance.
There is also much survey evidence from many countries that indicates that
compliance is strongly affected by the strength and commitment to the social
norm of compliance. These surveys conclude, among other things, that those who
comply view tax evasion as "immoral," that compliance is higher if a "moral
appeal" to taxpayer is made by government, that the low social standing of tax
evaders can be an effective deterrent, that individuals with tax evaders as friends
are more likely to be evaders themselves, and that compliance is greater in

communities with a stronger sense of social cohesion. Other survey evidence


suggests that some people won't pay their taxes if they dislike the way their taxes
are spent, if they feel they have no say in the decision process, if they feel that
government is unresponsive to their wishes, or if they feel that they are treated
unfairly by government; there is also some empirical, experimental, and
simulation evidence that compliance is affected by the nature of the collective
decision process, at least in democratic countries (Pommerehne and WeckHannemann, 1989; Alm, Jackson, and McKee, 1993; Pommerehne, Hart, and
Frey, 1994). It seems clear that such sentiments play an important, perhaps a
dominant role, in tax compliance.
In their entirety, these various influences on the "social norm" of compliance can
be classified into two basic categories. The first relates to how the taxpayer
judges his or her own compliance behavior in light of the individual's own
feelings about what is proper, acceptable, or moral behavior (what might be
termed "internal norms"). The second relates to how the taxpayer feels he or she
is treated by government in such areas as the payment of taxes, the receipt of
government services, or the responsiveness of government decisions (or "external
norms"). Both categories can be analyzed with prospect theory. They can also be
usefully examined from a number of other perspectives.
Still, applications of all of these alternative approaches to tax compliance remain
somewhat limited. Virtually all theoretical work on tax compliance continues to
rely in some form upon the expected utility model.
3. Conclusion
In its entirety, there is no question that the theoretical analysis of compliance
behavior has generated many insights, especially regarding how an individual
responds to greater enforcement activities. Paradoxically, however, this literature
is in a sense both too complex and too simple. It is too complex because it is only
in the simpler models that clear-cut analytical results can be generated on the
compliance impact of basic policy parameters. When more complex dimensions
of individual behavior are introduced, the theoretical results generally become
ambiguous. It is doubtful that theoretical analysis will yield more meaningful
results in the future.
The theoretical models of individual choice are also too simple. There are
numerous factors that affect the reporting decisions of individuals; for example,
the Internal Revenue Service (1978) has listed 64 factors that may affect the
reporting decisions of taxpayers. However, theoretical models are capable of
including only a few.

10

In short, and as emphasized above, the limited ability to incorporate many


relevant factors or to incorporate them in a meaningful way has meant that
theories based largely upon expected utility theory are unable to explain the level
of tax reporting, even when they have more success in explaining the change in
reporting in response to policy innovations. In particular, these models generally
imply that rational individuals should pay far less in taxes than they actually do.
This is not a mere quibble. It goes to the heart of the basic model, as well as its
many extensions, for explaining compliance.

IV. The Design and Administration of Tax Systems


All tax administrations exist to ensure compliance with the tax laws. This
administrative dimension of taxation has long been recognized by tax
administrators, especially those working on tax policy in developing countries
(Goode, 1981; Bird, 1989; Bird and Casangera de Jantscher, 1992). However,
there is relatively little systematic analysis of this dimension, at least by
economists. The available evidence from government budgetary information
clearly indicates that the budget cost of collecting individual income, business
income, and sales taxes is generally in excess of 1 percent of the revenues from
these taxes, and can sometimes be substantially higher (Sandford, 1995). In the
United States, for example, the IRS collected $1.3 trillion with a budget of $7.4
billion in 1994, so that the cost of collecting $100 of taxes was only $.58 (Internal
Revenue Service, 1995). Unfortunately, there is little information on how these
costs vary with various policy tools; that is, it seems likely that the administrative
costs change in large and discrete amounts with the scale of collections and that
they may also display economies of scale in their collections, but these aspects of
the collection cost technology are not known.
As emphasized by Bagchi, Bird, and Das-Gupta (1995), it is helpful to view the
tax administration process as a production function, in which "inputs" (e.g.,
personnel, materials, information, laws, procedures) are used to produce output
(e.g., government revenue, taxpayer equity, social welfare). The theoretical
analyses discussed above suggest a range of inputs that government can pursue to
increase at least one tax administrative output, or government revenues.
However, an equally important aspect is the desirability of these policies, or their
impacts on equity and welfare. Accordingly, I focus first on the positive analysis
of policies to increase compliance; that is, what are the effects of different policy
innovations on the level of compliance? I then examine the normative aspects of
these policies; that is, what policies should government pursue, under various
assumptions about the appropriate goals of government? Throughout, I focus on

11

the broad design of tax administration and say little about its actual, practical
details.
A. Government Efforts to Increase Tax Compliance
1. Increasing Tax Enforcement
The standard administrative prescription for increased tax compliance comes
directly and quickly from the basic model of individual choice. Recall that this
analysis shows that increases in penalty and audit rates unambiguously increase
tax compliance. Indeed, sufficient -- and draconian -- increases in penalty and
audit rates could substantially eliminate evasion. Of these two instruments,
penalty rates are often seen as the preferred tool, since penalties can be increased
by the simple passage of a law while higher audit rates require the commitment of
additional resources.
There are a number of associated measures that a tax agency can take to increase
enforcement efforts, as suggested by actual administrative experience in
identification, filing, reporting, and collection practices (Bagchi, Bird, and DasGupta, 1995). Source withholding has been universally found to increase tax
compliance, as long as the withholding agent is carefully monitored. Usage of
third-party sources of information, which document things like transactions made
and income received, are important in identifying taxpayers and in ensuring
accurate reporting; receipts from financial institutions are particularly helpful in
this regard. Record keeping within the tax agency is crucial for the most efficient
utilization of information. Computers can aid considerably in this task, especially
in the cross-checking and the analysis of information; as discussed later, the
appropriate analysis of tax return information is a vital aspect of audit selection.
Tax forms themselves may influence compliance (and taxpayer compliance costs)
if they are unduly complicated, although the precise impact of complexity on
compliance is unresolved.
It is interesting that actual enforcement efforts in the United States over the last
several decades have been somewhat inconsistent with these broad policy
suggestions (Dubin, Graetz, and Wilde, 1990a). The percentage of individual tax
returns subject to audit has fallen dramatically in recent years, from roughly 6
percent in the 1960s to only 1 percent in the 1990s. However, at the same time
penalties on detected evasion have increased significantly. Also, government use
of third-party sources of information, such as information returns and CP2000s, to
track income has also increased, despite continued and serious IRS problems with
its computer system. The effects of these policies on compliance are discussed
when empirical evidence is examined.

12

2. Selecting Returns for Audit


A factor intimately connected to enforcement efforts is the manner by which the
tax agency selects tax returns for audit. There are many ways to select individual
returns for a tax audit. The simplest and most widely studied is a random audit
rule, in which each individual faces a fixed, predetermined probability of audit,
regardless of his or her report. Many tax agencies do in fact randomly select some
returns for audit.
However, much audit selection is heavily dependent on the information received
from taxpayers on their tax returns; that is, the government tax authority does not
always select tax returns randomly for audit but instead often uses information
from the returns to determine strategically whom to audit. The probability of
audit is therefore not fixed and random, as assumed in the basic model, but rather
is variable and endogenous, depending in part on the behavior of both the
taxpayer and the tax agency. From this perspective, there are a number of ways in
which the tax agency can utilize the transmission of information from taxpayers in
the strategic selection of tax returns for audit.
Endogeneity arises largely because the tax agency may choose whom to audit
based on the information disclosed by the taxpayer in his or her tax return. The
agency might decide which returns to audit based on a previously determined
audit selection rule. For example, the IRS uses the results of its previous
experience with audited returns to devise a formula (the "Discriminant Index
Function" or DIF score) that determines which current tax returns to audit based
on items reported on the current returns. Selection of returns with a high DIF
score increases the high probability that an audit of the return will generate
additional assessments. In fact, IRS audits based on the DIF score generate
significantly higher amounts of additional assessments than purely random audits,
and roughly one-half of all audited returns are selected with this approach (United
States General Accounting Office, 1976); many other countries follow a similar
practice. Instead, the tax agency might decide which tax returns to audit only after
all returns are filed and without an implied commitment to a previously
determined audit selection rule. The first type of taxpayer-tax agency interaction
is similar to the standard principal-agent model, in which the "principal" (or the
tax agency) must design some rule to affect the behavior of the "agent" (or the
taxpayer). The second type of interaction can be examined using the standard
tools of game theory. Consider the theoretical analysis of each approach.
A. The Principal-Agent Approach. The principal-agent model generates an audit
selection rule typically referred to as a "cutoff rule" (Reinganum and Wilde, 1985;
Border and Sobel, 1987; Mookherjee and Png, 1989; Sanchez and Sobel, 1993).

13

Here the tax agency announces that any taxpayer who reports less than some
minimum, or cutoff, level of income will be audited with certainty; if the taxpayer
reports more than the cutoff level, then he or she will not be audited and will pay
only the reported tax liability. Theoretical analysis of this cutoff rule indicates that
it will raise at least as much revenue as a random audit policy if the cutoff level is
chosen appropriately, so that the cutoff rule weakly dominates the random audit
rule. Many tax agencies seem to follow an audit selection rule similar to a cutoff
rule; that is, within a given audit class, many agencies audit low reports with a
high probability, while high reports are not audited at all.
The optimal cutoff level of income Z can be derived as follows. Consider a
taxpayer with true income I. Income of all taxpayers is assumed to be a random
variable that is independently and identically distributed according to the
cumulative distribution function H(I), where h(I) = H'(I). If I<Z, then the
taxpayer will optimally declare D=I because the taxpayer knows with certainty
that he or she will be audited and subject to a fine on unpaid taxes if I is not
reported. On the other hand, if I Z, then the taxpayer will declare D=Z because
he or she will not be audited if Z is reported and a report higher than Z will merely
increase the tax payment. Denoting the lower and upper bounds on the support of
h () as IL and IU, government revenues R(Z) are therefore R(Z)= IL (tI-c)h(I)dI+
Z

U tZh(I)dI, where c is the constant cost per audit. The first term on the rightZ

hand side of this equation represents the net revenues from those taxpayers whose
reported (and true) incomes are less than the cutoff level and who are audited; the
second term is the revenues from those who report the cutoff level. Assuming
that the tax agency is not subject to a budget constraint and that its goal is revenue
maximization, the tax agency will choose the optimal cutoff level of income Z* to
maximize revenues. This first-order condition can be easily manipulated to give
the condition for the optimal cutoff level as h(Z*)/(1-H(Z*)) = t/c. In the special
case of a uniform distribution, Z* has the simple form Z*=I - c/t, so that Z*
decreases in the audit cost, while Z* increases in the tax rate and the upper bound
on income.
There are several important features of the cutoff rule. One major -- and troubling
-implication is that the tax agency will only audit those taxpayers who report
truthfully (e. g., those below the cutoff level); that is, only honest taxpayers will
be audited, and the agency knows before it selects the returns for audit that these
taxpayers are reporting truthfully. This implication is wildly inconsistent with
actual audit experiences, since a large percentage of audited taxpayers are in fact
shown only in the course of the audit to underreport their income. Another

14

implication is that underreported income increases with income for those above
the cutoff level, since individuals with income above the cutoff point will report at
the cutoff point; the empirical validity of this implication is not known precisely.
A third implication is that audit effort declines the greater is taxpayer reported
income; here, evidence is largely consistent with this prediction.
B. The Game Theory Approach. Audit rules generated by a game-theoretic
analysis are difficult to classify neatly, and depend upon the precise details of the
theoretical model. Typically, these models assume that the taxpayer and the tax
agency interact in a sequential-move game. At the beginning of the game, the
taxpayer learns his or her income and the tax agency learns its audit technology.
In the first stage of the game, the taxpayer decides how much income to report. In
the second stage, the agency decides which returns to audit, based upon
information contained in the return. The equilibrium of the game is one that
specifies a simultaneously determined strategy for both participants: for the
taxpayer the amount or the probability of underreporting, for the tax agency the
probability of audit. This equilibrium is called a Bayesian Nash equilibrium; that
is, both the individual's decision and the agency's decision must represent the best
response to the other's action, so that neither has any incentive to change strategy.
Unlike the principal-agent models, here the tax agency does not precommit to an
audit selection rule, but instead chooses its audit rule as a best-response to the
taxpayer's decision.
A variety of these models have been examined, whose features differ in
assumptions made about the information available to the individual and the
agency, the cost of an audit, the budget of the tax agency, the levels of taxpayer
income, the presence of "honest" taxpayers, and the nature of the tax and penalty
functions (Reinganum and Wilde, 1986; Graetz, Reinganum, and Wilde, 1986;
Beck and Jung, 1989b; Erard and Feinstein, 1994a; Cronshaw and Alm, 1995).
Typically, there are many possible equilibria. Because of this feature, no single
audit rule emerges from these analyses. Still, the general nature of the audit rule
tends to be broadly the same in these models: tax returns with high levels of
reported income will not be audited, while returns with low (and falling) reported
income will be audited with a positive (and increasing) probability.
Some more specific audit selection rules that emerge from these analyses can be
called a "conditional back audit" rule and a "conditional future audit" rule. Each
rule recognizes explicitly the dynamic aspect to compliance; that is, the tax
agency may be able to make use of a taxpayer's history in targeting whom to
audit. The conditional future audit rule says that taxpayers found to be
noncompliant in the past will be audited more frequently in the future
(Landsberger and Meilijson, 1982; Greenberg, 1984). Suppose instead that this

15

same approach is applied to previous periods (the conditional back audit rule).
Here individuals audited and found to be dishonest in the current period face the
certain prospect that the tax agency will go back in time to previous periods'
declarations (Rickard, Russell, and Howroyd, 1982). Both rules have been shown
to be more effective in deterring evasion than a simple random audit rule based
only on current period declarations.
There are some results from these various game-theoretic models that seem quite
realistic and that are consistent with much actual IRS audit experience. For
example, Erard and Feinstein (1994a) show that some audited taxpayers report
fully while others do not, that the level of underreporting increases with taxpayer
income, and that the agency does not know the true income of any taxpayer until
the agency performs an audit. Some, though not all, of the other models generate
similar results.
However, it is surprising that these models all imply that in equilibrium the
agency is indifferent between auditing and not auditing taxpayers. This is a
necessary implication of the solution concept of these models, but it is
nonetheless bothersome. Also, the comparative statics of these models are not
always very intuitive, largely because in a mixed strategy equilibrium (e.g., where
each player chooses a strategy with some probability) a player's strategy is chosen
not simply to maximize his or her own payoff but rather to ensure that the other
player is provided appropriate incentives to choose a mixed strategy. An
illustrative example is provided by Cronshaw and Alm (1995). Suppose that the
penalty on detected evasion increases. From the taxpayer's perspective, the
probability of an audit must then fall to offset the increased penalty, as long as the
taxpayer is to remain willing to follow a mixed strategy. From the tax agency's
perspective, a higher fine makes auditing more attractive, which raises the
probability of audit. However, as just noted, a higher audit probability will make
the taxpayer unwilling to follow a mixed strategy. Instead, the audit probability
must in fact fall, not rise, and the only way for the audit probability to fall is if the
taxpayer cheating probability falls. In short, a higher penalty generates a lower
cheating probability (which seems quite intuitive) but also a lower audit
probability (which may not seem intuitive). Other game-theoretic models often
generate similar counterintuitive results.
3. Changing Social Norms
As noted earlier, social norms (internal and external) are likely to play a major
role in the compliance decision. Evidence from other social sciences suggests that
these norms can be affected by a variety of government institutions and policies.

16

For example, there is much behavioral science evidence that implies that greater
individual participation in the decision process will foster an increased level of
compliance, in part because participation implies some commitment to the
institution and such commitment in turn requires behavior that is consistent with
words and actions. This notion implies that one dimension by which social norms
can be affected is via individual participation in the decision process, say, by
voting. Also, survey evidence suggests that compliance is higher when taxpayers
feel that they have a voice in the way their taxes will be spent. Under such
circumstances, they are likely to feel more inclined to pay their taxes.
Another dimension by which social norms may be affected by government actions
is related to the level of popular support for the government program.
Widespread support tends to legitimize the public sector, and so imposes some
social norm to pay taxes. This support may be obviously revealed through the
voting process. However, the level of support seems likely to affect compliance
even when the choice of the public good is imposed on members of the group.
Consequently, it seems likely that there will be more tax compliance when the
public good provided to a community is popular, even if individuals are unable to
articulate directly their support via voting. Survey evidence is largely consistent
with this hypothesis.
Still another dimension by which social norms can be changed is the
government's commitment to enforcing the tax laws. If the perception becomes
widespread that the government is not willing to detect and penalize evaders, then
such a perception legitimizes tax evasion. The rejection of sanctions sends a
signal to each individual that others do not wish to enforce the tax laws and that
tax evasion is in some sense socially acceptable, and the social norm of
compliance disappears. Such an outcome is common in many countries, such as
the Philippines and Italy where it seems to be accepted that tax evasion is the
norm. The introduction of a tax amnesty may also affect the social norm of
compliance. A tax amnesty gives individuals an opportunity to pay previously
unpaid back taxes without being subject to the penalties that the discovery of
evasion normally brings. Such amnesties may reduce compliance if honest
taxpayers resent the tax forgiveness given to tax cheats (and if individuals believe
that the amnesty may be repeated again).
B. Normative Considerations in Tax Administration
As noted above, the economics-of-crime approach to tax compliance suggests that
compliance can be increased by greater enforcement efforts. However, the
desirability of such a policy is not as obvious as it seems, for several reasons.

17

For one thing, there is a widespread belief that "the punishment should fit the
crime." Imposing draconian penalties on, say, small amounts of noncompliance
would likely violate most peoples' notions of tax equity. A similar notion is that
penalties should be chosen to ensure deterrence on the margin (Stigler, 1970).
Moreover, although higher penalty and audit rates entail benefits from increased
tax revenues and so expanded public services, they also involve costs, both to the
government that must use real resources in its efforts and to the individuals who
suffer a loss in utility from greater enforcement. Finally and relatedly, it may be
inappropriate to increase the tax authority's budget, even if a dollar spent on
enforcement increases collections by more than one dollar: the additional budget
allocation represents a real resource cost, while the additional revenues are simply
a transfer from the private to the public sector (Slemrod and Yitzhaki, 1987; Alm,
1988a). Put differently, a standard benefit-cost criteria for tax administration -increase the enforcement budget until another dollar of administrative
expenditures generates an additional dollar of revenues (Goode, 1981) -- is almost
certainly inappropriate. Instead, the optimal size of a tax administration agency
must involve equality between the marginal costs and benefits of its enforcement
budget, where the benefits should include the added revenues but should also
reflect the impact of greater induced honesty and the loss in individual expected
utility.
In short, it may well be that the best government policy is a pragmatic one, which
recognizes that evasion cannot -- and should not -- be completely eliminated
(Polinsky and Shavell, 1984). Such a policy should include greater enforcement,
but should also emphasize many of the factors noted above: the use of source
withholding, third-party sources of information, efficient record keeping,
computerization, appropriate audit selection, an emphasis on the social
obligations of compliance, the wise use of taxpayer dollars, and so on. Until more
is known about evasion, this strategy may well be the best that is available.

V. Empirical Evidence on Taxpayer Compliance and Tax


Administration
Empirical work has expanded enormously, especially in the last decade. The
obvious difficulty here is the absence of reliable information on individual
reporting behavior. This information is hard to come by, either for the United
States or for other countries: it is difficult to measure something that by its very
nature people want to conceal. This difficulty has not stopped researchers.
However, there are obvious problems with the data that make much of this
empirical work somewhat suspect.

18

For example, most empirical work for the United States has utilized data provided
by the IRS through its Taxpayer Compliance Measurement Program (TCMP). As
noted earlier, these audits yield an IRS estimate of the taxpayer's "true" income so
that a measure of individual tax evasion can be calculated. However, until
recently most researchers have not had access to the individual, micro-level data,
and instead have been forced to use TCMP data aggregated to the 3-digit zip code
level, an aggregate measure likely to comprise disparate elements of
underreporting that reflect very different motivational factors. Also, TCMP data
also have some well-known limitations, notably that the audits do not detect all
underreported (or unreported) income and that the audits cannot distinguish
between honest errors and intentional evasion. Importantly, as suggested by the
theoretical analyses of audit selection, the audit rate is almost certainly
endogenous, so that it cannot be used as an explanatory variable in evasion
equations unless appropriate econometric techniques are applied. Data for other
countries are even more flawed.
To avoid the problems with the TCMP data, some researchers have used
aggregate measures of evasion, such as the amount of income reported or the gap
between income reported on tax returns and income in the national income
accounts. By necessity, these studies focus on the aggregate, not the individual,
response. Other researchers have used surveys of taxpayers, in part to assess
factors such as taxpayer perceptions of the probability of detection, the fairness of
taxation, and the responsiveness of government in the respondent's reporting
decision. Although survey data often have much useful sociodemographic
information, these surveys are also subject to a number of methodological
problems that makes the reliability of their data highly suspect, as discussed
above. State amnesty data have also been used by researchers. Amnesty
participants must declare previously unreported income, so that their amnesty
declaration can be used as a measure of evasion. However, only some individuals
opt to participate in an amnesty, and these participants may not be representative
of all taxpayers.
In its entirety, this work generates a number of conclusions. Some of the more
important are discussed.
A. Estimating the Determinants of Taxpayer Compliance
1. Audit Rates
Estimation results suggest that a higher audit rate leads to more compliance, with
an estimated reported income-audit rate elasticity ranging from 0.1 to 0.2. Witte
and Woodbury (1985) and Dubin and Wilde (1988) use cross section information

19

from the 1969 TCMP aggregated to the 3-digit zip code level, and generally find
that higher audit rates discourage evasion. Dubin, Graetz, and Wilde (1990b) use
pooled time series-cross section information on actual IRS collections at the state
level for the period 1977-1986. They estimate that the decline in federal audit
rates from roughly 2 1/2 percent in 1977 to 1 percent in 1986 reduced tax income
tax collections by $41 billion; of this, $34 billion represented "spillover effects,"
or a reduction in payments independent of revenues generated directly from the
audits and penalties themselves. Kinsey (1992) and Sheffrin and Triest (1992)
examine individual survey data, and also find that compliance increases with a
greater (perceived) probability of audit.
2.Tax Rates
Most empirical evidence suggests that a higher tax rate generally leads to less
compliance, with an estimated underreported income-tax rate elasticity of -0.5 to 3.0. Clotfelter (1983) uses individual TCMP data for 1969 to estimate using Tobit
maximum likelihood methods the determinants of underreported income, for
several different types of taxpayers and several different audit classes. He finds
that noncompliance increases significantly with marginal tax rates. Crane and
Nourzad (1992) examine Michigan tax amnesty data; they also find that evasion is
positively affected by marginal tax rates. Slemrod (1985) finds with individual tax
return information that the proportion of taxpayers who cluster in the top quintile
of a tax reporting bracket tends to rise modestly with marginal tax rates, a result
that suggests that individual compliance falls with higher tax rates. In contrast to
these studies, Feinstein (1991) pools data from the 1982 and 1985 TCMPs in
order to separate more efficiently the effects of marginal tax rates from those of
income, and finds no significant impact of marginal tax rates on noncompliance.
Recall that the theoretical analysis of compliance generally suggests that a higher
tax rate will increase underreported income, so that the bulk of this research is
directly counter to the theory.
3. Income
Higher (true) income leads to higher reported income, with an estimated reported
income-income elasticity between 0 and 1 (Witte and Woodbury, 1985; Dubin,
Graetz, and Wilde, 1990b; Crane and Nourzad, 1992).
4. Tax Practitioners
An increase in tax complexity leads to greater use of a tax practitioner, users of
practitioners have different characteristics than nonusers, and the average level of
noncompliance is higher for returns prepared with paid assistance. Erard (1993)
estimates an endogenous switching model with a micro-level data from the 1979
TCMP, in which the taxpayer jointly chooses whether to seek assistance and

20

whether to comply. He finds that taxpayers who select professional assistance


tend to be those who have more (and more complex) tax forms to complete, are
older, are married, and face higher tax rates; these results are largely consistent
with other work using other data sources (Long and Caudill, 1987; Klepper,
Mazur, and Nagin, 1991; Dubin, Graetz, Udell, and Wilde, 1992). Importantly,
Erard (1993) also finds that the use of an accountant or an attorney significantly
increases cheating.
5. Sociodemographic Variables
Sociodemographic variables are important determinants of behavior. The analysis
of TCMP data suggests that compliance tends to be lower for individuals who are
younger, who are single, and who are self-employed (Clotfelter, 1983; Witte and
Woodbury, 1985; Dubin and Wilde, 1988; Dubin, Graetz, and Wilde, 1990b;
Feinstein, 1991; Beron, Tauchen, and Witte, 1992). The effects of many other
variables, such as race, sex, and education, are uncertain. Survey data generally
have much richer sociodemographic variables. Analyses of these data suggest
similar patterns.
B. Assessing the Effectiveness of Tax Administration
1. Audit Selection
There is strong evidence that audit rates are endogenous. Dubin and Wilde (1988)
and Dubin, Graetz, and Wilde (1990b) use instrumental variables methods to
control for the likely endogeneity of audit rates in their aggregate estimates.
Perhaps more convincingly, Erard and Feinstein (1996) and Alm, Erard, and
Feinstein (1996) combine micro-level TCMP data with similar data from the
Oregon Department of Revenue to estimate the factors that determine audit
selection. They find that the probability of audit selection is determined by a
number of individual tax return items, a result that is consistent with endogenous
audit selection. Surprisingly, Erard (1992) finds with micro-level TCMP data that
the impact of a prior audit on subsequent compliance behavior is statistically
insignificant. There is also evidence that the audit selection criteria of state and
federal enforcement agencies are somewhat different (Alm, Erard, and Feinstein,
1996), which suggests that information sharing could increase agency revenues.
These results are consistent with a revenue gain from strategic audit selection. In
fact, Alm, Erard, and Feinstein (1996) and Erard and Feinstein (1996) estimate
that the "shadow value" of additional IRS audit resources ranges from 1 to 8,
depending on the audit class; that is, providing $1 to the IRS generates
somewhere between $1 and $8 in additional tax collections. The shadow value of
additional state audit resources is lower, from 1 to 2.

21

2. Detection
There is convincing evidence that it is important in empirical work to control for
the inability of an audit to detect all tax evasion. Feinstein (1990, 1991) and
Erard (1997) use what they term "detection controlled estimation" methods to
estimate with micro-level TCMP data a two-stage system: first, is a taxpayer
noncompliant and, second (conditional on noncompliance), is noncompliance
detected? They find that the impact on noncompliance of numerous variables is
significantly altered relative to estimation methods that do not control for
detection. Their results also suggest that IRS auditors differ significantly in their
ability to detect noncompliance, with a detection rate of roughly one-half of true
taxpayer evasion.
3. Tax Amnesties
Most evidence shows that a tax amnesty generates relatively small amounts of
additional tax revenues, and also seems to have relatively small effects on postamnesty compliance (Mikesell, 1986; Fisher, Goddeeris, and Young, 1989; Alm
and Beck, 1992).
4. Social Norms
There is little empirical work on the role of social norms. Pommerehne and
Weck-Hannemann (1989) find that tax compliance in Swiss cantons is affected by
the process by which collective decisions are made and by individuals' attitudes
about the fairness with which they believe they are treated by government
officials.
C. Summary
These studies have expanded enormously our understanding of the factors that
affect taxpayer compliance. Given the underlying data and econometric problems,
this empirical work needs to be treated cautiously. Further, the various estimated
responses vary greatly across the different studies, both in magnitude and even
sometimes in sign. Still, these results indicate that taxpayer compliance decisions
are affected in largely predictable ways by the fiscal system in which they
operate. The results also suggest that the enforcement agency can increase
compliance by changing its enforcement strategy but that there are limits to
strategies based only on greater enforcement.

22

VI. Experimental Evidence on Taxpayer Compliance and


Tax Administration
Difficulties with the existing theoretical and empirical literatures have led to the
use of experimental economics as an additional approach to compliance research.
The use of laboratory experiments in economics began in the early 1960s with
work on resource allocation under alternative forms of market organization.
Growth in its applications came with the establishment of a well-defined
framework for experimental work by Smith (1976, 1982), and laboratory methods
are now widely accepted as a methodological approach in the analysis of theory
and policy.
Laboratory experiments seem particularly well-suited for the study of some
aspects of the taxpayer reporting decision. Unlike theoretical work, experiments
are not as constrained by the same degree of simplification required in analytical
studies of reporting, which allows the impact of numerous factors not amenable to
theoretical work to be examined precisely and unambiguously. Unlike empirical
work, experiments generate data under different settings in which there is control
over extraneous influences. As discussed later, there are some obvious limitations
of experimental methods. However, given the weaknesses of other methodologies,
there are compelling reasons for the use of experiments. In fact, experimental
work has examined a remarkedly rich range of factors in the individual
compliance decision, factors that to date have not proven amenable to either
theoretical or empirical analyses.
A. Creating a Microeconomic System in the Laboratory
Experimental economics involves the creation of a real microeconomic system in
the laboratory, one that parallels the naturally occurring world that is the subject
of investigation. The essence of such a system is control over the environment,
the institutions, and the preferences that subjects face. Of these, control over
preferences is particularly crucial. As emphasized by Smith (1976), "[s]uch
control can be achieved by using a reward structure to induce prescribed monetary
value on actions."
Smith (1982) identifies several (sufficient) conditions that must be satisfied for
control over preferences to be established:
Nonsatiation: subjects must prefer more to less
Saliency: the rewards received by subjects must be related to their
decisions, so that subjects recognize that their actions affect their
outcomes

23

Reward Dominance: rewards must be large enough to offset any


subjective costs or benefits that subjects place on participation in
the experiment, which requires the payment to subjects of an
amount comparable to what they could earn outside the laboratory
Privacy: each subject must know only his or her own payoffs so
that they do not receive any subjective value from the payoffs of
other subjects.

Several other procedures should also be followed in experimental economics. For


example, the experiment should be administered in a uniform and consistent
manner to allow replicability. The experiment should not be excessively long or
complicated, since subjects may become bored or confused. Subjects must
believe that the procedures described to them are the procedures actually
followed. The instructions provided to subjects should be understandable, should
avoid the use of examples that lead subjects to anchor on certain choices that are
the subject of the experiment, and should be phrased in "neutral" rather than
"loaded" terms, to mask the context of the experiment and to avoid direct
reference to the real-world phenomena under investigation. Neutrality increases
the experimenter's control over subject preferences and avoids leading subjects to
invoke different "mental scripts," which may enable them to fill in (potentially)
missing information in the instructions but which also may unpredictably
influence their choices. It is sometimes claimed that the use of neutral
instructions limits the ability to generalize from the experimental to the naturally
occurring setting. In fact, however, it is not possible to generalize beyond the
laboratory unless one uses neutral instructions, since the experimenter cannot
control (or induce) the values that subjects associate with loaded terms.
Applying this framework to the study of tax compliance is straightforward, and
the basic design of most compliance experiments is similar. Human subjects in a
controlled laboratory are told that they should feel free to make as much income
as possible. At the beginning of each round of the experiment, each subject is
given income and must decide how much income to report. Taxes are paid at
some rate on all reported, but not on underreported, income. However,
underreporting is discovered with some probability, and the subject must then pay
a fine on unpaid taxes. This process is repeated for a given number of rounds. At
the completion of the experiment, each subject is paid an amount (the
accumulated earnings) that depends on his or her performance during the
experiment. Into this microeconomic system, various policy changes can be
introduced, such as changes in audit probabilities or audit rules, in penalty rates,
in tax rates, in public good provision, in institutions that affect tax equity or social

24

norms, and in other relevant policies or institutions. Results from experimental


analyses of tax compliance are discussed next.
B. Experimental Results
The first experimental study of tax compliance was conducted by Friedland,
Maital, and Rutenberg (1978), who examined subject responses to changes in tax,
penalty, and audit rates. Numerous experimental analyses have followed. These
experimental studies suggest several conclusions, many of which parallel those of
the empirical studies but several of which are quite different.
1. Experimental Results on Taxpayer Compliance
Audit Rates. Nearly all studies have found that a higher (random) audit rate leads
to more compliance, with an estimated reported income-audit rate elasticity
ranging from 0.1 to 0.2 (Friedland, Maital, and Rutenberg, 1978; Beck, Davis,
and Jung, 1991; Alm, Jackson, and McKee, 1992a, 1992b; Alm, Cronshaw, and
McKee, 1993). However, Alm, McClelland, and Schulze (1992) find that this
impact appears to be small and nonlinear, so that the deterrent effect of a higher
audit rate eventually diminishes. They also find that many subjects appear to
substantially overweight the probability of an audit, so that there is far more
compliance than is predicted by expected utility theory; indeed, this result is
nearly universal across all experimental designs. Also, Spicer and Hero (1985)
and Webley (1987) find that individuals who have been audited report more
income post-audit than individuals who have not been audited.
Penalty Rates. In the relevant range of penalty rate changes, compliance
increases but only slightly with increases in the fine rate on unpaid taxes
(Friedland, Maital, and Rutenberg, 1978; Beck, Davis, and Jung, 1991; Alm,
Jackson, and McKee, 1992a, 1992b; Alm, McClelland, and Schulze, 1992). A
higher fine rate leads to marginally more compliance, with an estimated reported
income-fine rate elasticity less than 0.1.
Tax Rates. A higher marginal tax rate often leads to less compliance, with an
estimated underreported income-tax rate elasticity of roughly -0.5 (Friedland,
Maital, and Rutenberg, 1978; Alm, Jackson, and McKee, 1992b). However,
Beck, Davis, and Jung (1991) and Alm, Sanchez, and de Juan (1995) find that
reported income rises with higher tax rates.
Income. Higher (true) income leads to higher reported income, with an estimated
reported income-income elasticity of roughly 3/4 (Alm, Jackson, and McKee,
1992b).

25

Public Good Provision. The presence of a public good financed by voluntary tax
payments increases subject tax compliance in a nonlinear manner (Becker,
Buchner, and Sleeking, 1987; Alm, McClelland, and Schulze, 1992; Alm,
Jackson, and McKee, 1992a, 1992b).
Positive Rewards. Like group rewards, individual rewards can provide a
significant positive inducement for greater compliance (Alm, Jackson, and
McKee, 1992b). When audited and fully compliant taxpayers are eligible for a
lottery whose expected value equals the average subject per round income, or
when audited and fully compliant taxpayers are given an immediate reward of
comparable value, compliance is significantly higher than in other experiments in
which rewards were not given but the fine and audit rates were adjusted to keep
the expected value of the evasion gamble constant.
Tax Complexity and Uncertainty. The presence of taxpayer uncertainty about
taxable income and the various fiscal parameters has an ambiguous impact on
subject compliance (Spicer and Thomas, 1982; Friedland, 1982; Beck, Davis, and
Jung, 1991; Alm, Jackson, and McKee, 1992c). For example, Beck, Davis, and
Jung (1991) generally find that greater taxpayer uncertainty about true taxable
income leads the taxpayer to report higher taxable income, although the effect
depends upon the level of tax, penalty, and audit rates, as well as upon the
taxpayer's attitude toward risk. Alm, Jackson, and McKee (1992c) find instead
that the impact of greater taxpayer uncertainty about the tax, penalty, and audit
rate depends upon the presence of absence of a public good, financed by
individuals' tax payments. When subjects receive something for their tax
payments, greater uncertainty always lowers compliance; when there is no public
good greater uncertainty raises compliance.
Sociodemographic Variables.
The impacts of only a small number of
sociodemographic variables have been examined in the laboratory. Friedland,
Maital, and Rutenberg (1978) and Baldry (1987) find that older individuals are
more compliant than younger ones. Baldry (1987) also finds that women evade
less than men.
2. Experimental Results on Tax Administration
Audit Selection. Audit selection methods that utilize information provided on the
tax return are far more effective in generating tax compliance than purely random
selection methods, even when the random audit rate is 20, 30, or 50 percent
(Collins and Plumlee, 1991; Alm, Cronshaw, and McKee, 1993). A cutoff rule, in
which a taxpayer who reports less than some cutoff level of income is audited
with certainty, is the most effective in increasing compliance, although it requires

26

a large number of audits. Another rule requires that an audited individual will
face some back audits if found to be noncompliant in the current period (a
conditional back audit rule); this rule is also able to increase compliance
significantly, and the audit rate is far lower than the cutoff rule. Auditing an
individual found to be noncompliant in the current period with certainty for a
number of future periods (a conditional future audit rule) appears to be the least
effective of the endogenous rules, although compliance still exceeds that under all
random audit rules.
Fiscal Institutions. Compliance is affected by the uses of tax revenues and the
decision process by which these uses are chosen. Alm, Jackson, and McKee
(1993) find that subjects pay more in taxes when they choose via voting the use of
their taxes than when the identical use is imposed upon them, that compliance is
somewhat greater when the vote is decisive than when the vote is close, and that
compliance is significantly and dramatically lowered by the imposition of an
unpopular program. Surprisingly, Martinez-Vazquez, Harwood, and Larkins
(1992) find that compliance does not seem to be significantly affected by
withholding systems, at least beyond the obvious channel that withholding
reduces the opportunities for evasion.
Subjects who find themselves
unexpectedly underwithheld do not behave much differently than subjects who
are correctly anticipate the underwithholding, and there is little difference in
compliance between taxpayers who are under- or overwithheld.
Tax Amnesties. Alm, McKee, and Beck (1990) find that a tax amnesty lowers
post-amnesty tax compliance, largely because the introduction of an amnesty
increases taxpayers' expectations of another, future amnesty. However, they also
find that a "welldesigned" amnesty, or one in which post-amnesty enforcement
efforts increase, can overcome and even reverse the typical post-amnesty decline
in compliance. In fact, their results show that compliance is higher when an
amnesty is followed by greater enforcement than when enforcement alone
increases by an equal amount.
Social Norms. The social norm of compliance can be affected by the institutions
that face individuals, by individuals' attitudes toward these institutions, and by
individual participation in the selection of those institutions. Webley, et al.
(1991) find that individuals who have a negative attitude toward government
comply less in taxes as a result. Alm, Jackson, and McKee (1993) demonstrate
that government can affect compliance by ensuring that individuals have a say in
the decision process and by spending taxes in ways consistent with citizen
preferences. Also, Alm, McClelland, and Schulze (1997) find that compliance is
decreased, often collapsing virtually to zero, when there is a social expression via

27

group selection of the fiscal regime of a willingness to tolerate tax evasion. It is


as if the group decision ratifies each individual's decision to evade his or her
taxes, and post-vote individual noncompliance is in some sense now justified by
the revealed actions of others. However, they also find that compliance can be
increased when there is a social expression of an unwillingness to tolerate tax
evasion. Spicer and Becker (1980) find that perceptions of "fiscal inequity" affect
compliance: compliance is lower (higher) among subjects who are told that their
tax rate is higher (lower) than that of others. Alm, Sanchez, and de Juan (1995)
demonstrate that there are significant differences in the compliance behavior of
student subjects in similar experiments but in different countries (Spain versus the
United States), thereby suggesting that societal attitudes toward compliance exert
a measurable impact on tax compliance.
C. Limitations of Experimental Economics
There are sound reasons for caution in interpreting and generalizing experimental
results. Some early compliance experiments did not follow some now widely
accepted procedures of the experimental paradigm, such as the use of repeated
experiments and neutral instructions. Much early work also lacked realism
because values of the various policy parameters did not approximate real-world
values.
Although more recent experimental research has generally addressed these
problems, some concerns remain, some of which are more real than others. A
common criticism of experimental economics is that the student subjects typically
used may not be representative of taxpayers. However, there is now much
evidence that the experimental responses of students are no different than the
responses of other subject pools (Plott 1987); there is also no reason to believe
that the cognitive processes of students are different from those of "real" people.
Another common criticism is that it is not possible to control for many relevant
factors in the laboratory. However, if one cannot control for such factors in the
laboratory where the experimenter establishes the institutions, the rules, and the
reward structure, then one cannot hope to control for these factors in the
"naturally occurring world."
Of more legitimate concern, the results may well be sensitive to the specific
experimental design, so that replication is crucial. It is also possible that subjects
may modify their behavior simply because they know that they are participating
in an experiment. Most importantly, there is a certain artificiality in a laboratory
setting. A decision to report $2 in an experiment is clearly different from a
decision to report actual income on an annual tax return, even if the laboratory
incentives are salient. In particular, the laboratory setting cannot capture a

28

catastrophic loss such as jail, and it cannot capture the social stigma that some
surveys suggest is an important factor in taxpayer reporting.
In short, one must use the results from laboratory experiments with some care.
However, such use depends largely upon the purpose of the experiment.
According to Roth (1987), experiments can be classified into three broad
categories that depend upon the dialogue in which they are meant to participate.
"Speaking to Theorists" includes those experiments designed to test wellarticulated theories. "Searching for Facts" involves experiments that examine the
effects of variables about which existing theory has little to say. "Whispering in
the Ears of Princes" identifies those experiments motivated by specific policy
issues. To date, most experiments on taxpayer reporting have fallen into the first
two categories. Although this now seems to be changing somewhat, there
remains a natural skepticism among policy makers about the ability of
experimental analyses to illuminate some aspects of tax compliance.

VII. Conclusions
It is, I hope, apparent that enormous amounts have been learned about tax
compliance and administration in the last 25 years. We have more and better
estimates of the extent of tax evasion. We have a much deeper understanding of
the factors that motivate individuals to cheat on their income taxes. We have a
better comprehension of the tradeoffs that face government in the design and
enforcement of tax laws. We know much more about the magnitude of individual
behavioral responses, to changes in audit rates, tax rates, and many other policies,
both from empirical and experimental work. However, it should also be apparent
that enormous amounts remain to be learned about tax compliance and
administration. Let me conclude by discussing ten particular areas that I believe
require more attention.
First, nearly all theoretical analyses of taxpayer behavior, and the empirical
analyses that follow from them, are based upon expected utility theory. As
emphasized earlier, there are significant limitations in the ability of this theory to
explain major aspects of individual compliance behavior; in fact, there is growing
dissatisfaction with this approach in the analysis of many other individual choices
under uncertainty. Clearly, it is important to apply other theories of behavior to
tax evasion, theories that allow the introduction into the compliance decision of
numerous factors beyond simply enforcement: overweighting of low probabilities,
differential responses to gains versus losses, the presence of social norms and
moral sentiments, notions of fairness, satisfaction with government programs, and
so on. Approaches that seem particularly useful for detailed investigation include

29

prospect theory (Kahneman and Tversky, 1979) and generalized expected utility
theory (Quiggin, 1993).
Second, the intertemporal aspects of tax compliance have received little attention.
Individuals pay taxes over a number of years, and they undoubtedly recognize
that their decisions today affect their chances of audit, both now and in the future.
With the exception of work by Landsberger and Meilijson (1982), Rickard,
Russell, and Howroyd (1982), and Greenberg (1984), the ways in which
individuals and the tax agency respond and interact to such intertemporal
incentives have largely been ignored. Also, an individual's willingness to comply
can change over time in response to such events as the establishment of a tax
amnesty or the erosion of other people's willingness to pay taxes. Evolutionary
game theory seems a promising avenue for the investigation of these dynamics of
tax evasion (Friedman, 1991).
Third, the basic model of individual choice has centered on the individual's choice
of the total amount of income to declare. However, this decision is not really a
single choice, but actually consists of a number of other decisions on the reporting
of income types, personal exemptions, deductions, credits, tax schedules, and the
like, and the chances of detection of fraudulent claims certainly vary across these
many choices. More analysis of the multidimensional nature of reporting
decisions is needed. In this regard, the work of Klepper and Nagin (1989a) and
Martinez-Vazquez and Rider (1995) represent important recent contributions.
Fourth, nearly all analyses of tax evasion in the United State have examined the
factors that lead individuals who file tax returns to underreport their incomes.
However, there is now some evidence that nonfiling of tax returns is also a
serious problem, especially for some occupations (Erard and Ho, 1995). More
analysis of nonfilers is an important area for future research.
Fifth, most analyses of tax evasion have focused on individual compliance with
the income tax. Clearly, however, compliance is an important issue for all other
taxes, especially the firm decision to comply with the corporate income tax and
with sales and excise taxes. For example, Rice (1992) and Greene (1997) have
examined some aspects of corporate tax compliance, and Murray (1995) has
analyzed firm compliance with sales taxes. More such work is required.
Sixth, although individuals may cheat on the taxes they pay to government, they
also may cheat on the benefits they receive from government, especially for
welfare transfers and tax credits. Compliance with government transfer and

30

benefit programs has only recently begun to receive much needed attention
(Scholz, 1994; Joulfaian and Rider, 1996).
Seventh, there have been relatively few systematic analyses of compliance in
other countries, despite the fact that noncompliance is almost certainly more of a
problem in, say, developing countries than in the United States. Data availability
is obviously an important reason for this omission. However, recent work by Alm,
Bahl, and Murray (1990, 1993) for Jamaica, as part of a comprehensive tax
reform, indicates the potential for such work. Working with the full cooperation
of the Government of Jamaica, they were able to construct data sets that allowed
them to estimate the responses both of employees and of the self-employed to
changes in tax, penalty, and audit rates in the individual income tax; they were
also able to estimate the criteria by which self-employed income tax returns were
selected for audit. Their results demonstrate that Jamaican taxpayers respond in
significant ways to the various incentives and that audit selection is endogenous.
Eighth, the analysis of strategic tax agency audit selection has considerably
expanded our understanding of agency behavior. However, as emphasized earlier,
both the principal-agent and the game-theoretic approaches have major
weaknesses, especially in their implications for optimal audit selection rules.
Further analysis of strategic behavior is sorely needed. In particular, the work by
Erard and Feinstein (1994a) illustrates the vital role of assumptions about the
prevalence of honest and dishonest taxpayers. Also, the ways in which federal
and state tax agencies can share information in their selection and audit of returns
merits further work (Alm, Erard, and Feinstein, 1996). Experimental analysis of
these models is a promising approach.
Ninth, although there is budgetary information on the administrative costs of
various taxes, there has been relatively little systematic analysis of the
determinants of these administrative costs and the effects of these determinants on
the quality of tax administration. Put differently, the nature of the tax agency
production and cost functions are largely unknown. It seems likely that there are
significant fixed costs when a new tax is imposed or when the features of an
existing tax are changed, that these policies will involve a stepwise increase in
costs, and that there will be economies of scale in tax administration. These costs
must clearly affect the ability of the tax administration to enforce compliance with
these taxes. As shown by Allers (1994), Sandford (1995), and Hunter and Nelson
(1996), there is much scope for systematic empirical analyses of these issues.
Especially important here is the estimation of the shadow value of an additional
dollar of audit resources.

31

Lastly, recent work by Gould (1996) emphasizes that it is grossly misleading to


represent a complex system by a single, so-called representative agent, who
behaves in some average or typical way. Instead, most systems have incredible
variety --or a "full house" of individual behaviors -- and the proper understanding
of any system requires recognition of this basic fact. Indeed, Gould (1996) argues
that the way in which a system changes over time is attributable largely to
changes in the amount of variation within the system, rather than to changes in
some largely meaningless "average" behavior across its individual members.
This lesson is especially apt for tax compliance. People exhibit a remarkable
diversity in their behavior. There are individuals who always cheat and those who
always comply, some who behave as if they maximize the expected utility of the
tax evasion gamble, others who seem to overweight low probabilities, individuals
who respond in different ways to changes in their tax burden, some who are at
times cooperative and at other times free-riders, and many who seem to be guided
by such things as social norms, moral sentiments, and tax equity. These findings
suggest that it is unlikely that a single unifying theory of tax compliance can ever
be devised, one that incorporates the incredible variation in individual behavior
exhibited by the many analyses of taxpayer compliance, one that explains the
behavior of all individuals at all times, or even one that explains the actions of the
same person at all times. Perhaps our research should still be devoted to the
pursuit of such a holy grail. More importantly, however, our research needs to
recognize that a "theory" of taxpayer compliance must really consist of a "full
house" of theories, each explaining the behavior of different individuals at
different times. Any tax administration must also recognize that it must address
this "full house" of behaviors in devising policies to ensure compliance.
Consequently, a government compliance strategy based only on detection and
punishment may well be a reasonable starting point for tax administration but not
a good ending point. Instead, what is needed is a multi-faceted approach that
emphasizes enforcement, but that also emphasizes the much broader range of
actual motivations that explain why people pay taxes.

32

VIII. References
Allers, M. (1994). Administrative and Compliance Costs of Taxation and Public
Transfers in the Netherlands, Woltersgroep Groningen, The Netherlands.
Allingham, M. G., and Sandmo, A. (1972). Income tax evasion: A theoretical
analysis, Journal of Public Economics, 1: 323-338.
Alm, J. (1988a). Compliance costs and the tax avoidance-tax evasion decision,
Public Finance Quarterly, 16: 31-66.
Alm, J. (1988b). Uncertain tax policies, individual behavior, and welfare, The
American Economic Review, 78: 237-245.
Alm, J., Bahl, R., and Murray, M. N. (1990). Tax structure and tax compliance,
The Review of Economics and Statistics, 72: 603-613.
Alm, J., Bahl, R., and Murray, M. N. (1993). Audit selection and income tax
underreporting in the tax compliance game, Journal of Development Economics,
42: 1-33.
Alm, J., and Beck, W. (1993). Tax amnesties and compliance in the long run: A
time series analysis, National Tax Journal, 46: 53-60.
Alm, J., Cronshaw, M. B., and McKee, M. (1993). Tax compliance with
endogenous audit selection rules, Kyklos, 46: 27-45.
Alm, J., Erard, B., and Feinstein, J. S. (1996). The relationship between state and
federal tax audits, in Empirical Foundations of Household Taxation (M. Feldstein
and J. M. Poterba, eds.), University of Chicago Press, Chicago, IL, 235-273.
Alm, J., Jackson, B. R., and McKee, M. (1992a). Deterrence and beyond: Toward
a kinder, gentler IRS, in Why People Pay Taxes (J. Slemrod, ed.), University of
Michigan Press, Ann Arbor, MI, 311-329.
Alm, J., Jackson, B. R., and McKee, M. (1992b). Estimating the determinants of
taxpayer compliance with experimental data, National Tax Journal, 45: 107-114.
Alm, J., Jackson, B. R., and McKee, M. (1992c). Institutional uncertainty and
taxpayer compliance, The American Economic Review, 82: 1018-1026.

33

Alm, J., Jackson, B. R., and McKee, M. (1993). Fiscal exchange, collective
decision institutions, and tax compliance, Journal of Economic Behavior and
Organization, 22: 285-303.
Alm, J., McClelland, G. H., and Schulze, W. D. (1992). Why do people pay
taxes? Journal of Public Economics, 48: 21-38.
Alm, J., McClelland, G. H., and Schulze, W. D. (1997). Changing the social norm
of tax compliance by voting, Working Paper, Department of Economics,
University of Colorado at Boulder.
Alm, J., McKee, M., and Beck, W. (1990). Amazing grace: Tax amnesties and tax
compliance, National Tax Journal, 43: 23-37.
Alm, J., Sanchez, I., and de Juan, A. (1995). Economic and noneconomic factors
in tax compliance, Kyklos, 48: 3-18.
Andreoni, J. (1991). On the desirability of a permanent tax amnesty, Journal of
Public Economics, 45: 143-159.
Andreoni, J., Erard, B., and Feinstein, J. (1996). Tax compliance, Working Paper,
Department of Economics, University of Wisconsin-Madison.
Axelrod, R. (1984). The Evolution of Cooperation, Basic Books, Inc., New York,
NY.
Bagchi, A., Bird, R., and Das-Gupta, A. (1995). An economic approach to tax
administration reform, Discussion Paper No. 3, International Centre for Tax
Studies, University of Toronto.
Bagnoli, and McKee, M. (1991). Voluntary contribution games: Efficient private
provision of public goods, Economic Inquiry, 29: 351-366.
Baldry, J. C. (1987). Income tax evasion and the tax schedule: Some experimental
results, Public Finance/Finances Publiques, 42: 357-383.
Beck, P. J., Davis, J. S., and Jung, W. (1991). Experimental evidence on taxpayer
reporting behavior, The Accounting Review, 66: 535-558.
Beck, P. J., and Jung, W. (1989a). An economic model of taxpayer compliance
under uncertainty, Journal of Accounting and Public Policy, 8: 1-27.

34

Beck, P. J., and Jung, W. (1989b). Taxpayers' reporting decisions and auditing
under information asymmetry, The Accounting Review, 66: 468-487.
Becker, G. S. (1968). Crime and punishment - An economic approach, The
Journal of Political Economy, 76: 169-217.
Becker, W., Buchner, H., and Sleeking, S. (1987). The impact of public transfer
expenditures on tax evasion: An experimental approach, Journal of Public
Economics, 34: 243-252.
Beron, K., Tauchen, H. V., and Witte, A. D. (1992). The effect of audits and
socioeconomic variables on compliance, in Why People Pay Taxes (J. Slemrod,
ed.), University of Michigan Press, Ann Arbor, MI, 67-89.
Bird, R. M. (1989). The administrative dimension of tax reform in developing
countries, in Tax Reform in Developing Countries (M. Gillis, ed.), Duke
University Press, Durham, NC, 315-346.
Bird, R. M., and Casangera de Jantscher, M., eds. (1992). Improving Tax
Administration in Developing Countries, International Monetary Fund,
Washington, D.C.
Border, K., and Sobel, J. (1987). Samurai accountant: A theory of audit and
plunder, The Review of Economic Studies, 54: 525-540.
Clotfelter, C. T. (1983). Tax evasion and tax rates: An analysis of individual
returns, The Review of Economics and Statistics, 65: 363-373.
Collins, J. H., and Plumlee, R. D. (1991). The taxpayer's labor and reporting
decisions: The effect of audit schemes, The Accounting Review, 66: 559-576.
Cowell, F. A. (1981). Taxation and labour supply with risky activities,
Economica, 48: 365379.
Cowell, F. A. (1990). Cheating the Government: The Economics of Tax Evasion,
The MIT Press, Cambridge, MA.
Cowell, F. A., and Gordon, J. P. F. (1988). Unwillingness to pay: Tax evasion and
public good provision, Journal of Public Economics, 36: 305-321.

35

Crane, S. E., and Nourzad, F. (1992). Analyzing income tax evasion using
amnesty data with self-selection correction: The case of the Michigan tax amnesty
program, in Why People Pay Taxes (J. Slemrod, ed.), University of Michigan
Press, Ann Arbor, MI, 167-189.
Crane, S. E., and Nourzad, F. (1994). An empirical analysis of factors that
distinguish those who evade on their tax return from those who choose not to file
a return, Public Finance/Finances Publiques, 49 (Supplement): 106-116.
Cremer, H., and Gahvari, F. (1995). Tax evasion and the optimal general income
tax, Journal of Public Economics, 60: 235-249.
Cremer, H., Marchand, M., and Pestieau, P. (1990). Evading, auditing, and taxing:
The equity-compliance tradeoff, Journal of Public Economics, 43: 67-92.
Cronshaw, M. B., and Alm, J. (1995). Tax compliance with two-sided
uncertainty, Public Finance Quarterly, 23: 139-166.
Cross, R. B., and Shaw, G. K. (1982). On the economics of tax aversion, Public
Finance/Finances Publiques, 37: 36-47.
Davis, D. D., and Holt, C. A. (1993). Experimental Economics, Princeton
University Press, Princeton, NJ.
de Juan, A., Lasheras, M. J., and Mayo, R. (1994). Voluntary compliance and
behavior of Spanish taxpayers, Public Finance/Finances Publiques, 49
(Supplement): 90-105.
Dubin, J. A., Graetz, M. J., and Wilde, L. L. (1990). The effect of audit rates on
the federal individual income tax, 1977-1986, National Tax Journal, 43: 395-409.
Dubin, J. A., Graetz, M. J., Udell, M. A., and Wilde, L. L. (1992). The demand
for tax preparation services, The Review of Economics and Statistics, 74: 75-82.
Dubin, J. A., Graetz, M. J., and Wilde, L. L (1990). The changing face of tax
enforcement, 1978-1988, Tax Lawyer, 43: 893-913.
Dubin, J. A., and Wilde, L. L (1988). An empirical analysis of federal income tax
auditing and compliance, National Tax Journal, 41: 61-74.

36

Elffers, H. (1991). Income Tax Evasion: Theory and Measurement, Kluwer,


Deventer.
Elffers, H., Weigel, R. H., and Hessing, D. J. (1987). The consequences of
different strategies for measuring tax evasion behavior, Journal of Economic
Psychology, 8: 311-337.
Elster, J. (1989). Social norms and economic theory, Journal of Economic
Perspectives, 3: 99-117.
Erard, B. (1992). The influence of tax audits on reporting behavior, in Why
People Pay Taxes
(J. Slemrod, ed.), University of Michigan Press, Ann Arbor, MI, 95-114.
Erard, B. (1993). Taxation with representation: An analysis of the role of tax
practitioners in tax compliance, Journal of Public Economics, 52: 163-197.
Erard, B. (1997). Self-selection with measurement errors: A microeconometric
analysis of the decision to seek tax assistance and its implications for tax
compliance, Journal of Econometrics, forthcoming.
Erard, B., and Feinstein, J. S. (1994a). Honesty and evasion in the tax compliance
game, RAND Journal of Economics, 25: 1-19.
Erard, B., and Feinstein, J. S. (1994b). The role of moral sentiments and audit
perceptions in tax compliance, Public Finance/Finances Publiques, 49
(Supplement): 70-89.
Erard, B., and Feinstein, J. S. (1996). Reporting behavior and audit selection
decision, Working Paper, Yale School of Management.
Erard, B., and Ho, C.-C. (1995). Searching for ghosts: Who are the nonfilers and
how much tax do they owe? Working Paper, Department of Economics, Carleton
University.
Falkinger, J., and Walther, H. (1991). Rewards versus penalties: On a new policy
against tax evasion, Public Finance Quarterly, 19: 67-79.
Feige, E. L. (1989). The Underground Economies: Tax Evasion and Information
Distortion, Cambridge University Press, Cambridge, England.

37

Feinstein, J. S. (1990). Detection controlled estimation, Journal of Law and


Economics, 33: 233-276.
Feinstein, J. S. (1991). An econometric analysis of income tax evasion and its
detection, RAND Journal of Economics, 22: 14-35.
Fisher, R. C., Goddeeris, J. H., and Young, J. C. (1989). Participation in tax
amnesties: The individual income tax, National Tax Journal, 42: 15-27.
Friedland, N. (1982). A note on tax evasion as a function of the quality of
information about the credibility of threatened fines: Some preliminary research,
Journal of Applied Social Psychology, 12: 54-59.
Friedland, N., Maital, S., and Rutenberg, A. (1978). A simulation study of income
tax evasion, Journal of Public Economics, 10: 107-116.
Friedman, D. (1991). Evolutionary games in economics, Econometrica, 59: 637666.
Goode, R. (1981). Some economic aspects of tax administration, IMF Staff
Papers, 28: 249274.
Gordon, J. P. F. (1989). Individual morality and reputation costs as deterrents to
tax evasion, European Economic Review, 33: 797-805.
Gould, S. J. (1996). Full House, Harmony Books, New York, NY.
Graetz, M. J., and Wilde, L. L. (1985). The economics of tax compliance: Fact
and fantasy, National Tax Journal, 38: 355-363.
Graetz, M. J., Reinganum, J. F., and Wilde, L. L. (1986). The tax compliance
game: Toward an interactive theory of tax enforcement, Journal of Law,
Economics, and Organization, 2: 1-32.
Greene, P. A. (1997). Corporate Tax Compliance and IRS Audit Strategy, Ph.D.
Dissertation, University of Colorado at Boulder, Boulder, CO.
Greenberg, J. (1984). Avoiding tax avoidance: A (repeated) game-theoretic
approach, Journal of Economic Theory, 32: 1-13.
Harris and Associates, Inc. (1988). 1987 Taxpayer Opinion Survey, Conducted
for the

38

U.S. Internal Revenue Service, Internal Revenue Service Document 7292,


Washington, D.C.
Herschel, F. J. (1978). Tax evasion and its measurement in developing countries,
Public Finance/Finances Publiques, 33: 232-268.
Hunter, W. J., and Nelson, M. A. (1996). An IRS production function, National
Tax Journal,
49: 105-115.
Internal Revenue Service (1978). A Dictionary of Compliance Factors, United
States Department of the Treasury, Washington, D.C.
Internal Revenue Service (1988). Income Tax Compliance Research: Gross Tax
Gap Estimates and Projections for 1973-1992, Publication 7285, United States
Department of the Treasury, Washington, D.C.
Internal Revenue Service (1990). Income Tax Compliance Research: Net Tax
Gap and Remittance Gap Estimates (Supplement to Publication 7285),
Publication 1415, United States Department of the Treasury, Washington, D.C.
Internal Revenue Service (1995). Internal Revenue Service 1993-94 Data Book,
Publication 55B, United States Government Printing Office, Washington, D.C.
Internal Revenue Service (1996). Federal Tax Compliance Research: Individual
Income Tax Gap Estimates for 1985, 1988, and 1992, Publication 1415, United
States Department of the Treasury, Washington, D.C.
Joulfaian, D., and Rider, M. (1996). Tax evasion in the presence of negative
income tax rates, National Tax Journal, 49: 553-570.
Kahneman, D., and Tversky, A. (1979). Prospect theory: An analysis of decision
under risk, Econometrica, 47: 263-291.
Kaplow, L. (1990). Optimal taxation with costly enforcement and evasion,
Journal of Public Economics, 43: 221-236.
Kesselman, J. (1989). Income tax evasion: An intersectoral analysis, Journal of
Public Economics, 38: 137-182.

39

Kinsey, K. A. (1992). Deterrence and alienation effects of IRS enforcement: An


analysis of survey data, in Why People Pay Taxes (J. Slemrod, ed.), University of
Michigan Press, Ann Arbor, MI, 259-285.
Klepper, S., and Nagin, D. (1989a). The anatomy of tax evasion, Journal of Law,
Economics, and Organization, 5: 1-24.
Klepper, S., and Nagin, D. (1989b). The role of tax practitioners in tax
compliance, Policy Sciences, 22: 167-192.
Klepper, S., Mazur, M., and Nagin, D. (1991). Expert intermediaries and legal
compliance: The case of tax preparers, Journal of Law and Economics, 34: 205229.
Landsberger, M., and Meilijson, I. (1982). Incentive generating state dependent
penalty system, Journal of Public Economics, 19: 333-352.
Lewis, A. (1989). An empirical assessment of tax mentality, Public
Finance/Finances Publiques, 43: 245-257.
Long, J. E., and Caudill, S. B. (1987). The usage and benefits of paid tax return
preparation, National Tax Journal, 40: 35-46.
Long, S. B., and Swingen, J. A. (1991). Taxpayer compliance: Setting new
agendas for research, Law and Society Review, 25: 637-683.
Machina, M. J. (1987). Choice under uncertainty: Problems solved and unsolved,
Journal of Economic Perspectives, 1: 121-154.
Malik, A., and Schwab, R. (1991). The economics of tax amnesties, Journal of
Public Economics, 46: 29-49.
Manasan, R. G. (1988). Tax evasion in the Philippines, 1981-1985, Journal of
Philippine Development, 15: 167-189.
Martinez-Vazquez, J., Harwood, G. B., and Larkins, E. R. (1992). Withholding
position and income tax compliance: Some experimental evidence, Public Finance
Quarterly, 20: 152-174.

40

Martinez-Vazquez, J., and Rider, M. (1995). Multiple modes of tax evasion:


Theory and evidence from the TCMP, Working Paper, Department of Economics,
Georgia State University.
Mikesell, J. L. (1986). Amnesties for state tax evaders: The nature of and
response to recent programs, National Tax Journal, 39: 507-525.
Mookherjee, D., and Png, I. (1989). Optimal auditing, insurance, and
redistribution, Quarterly Journal of Economics, 104: 399-415.
Murray, M. N. (1995). Sales tax compliance and audit selection, National Tax
Journal, 48: 515:530
Myles, G. D., and Naylor, R. A. (1996). A model of tax evasion with group
conformity and social customs, European Journal of Political Economy, 12: 4966.
Pencavel, J. H. (1979). A note on income tax evasion, labor supply, and nonlinear
tax schedules, Journal of Public Economics, 12: 115-124.
Pestieau, P., and Possen, U. M. (1991). Tax evasion and occupational choice,
Journal of Public Economics, 45: 107-125.
Polinsky, A. M., and Shavell, S. (1984). The optimal use of fines and punishment,
Journal of Public Economics, 24: 89-99.
Pommerehne, W. W., Hart, A., and Frey, B. S. (1994). Tax morale, tax evasion,
and the choice of tax policy instruments in different political systems, Public
Finance/Finances Publiques, 49 (Supplement): 52-69.
Pommerehne, W. W., and Weck-Hannemann, H. (1989). Tax rates, tax
administration and income tax evasion in Switzerland, Discussion Paper No.
B8904, Department of Economics, University of Saarland, Saarbrucken.
Quiggin, J. (1993). Generalized Expected Utility Theory: The Rank-Dependent
Model, Kluwer Academic Publishers, Boston, MA.
Reinganum, J. F., and Wilde, L. L. (1985). Income tax compliance in a principalagent framework, Journal of Public Economics, 26: 1-18.

41

Reinganum, J. F., and Wilde, L. L. (1986). Equilibrium verification and reporting


policies in a model of tax compliance, International Economic Review, 27: 739760.
Reinganum, J. F., and Wilde, L. L. (1991). Equilibrium enforcement and
compliance in the presence of tax practitioners, Journal of Law, Economics, and
Organization, 7: 163-181.
Rice, E. M. (1992). The corporate tax gap: Evidence on tax compliance by small
corporations, in Why People Pay Taxes (J. Slemrod, ed.), University of Michigan
Press, Ann Arbor, MI, 125-161.
Rickard, J. A., Russell, A. M., and Howroyd, T. D. (1982). A tax evasion model
with allowance for retroactive penalties, Economic Record, 58: 379-385.
Roth, A. E., ed. (1987). Laboratory Experimentation in Economics: Six Points of
View, Cambridge University Press, New York, NY.
Roth, J. A., Scholz, J. T., and Witte, A. D., eds. (1989). Taxpayer Compliance,
Volume 1: An Agenda for Research and Volume 2: Social Science Perspectives,
University of Pennsylvania Press, Philadelphia, PA.
Sanchez, I., and Sobel, J. (1993). Hierarchical design and enforcement of income
tax policies, Journal of Public Economics, 50: 3435-369.
Sandford, C., ed. (1995). Tax Compliance Costs: Measurement and Policy, Fiscal
Publications, Bath, United Kingdom.
Sandmo, A. (1981). Income tax evasion, labour supply, and the equity-efficiency
tradeoff, Journal of Public Economics, 16: 265-288.
Scholz, J. K. (1994). The earned income tax credit: Participation, compliance, and
antipoverty effectiveness, National Tax Journal, 47: 59-81.
Scotchmer, S. (1989). The effects of tax advisors on tax compliance, in Taxpayer
Compliance, Volume 2: Social Science Perspectives, University of Pennsylvania
Press, Philadelphia, PA, .
Scotchmer, S., and Slemrod, J. (1989). Randomness in tax enforcement, Journal
of Public Economics, 38: 17-32.

42

Sheffrin, S. M., and Triest, R. K. (1992). Can brute deterrence backfire?


Perceptions and attitudes in taxpayer compliance, in Why People Pay Taxes (J.
Slemrod, ed.), University of Michigan Press, Ann Arbor, MI, 193-218.
Slemrod, J. (1985). An empirical test for tax evasion, The Review of Economics
and Statistics, 67: 232-238.
Slemrod, J., ed. (1992). Why People Pay Taxes, The University of Michigan
Press, Ann Arbor, MI.
Slemrod, J., and Yitzhaki, Y. (1987). The optimal size of a tax collection agency,
Scandinavian Journal of Economics, 89: 183-192.
Smith, K. W., and Kinsey, K. A. (1987). Understanding taxpayer behavior: A
conceptual framework with implications for research, Law and Society Review,
21: 639-663.
Smith, V. L. (1976). Experimental economics: Induced value theory, The
American Economic Review Papers and Proceedings, 66: 274-279.
Smith, V. L. (1982). Microeconomic systems as an experimental science, The
American Economic Review, 72: 923-955.
Spicer, M. W., and Becker, L. A. (1980). Fiscal inequity and tax compliance: An
experimental approach, National Tax Journal, 33: 171-175.
Spicer, M. W., and Hero, R. E. (1985). Tax evasion and heuristics: A research
note, Journal of Public Economics, 26: 263-267.
Spicer, M. W., and Thomas, J. E. (1982). Audit probabilities and the tax evasion
decision: An experimental approach, Journal of Economic Psychology, 2: 242245.
Srinivasan, T. N. (1973). Tax evasion: A model, Journal of Public Economics, 2:
339-346.
Stigler, G. J. (1970). The optimum enforcement of laws, The Journal of Political
Economy,
78: 526-536.

43

United States General Accounting Office (1976). How the Internal Revenue
Service Selects Individual Income Tax Returns for Audit, Report to the Joint
Committee on Internal Revenue Taxation, Congress of the United States,
Washington, D.C.
Usher, D. (1986). Tax evasion and the marginal cost of public funds, Economic
Inquiry, 24: 563-586.
Vogel, J. (1974). Taxation and public opinion in Sweden: An interpretation of
recent survey data, National Tax Journal, 27: 499-513.
Webley, P. (1987). Audit probabilities and tax evasion in a business simulation,
Economics Letters, 25: 267-270.
Webley, P., Robben, H., Elffers, H., and Hessing, D. (1991). Tax Evasion: An
Experimental Approach, Cambridge University Press, Cambridge, England.
Westat, Inc. (1980). Individual Income Tax Compliance Factors Study Qualitative
Research Results, Prepared for the Internal Revenue Service, February 4, 1980,
by Westat, Inc., Rockville, MD.
Witte, A. D., and Woodbury, D. F. (1985). The effect of tax laws and tax
administration on tax compliance, National Tax Journal, 38: 1-13
Yankelovich, Skelly, and White, Inc. (1984). Taxpayer Attitudes Survey: Final
Report. Public Opinion Survey Prepared for the Public Affairs Division, Internal
Revenue Service, December 1984, by Yankelovich, Skelly, and White, Inc., New
York, NY.
Yaniv, G. (1988). Withholding and non-withheld tax evasion, Journal of Public
Economics, 35: 183-204.
Yitzhaki, S. (1974). A note on 'Income tax evasion: A theoretical analysis',
Journal of Public Economics, 3: 201-202.

44

Vous aimerez peut-être aussi