AUDIT THE TAXPAYER NOT THE RETURN TAX AUDITING AS AN EXPRESSION

Description
This paper cxamincsth e practice of tax iulditingi n the ExaminationD ivisiono f the Internal Revenue
Service (IRS). Using data from 142 interviews of revemte agents in f&r dHerent geographical locations,
we analyze the audit procew aa an example of an “expression game” [G~flhun, E., Strategic Interaction
@hUadelphia,P A: Universityo f PennsylvaniaP ress, 1969)] in which the taxpayer attempts to maintaina
front an an honeat, compliant citizen and the revenue ?genta attempt to maintain a front as a fair,
competent professional who is just doing his or her job.

Accounting, ~gani.&t~ and society, vol. 21, No. 213, PP. 269-287, 19%
capyrishf ca 19% Elsevier sckncc Ltd
printed in Great ~rltain. Au rights resewed
0361~3fi82/96 $15.00+0.00
AUDIT THE TAXPAYER, NOT THE RETURN: TAX AUDITING AS AN EXPRESSION
GAME
BRIAN T. PENTLAND
Michigan State University
and
PAUL CARLILE
University of Michigan
This paper cxamincs the practice of tax iulditing in the Examination Division of the Internal Revenue
Service (IRS). Using data from 142 interviews of revemte agents in f&r dHerent geographical locations,
we analyze the audit procew aa an example of an “expression game” [G~flhun, E., Strategic Interaction
@hUadelphia, PA: University of Pennsylvania Press, 1969)] in which the taxpayer attempts to maintain a
front an an honeat, compliant citizen and the revenue ?genta attempt to maintain a front as a fair,
competent professional who is just doing his or her job. Expression games are a generic kind of strategic
Metaction in which participanta attempt to man?gc the impressions that others have of them and their
credibility. While certain aapecta of an expression game could be reduced to a formal dynamic game with
asymmetric information, the sociological perspective on the audit interaction seeks to preserve the
complexity of the moves and motivations that drive the interaction. Unlike formal game theoretic
models, this perspective draws attention to the practical, working knowledge required in audit work,
as well as the so&-emotional aspectn of the interaction. We use this framework to help understand the
limits of obiectivity, formnllzption, and technology in work that involves strategic interactions of this kind.
copyright (D 19% Else&r science L.td.
_.
In the relationship between the United States
Federal Government and the average citizen,
the Internal Revenue Service (IRS) is where
the rubber hits the road. The regularity and
inevitability with which we all pay taxes
makes the IRS one of the best known and
most widely disliked institutions in American
life (Watterson, 1988). At a macro level, we
can interpret the practice of tax auditing as
monitoring and enforcing one aspect of the
social contract between the people and the
government. As citizens, we are obliged to pro-
vide financial support for the activities of our
government. Although the obligation to pay
taxes is statutory and not contractual, one can
interpret the work of the IRS tax auditors in a
standard agency theory framework: they moni-
tor the behavior of taxpayers to insure compli-
ance. In this framework, the government is in
the role of principal, the taxpayer is in the role
of agent, and the job of the IR!3 Examination
Division is to monitor compliance with the
tax code (Reinganum & Wilde, 1985). Even
more so than other accounting professionals
(SingletonGreen, 1992), the IRS espouses the
ideal of objectivity in their auditing: catching
cheaters is important, but the uniform applica-
tion of the law is considered essential.’
’ Readers who doubt the sincerity of this objective need only consider the scandal that would emerge if systematic bias
were discovered. WC have no data concerning how well the IRS acmally meets this objective, but there is no question that,
like any public bureaucracy, they espouse it strongly.
269
270 B. T. PENTLAND and P. CARLILE
At a micro level, this monitoring process
unfolds through an elaborate set of interac-
tions between particular taxpayers and various
employees and systems of the IRS. As a practi-
cal matter, it is difficult for the IRS to capture a
complete and accurate assessment of what is
owed. Corporate taxpayers have elaborate
records which are simply too voluminous to
analyze completely, and individual taxpayers
may have no records at all. The goal of the
revenue agent is to look beyond the superficial
appearance as presented in the tax forms and
the supporting documentation. Records can be
forged and books can be cooked, so the
revenue agent must adopt a skeptical stance
that gives rise to the informal motto of the
Examination Division: “Audit the taxpayer,
not the return.”
The purpose of this paper is to explore the
game-like qualities of the audit interaction as
described by revenue agents. In a tax audit,
the taxpayer is generally attempting to main-
tain a “front” of honesty and compliance,
while the auditor is attempting to test that
front. At the same time, the auditor is attempt-
ing to maintain the appearance of a fair, com-
petent, dispassionate professional who is
simply doing his or her job. Even where hon-
esty, compliance, fairness and competence are
not explicitly at stake, a contest over facts
necessarily occurs in the taxpayer-agent inter-
action. The outcome of an audit (a revised tax
assessment) is the product of a variety of fac-
tual inputs, such as the valuation of an asset
that was donated to charity, the categorization
of certain deductions, and the amount of losses
to be claimed. Auditors are not allowed to inter-
pret the law, but they are required to interpret
the facts; in a sense, this interpretative process
is the primary skill of auditing (J ournal of
Accountancy, 1986). The ability to arrive at
an interpretation that is good enough for all
practical purposes is a critical part of the
knowledge developed and used by revenue
agents as they go about their work.
Audits are a classic example of a non-coop
erative game and have been analyzed using
formal game-theoretic techniques (e.g. Felling-
ham & Newman, 1985; Morton, 1993). As we
shall demonstrate, however, the realities of the
audit situation stretch the limits of applicability
of formal game theoretic concepts. For this
reason, we have chosen to approach the analy-
sis of the audit situation using Goffman’s < 1969)
concept of an “expression game.” This frame-
work allows us to analyze the audit process in
terms of a set of generic moves that auditors
and taxpayers use as they negotiate the facts of
the case and the outcome of the audit. These
moves are enabled and constrained by features
of the situation (Pentland, 1992) and are used
to enact the power relations between the IRS
and the taxpayer. This forms the basis of the
working knowledge that comprises the day-to-
day performance of the job. As we shall see,
revenue agents can have a thorough knowl-
edge of the tax law (formal knowledge), but
unless they can get cooperation from the tax-
payer, they will never complete an audit. The
distinction between formal knowledge and
working knowledge (Kusterer, 1978; Orr,
1990) helps explain some of the ways in which
the audit diverges from an idealized game.
The paper begins by introducing the theore-
tical background needed to interpret auditing
as a strategic interaction. We contrast the for-
mal game-theoretic interpretation to the less
formal, but more descriptively adequate frame-
work from Goffman (1969). This framework
provides a convenient way to express the
working knowledge of auditors and to contrast
that knowledge with the formal knowledge of
accounting rules and tax law that dominates
the traditional accounting literature (Sylph,
1988). Second, we introduce the setting for
this research and describe the methods used
in this study. Third, we present a detailed
description and analysis of the field audit pro
cess as experienced by revenue agents. The
emphasis in this description is on the interac-
tions with the taxpayer and the practical con-
straints that shape those interactions. Finally,
we discuss the implications of this analysis for
auditing and other strategic interactions where
an interpretation of facts is required.
TAX AUDITING 271
THEORY: STRATEGIC INTERACTIONS
To say that tax auditing is a game merely
states the obvious: taxpayers are motivated to
minimize their tax burden and the IRS is
responsible for making sure they do so within
the limits of the law. The question is, what kind
of game is a tax audit and what insights can we
gain about auditing (and similar kinds of inter-
actions) from an analysis of its game-like prop
erties? To answer this question, we will
explore two alternative perspectives on strate-
gic interactions that might be applied to the
audit situation: formal games and expression
games.
Formal games
Audit situations can be interpreted as formal
games (Graetz et al ., 1986; Fellingham &
Newman, 1985; Morton, 1993), where the audi-
tor and the taxpayer attempt to construct opti-
mal strategies based on their knowledge of
each other’s options and payoffs. Formal
game-theoretic analyses have provided a help
fi_tl corrective to traditional, decision-theoretic
approaches to auditing which overlook the
possibility that the entity being audited may
be reacting strategically to the auditor (Eelling-
ham & Newman, 1985). They also constitute an
improvement over principle-agent models (as
in Reinganum & Wilde, 1985), where the prin-
ciple commits to a particular auditing policy in
advance and the agent may respond strategi-
cally to that policy. By acknowledging that
each participant in the audit game may behave
strategically, the game-theoretic framework
provides the greatest face validity of any of
the economic models of auditing (Graetz et
al ., 1986). But as Fellingham and Newman
(1985, p. 636) acknowledge, “the game-theo-
retic approach promises to be equally intransi-
gent in application.” The burden of
constructing an analytically tractable set of
moves and payoffs limits the applicability of
the formal framework to a highly simplified
and decontextualized set of situations. The
basic structure of the audit situation is espe-
cially difficult to capture from a game-theoretic
point of view, for the following reasons.
First, there is imperfect and incomplete
information because the auditor, in particular,
is ignorant of what the taxpayer has actually
done. This asymmetry of information is the
fundamental motivation for auditing, of
course. Some taxpayers design their actions
to be deceptive or unobservable, through the
use of cash or barter, for example. Other tax-
payers simply make mistakes. The job of the
auditor is similar in either case, but the details
of how it is carried out depend on the auditor’s
perceptions of the taxpayer’s willingness or
ability to deceive. Auditing a well-intentioned
but ignorant taxpayer is quite different from
auditing someone who is actively attempting
to commit tax fraud. In theoretical treatments
of the subject (e.g. Graetz et aZ., 1986), it is
assumed that an IRS audit eliminates the infor-
mation asymmetry and results in the correct
determination of taxable income. As we shall
see, this is not necessarily the case.
Second, the audit game is dynamic on at least
two levels: within a given year and between
years. Within a given year’s audit, the auditor
and the taxpayer typically visit and revisit var-
ious issues; it is rare to completely settle a case
on a single visit. Since the audit interaction may
take place over a period of weeks, events from
one encounter can influence actions at the
next. Similar dependencies exist between
years, as well, since the taxpayer may readjust
his or her strategy in reaction to the outcome
of the prior year’s audit. In current formula-
tions (e.g. Graetz et al ., 1986) the audit game
is conceived of as a single round of play in
which each party (the taxpayer and the IRS)
gets just one move: the taxpayer reports, and
the IRS either audits or does not.
Third, because the IRS is a public agency and
taxpayers often share information (through
their accountants, for example), the strategy
and outcome of a particular audit can have
spillover effects to other audits. The IRS
makes strategic use of this tendency when
they target particular occupational commu-
nities (e.g. hairdressers, auto mechanics,
272 B. T. PENTLAND and P. CARLILE
restaurants) in specific geographic areas. By
signaling owners of cash businesses that they
may be subject to audit, the IRS attempts to
encourage “voluntary compliance” among the
entire community (Klepper & Nagin, 1989).
The strategy can backfire, however, if it
encourages more careful concealment of the
underground economy. Because millions of
taxpayers are playing this game every year,
the practical payoff of a particular audit is diffi-
cult to predict.
Finally, in spite of the fact that tax audits are
primarily economic encounters, they have
enormous costs and potential payoffs for each
party that are difficult to measure in economic
terms. Taypayers, even honest ones, generally
experience audits as extremely stressful inter-
actions and are often willing to concede issues
to which they might be entitled just to get the
process over with or to avoid the audit in the
first place (Carroll, 1987). And as we shall see,
auditors have incentives to complete audits
that do not seem to correspond to the objec-
tive of revenue maximization. Some of these
incentives are the natural result of working in
a large bureaucracy that strives for efficiency,
but others have to do with the auditors’ need to
cope with an exceedingly stressful job. These
factors further complicate the problem of spe-
cifying well-defined payoffs for the partici-
pants. Thus, while the formal game-theoretic
approach yields important insights in simpli-
fied situations, it cannot easily be applied to
real audits.
Expression games
An equally general, but less formal model for
strategic interaction was proposed by Goffman
(1969). In most face-to-face communications
there is a transmission of expressed informa-
tion between participants. But when these
expressions are contested, game-like condi-
tions prevail (Goffman, 1969, p. 10). Goffman
argues that while these features are most
vividly illustrated in situations like espionage,
they are general properties of all human inter-
action (Goffman, 1967). In any kind of face-to-
face interaction, individuals engage in some
degree of impression management (Goffman,
1959). This is true whether the interaction
occurs in public or a highly institutionalized
setting (Goffman, 1961). Usually, participants
collude to support the everyday expressions
and behaviors that are used to create and main-
tain face; this collusion allows participants to
sustain normal social interaction (Goffman,
1967). But these same expressions may also
be scrutinized as to who or what is to be
believed, and in these circumstances, the nor-
mal expression game may become overtly stra-
tegic (Goffman, 1969). It is important to note
that in expression games, it is not just the
“facts” that are at stake: the identities of the
participants are being tested and negotiated, as
well.
The practice of tax auditing can be framed as
a strategic game where agent and taxpayer
engage in a series of “moves” (Goffman,
1969, pp. 11-28) where each individual plays
off the “moves” of the other - like a fencing
duel, but one that is communication or expres-
sion based (Mead, 1934, p. 42). Goffman
describes five generic moves that can be used
to analyze this interaction. This list is not meant
to be exhaustive in a technical sense, but as a
helpful frame of reference to understand the
properties of these kinds of social interactions
(Goffman, 1969, p. 11). While Goffman intro-
duced these moves using examples from the
world of espionage (spies and counter-spies),
we will introduce them here using examples
that foreshadow our analysis of the tax audit
game. Note that while it is natural to cast the
taxpayer in the role of the “spy” who is
attempting to cover the truth and the revenue
agent in the role of the truth-seeker, these
moves can be made by either party to the inter-
action.
Unwitting moves occur when the subject is
unaware that he or she is being observed. Tay-
payers normally go about their business with-
out being directly observed and may develop a
false sense of privacy that encourages them to
believe that “big brother” is not, in fact, watch-
ing. For this reason, taxpayers can be viewed as
making unwitting moves throughout the year.
TAX AUDITING 273
But when a taxpayer fills out a set of tax forms,
he or she offers the revenue agent observable
data and is typically very conscious of how it
may be interpreted (Carroll, 1987). There is no
way to make an unwitting move on a tax form.
Naive moves occur when the observer
believes that he can take the subject as he or
she appears. Revenue agents report that they
rarely adopt a purely naive stance towards tax-
payers because they believe that nearly every-
one cuts comers on tax returns. In principle,
nothing can be taken at face value. Of course,
in practice, the revenue agent is forced by the
constraints of the situation to take a great deal
at face value and the ability to do so judiciously
lies at the heart of their interpretive skills and
their ability to “read people”. The taxpayer, on
the other hand, may wonder if a revenue
agent’s friendly question about where he spent
his vacation is really just idle conversation or
part of a systematic attempt to gather informa-
tion.
Control moves happen when the subject
intentionally produces an expression or beha-
vior that he thinks will improve his situation
with the observer. The audit game is filled
with moves that embody attempts on the part
of the taxpayer (or in some cases, the Revenue
Agent) to control the impression being con-
veyed. In the audit this might consist of stal-
ling, having unorganized records, getting
angry, or just not answering questions. One
agent mentioned that a taxpayer refused to let
her plug in her computer “because electricity
costs money”. The time constraints that rev-
enue agents work under can make these delay-
ing tactics pay off for the taxpayer by forcing
the agent to limit the investigation or in some
cases, to say “to hell with it” and move on to
another audit. Control moves are thus an
important part of impression management.
Uncovering moves take place when the
observer tries to uncover the true motives of
the subject. To a large extent, uncovering is
what auditing is all about. Information that a
taxpayer gives early on in the audit that was
originally seen as sincere (i.e. an unwitting or
naive move) may over the course of the audit
be seen as something that needs to be uncov-
ered. Uncovering moves can be covert (e.g.
monitoring a taxpayer’s facial expressions
while they talk) or overt (e.g. collection and
analysis of bank statements). Agents often men-
tion that they “have to be aware of body lan-
guage in this job to tell if a taxpayer is being
honest.” Body language can only go so far, of
course; revenue agents must not only collect
and analyze specific data, they must know
what the numbers they uncover mean (i.e.
that a small refrigerated delivery truck doesn’t
cost $500,ooo).
Counter-uncovering moves occur when the
subject realizes that the control moves they are
performing might be detected and that they
need to meet this uncovering move with a
counter move. Counter-uncovering moves con-
sist of some of the control moves mentioned
before. More specifically, guarding against
detection consists of knowing the tax law as
well or better than the revenue agent perform-
ing the audit. This is why accountants and tax
lawyers (who act on behalf of taxpayers being
audited) present an especially prominent chal-
lenge to the revenue agent. Sorting all of this
out empirically is difficult for the participants,
and practically impossible for an observer, for
as Goffman points out, “the best evidence for
the observer is also the best evidence for the
subject to tamper with” (Goffman, 1969, p.
28). The point is that the game consists of
both reactive and anticipatory moves and that
in general, it starts before the tax return is filed,
through the preparation of records, the struc-
turing of deals and contracts, and the conduct
of business in general.
Comparison of perspectives on strategic
interaction
While Goffman’s expression games are expli-
citly dynamic, with asymmetric information
driving the action, there are some fundamental
differences between the sociological and the
economic approaches. The economic model
is concerned, first and foremost, with “sol-
ving” the game to identify the optimal strategy
for each participant. Since the publication of
274 B. T. PENTLAND and P. CARLILE
Becker’s (1968) normative economic analysis
of optimal enforcement levels, in which tax
auditing was explicitly mentioned as a possible
application, this approach has dominated the
literature on tax auditing (Allingham &
Sandmo, 1972; Reinganum & Wilde, 1988).
Most of this literature has assumed that the
IRS and its audit policy are exogenously given
and the behavior of the taxpayer is determined
by rational choice. The inclusion of “strategic”
audit selection criteria that vary based on the
data provided by the taxpayer has been a
recent innovation in the formal analysis of tax
compliance (Graetz et al ., 1986). But selection
of a return for audit is conceived of as the last
move the IRS makes in the audit game, because
the results of the audit are assumed to be an
accurate tax assessment. At best, audit results
are treated by a kind of variable payoff func-
tion. Thus, where the formal game ends, the
expression game begins.
While they lack computational elegance,
expression games provide a much richer voca-
bulary with which to describe the audit game.
The generic moves in an expression game pro
vide no guidance at all concerning optimal stra-
tegies, but they capture the texture and feel of
the practice as experienced by the participants
in terms that can be generalized across situa-
tions. As with formal game theory, there is an
interest in the structure of the situation, but
also in the ongoing practices, working knowl-
edge, and emotional states of the participants.
The expression game framework also calls
attention to the ways in which revenue agents
themselves may engage in impression manage-
ment during an audit. The key difference is that
while formal game theory treats the facts as
exogenous to the audit - waiting to be discov-
ered by the auditor - the expression game
treats the facts as endogenous to the audit -
a product of the game itself.
It is important to note that each framework
tends to simplify the audit situation in some
respects. Neither framework explicitly
accounts for the organizational setting; while
the expression game framework is less formal
and therefore provides more context, it none-
theless focuses on the audit interaction itself.
Likewise, neither framework explicitly
accounts for the possible presence of multiple
individuals on each side of the audit interac-
tion. Taxpayers often have accountants or
attorneys who represent them, and for larger
cases, auditors may work in teams. For these
reasons, each framework tends to portray a
rather stylized version of a very complex
situation.
RESEARCH SITES AND METHODS
The I RS Exami nati on Di vi si on
The United States tax system depends on
“voluntary compliance”: taxpayers are
expected to tile returns and pay what they
owe (Klepper & Nagin, 1989). Of course, it is
never really quite that simple. In a multi-
method field study of taxpayer behavior, Car-
roll (1987) reports that many taxpayers know-
ingly cheat on their returns but rationalize this
behavior by saying that the net effect is that
they basically “pay what they owe”. Because
taxpayers may espouse a standard of fairness
and obligation that departs significantly from
the Internal Revenue Code, it is necessary to
conduct audits, as this IRS manager explains:
Overall, our biggest effort is voluntary compliance. To
get compliance, we have to do audits. There has to be
the possibility of an audit or else the whole system will
go to hell.
The Examination Division is the part of the
IRS that is responsible for conducting these
audits. The IRS is organized as a matrix of func-
tional units (such as examination, collection,
taxpayer service, criminal investigation, and
so on), and geographical units. Thus, for any
given geographical location, a cross-section of
functional units is generally present. Geogra-
phically, the IRS is organized into seven
regions, each of which contains several dis-
tricts. There are a total of 64 districts, some
of which cover entire states (e.g. Massachu-
setts), others of which cover only part of a
TAX AUDITING 275
major metropolitan area (e.g. Brooklyn).
Within a district, there are generally several
physical locations, called posts of duty (POD).
Finally, at any particular post of duty, there may
be one or more field audit groups, each of
which has a group manager and a roster of
lo- 12 revenue agents.’
Methods and data
The data reported here come from two
rounds of interviews conducted by the first
author, approximately one year apart. The first
round consisted of a series of 15 days of field
visits between February and July, 1987. Alto-
gether, over 80 persons were interviewed,
either individually or in small groups, repre-
senting four different IRS district offices, two
in the Northeast and two in the Midwest.
Within each district, agents from at least two
field audit groups were interviewed. Several
examination division personnel in the national
office were also interviewed. Interviews were
unstructured, lasting between one and two
hours each. Taping was generally not
allowed, so the raw data consist of expanded
interview notes. Quotes are taken directly from
these notes, with names, places and other iden-
tifying references omitted to protect confiden-
tiality. The second round of interviews
consisted of 16 days of field visits conducted
between June and October, 1988. The objec-
tive of these interviews was to follow up on
the use of technology in each of the four sites
visited in the first round. During the second
round, an additional 62 interviews were con-
ducted. Recent published reports of IRS audit-
ing practices (Sanders, 1995; Lohse, 1995)
suggest that auditing procedures in the Exam-
ination Division have not changed substantially
since the time these interviews were
conducted.
This research was originally intended to
study the use of portable computers by
accounting professionals within the IRS (Pent-
land, 1989). Thus, while some interview time
was devoted to the technology per se, a con-
siderable amount of time was devoted to under-
standing the nature of field auditing in general.
Towards this end, revenue agents were asked
to describe typical audits, easy audits and diffi-
cult audits. These descriptions provide a rare
window into the practice of tax auditing,
which most people can only see from the
uncomfortable position of the taxpayer being
audited. Since these data were collected
through interviews rather than observations,
however, they must be treated as accounts pre-
sented for an outsider, subject to some degree
of impression management on the part of the
IRS interviewees. Although most respondents
seemed quite candid in their views, and most
of the data has a distinctly “backstage” flavor to
it (Goffman, 1959) it would have been desir-
able to supplement the research with observa-
tions of actual audits. This was not possible,
unfortunately, due to IRS regulations on tax-
payer confidentiality. Consequently, the analy-
sis that follows represents that audit game from
the viewpoint of only one of its players: the
revenue agent.
THE FIELD AUDIT
The IRS conducts three basic kinds of audits
which are distinguished by the venue in which
they take place. Service center audits are con-
ducted completely by correspondence and
typically involve correction of errors on a
return. Office audits are conducted in a local
IRS office and are typically limited to a narrow
set of issues identified by computer analysis or
a Some small posts of duty in rural areas may have only part of a field audit group (i.e. one or two agents). All of the
interviews reported here were conducted at posts of duty with at least two revenue agent groups.
’ Only 14 of these individuals were also included ln the original round of interviews. Although the second round of visits
targeted the same locations, revenue agents generally spend most of their time out of the office, and they arc often given
temporary assignments in other functions, such as taxpayer assistance, tralnlng, or quality control. Given these limitations,
we make no attempt to analyze these interviews as a longitudlnal panel.
276 B. T. PENTLAND and P. CARLILE
by human reviewers. Field audits are con-
ducted in the taxpayer’s home or place of busi-
ness and generally involve a wider range of
issues that can be expanded or narrowed at
the discretion of the revenue agent handling
the case. In this paper, we will be concerned
exclusively with the field audit process.
Revenue agents
Revenue agents conduct field audits of high-
dollar individuals and sole proprietorships
(form 1040) partnerships (form 1065) and
corporations (form 1120). The term “field
audit” derives from the fact that the revenue
agent visits the taxpayer’s home or business;
Revenue agents are expected to spend as
much as 80% of their time in the field, with
the balance spent on office paperwork, legal
research, and other activities. The actual
work of conducting a field audit is extremely
varied, because revenue agents cover the
entire range of taxpayers present in the mod-
ern American economy, from investment banks
to drug dealers. Every taxpayer has a unique set
of books, circumstances, and tax issues.
To deal with this potential variety, revenue
agents are required to have a college level
accounting degree (or equivalent experience)
and many are certmed public accountants.
There are approximately 18,000 revenue
agents employed by the Examination Division.
Revenue agents are graded according to the
civil service pay grades, starting at GS7 (for
an entry level agent) and moving up to GS-13
(for an agent with many years of experience).
As revenue agents gain seniority, they are given
more and more complex cases. Moving up to a
higher grade is a function of experience and
productivity (timely completion of audits).
Newer agents, who audit mostly individual
returns and sole proprietorships, tend to say
that the most important part of the job is deal-
ing with taxpayers: “The biggest aspect is the
meet and deal skills. Without that, you go abso-
lutely nowhere on this job.” Senior agents,
whose work consists almost exclusively of lar-
ger corporations, tend to emphasize technical
skills: “Accounting is the prime skill. Without
that, you can do a 1040 and the schedule C, but
on a big exam, you’ve gotta be able to under-
stand the books.” These seemingly contradlc-
tory responses reflect the difference in the
kinds of assignments that revenue agents are
given as they progress in their careers.
A critical part of becoming a revenue agent is
adjusting to work in a stigmatized organization.
This is a difficult transition for many revenue
agents, as this manager explains.
We’re not exactly the most popular organization in the
world and it takes a while to get used to workiug here.
This job isn’t for everyone. There are some people who
just can’t take the stress of having to go out and con-
front people day ln and day out. And not everyone tteats
us real nicely, either.
This point is illustrated vividly by this story
from a revenue agent in the mldwest.
But I had this thing recently when my fiancee and I
were looking at a condo. The owner and his wife
were showing us the place and at some point my flan-
tee said something like, “Watch what you say around
Hemy, he works for IRS.” The woman looks at me and
says, “I knew there was something I didn’t llke about
you.” Now usually people say things like that ln a joking
kind of way, and it’s funny. But she was really mean, llke
she really meant it. I felt very uncomfortable.
Revenue agents have many stories of thls
kind, although they seem to be most sallent
among newer members of the organization.
In an effort to protect himself against this
stigma, one revenue agent reported that
whenever someone asked where he worked,
he would reply “for the government”, or “for
the treasury department”, or “downtown”, but
never “for the IRS.” Our point in mentioning
this here is that it would be wrong to assume
that revenue agents have, as a group, fully
internalized the legal framework which they
have been hired to apply and enforce. They
are people, and they are also taxpayers, like
anyone else. To lose sight of the emotional
side of audit work (as reflected, in part, in the
“meet and deal” skills mentioned above)
would be to overlook an important aspect of
the field audit process.
TAX AUDITING 277
Description of tbe field audit process
Comparing our observations of IRS to other
published accounts (Surface, 1967; Chommie,
1970; Diogenes, 1973) shows that the basic
structure, role, and daily life of the Rxamina-
tion Division has remained largely unchanged
for nearly thirty years. In the following
sections, we describe the details of the field
audit process and analyze it as an expression
game.
Selecting returns for audit. Returns are filed
at one of ten IRS service centers, where they
are classified and selected for audit. Returns are
selected for a variety of reasons; the decision
process can involve computer scoring, random
sampling, or human judgment. Once selected,
returns are sent to the geographical district
where the taxpayer resides or does business.
From there, returns are assigned to an appro-
priate audit group for examination. These
returns are the “inventory” for the field audit
group, and the manager’s main job is to “move
the inventory.”
In Rling a return, the taxpayer makes the first
control moves in the expression game. Tax-
payers present a limited, stylized set of informa-
tion that purports to summarize their economic
activities for the year in question. In selecting a
return for audit, the IRS makes its first uncover-
ing move. At this level, from the taxpayer’s
point of view, the game consists of a single
question: “Will I get audited if I include (or
omit) this information on my return?” (Car-
roll, 1987). But filing and selecting returns is
only the beginning of the audit game.
Assigning returns to individual agents.
Agents in each group are assigned cases by
their group manager based on their training
level and experience. The rule is one agent
per case; with the exception of very large
cases and certain specialty areas, revenue
agents work alone. On one hand, managers
are supposed to assign cases that develop the
skills of the examiners and agents need
“developmental work” in order to qualify for
promotions. On the other hand, the need to
“move the inventory” also creates a pressure
to assign cases to agents who are already famil-
iar with the issues. However, the incredible
diversity of work guarantees that most agents
cannot specialize very much, if at all. Accord-
ing to one group manager, “Private accounting
firms generally specialize, they get to know the
details of certain lines of business. But the IRS
doesn’t have that luxury.”
Common wisdom has it that the public is
powerless against the omnipotent hand of the
IRS. While the IRS does have extraordinary
powers to collect taxes which seem to go
beyond our concept of “due process”, the rea-
lity of the audit situation is not quite as one-
sided as one might believe. A group manager
in the Northeast region explains his perspec-
tive as follows.
On any given case, there’s only one ‘agent assigned.
Then if that case goes unagreed, there’s only one
appeals officer. And lf it goes to tax court, one prose-
cutor. People have the idea that the government has all
this power, but ln real@ the IRS is really outgunned
with it comes to legal and technical talent. The tax-
payer almost always has more resources lf they choose
to apply them ln a particular case.
One may question this manager’s efforts to
downplay IRS enforcement powers, because
not every taxpayer has access to the kind of
resources and legal representation that are
required to “outgun” the IRS. At the same
time, there are other more mundane ways in
which every taxpayer holds some important
cards. First of all, the taxpayer knows more of
the details of his or her own finances. The
taxpayer knows if another set of books has
been kept, and if so, where it is. The revenue
agent only knows what has been reported on
the return, which can be viewed as a covering
move intended to create the impression of
compliance. Second, the taxpayer is likely to
have more time and attention to devote to
the case than the revenue agent does. Agents
must allocate their time among several audits in
their inventory. The taxpayer, on the other
hand, has only one case that concerns them.
Of course, if a taxpayer is relying on a paid
representative, the cost of these representa-
tions may become significant.
278 8. T. PENTLAND and P. CARLILE
The i ni ti al i ntervi ew. Once assigned a case,
the agent has to schedule the initial meeting
with the taxpayer and, in many cases, the
accountant or attorney representing the tax-
payer. The initial interview sets the tone for
the rest of the audit process. The agent forms
impressions about what to expect from the
taxpayer, and may elicit important informa-
tion: “This is a critical part of the audit. People
will often tell you a lot of things up front, and
then clam up later, if they think we’re onto
something.” The taxpayer’s reaction is also
important: “If you lose the respect of the tax-
payer, you’re dead in the water. If you mess up
the interaction, you may get nowhere.” This
comment underscores the revenue agent’s
dependence on the taxpayer and the need for
what they call “meet and deal” skills.
The initial interview provides an occasion for
a rich set of moves. There will surely be an
array of control moves and uncovering
moves, as the taxpayer attempts to establish a
trustworthy image and the auditor attempts to
test it. There may also be unwitting moves,
where the taxpayer fails to realize that off-
handed comments about a vacation may be
an important clue to an auditor. A critical
aspect of the initial interview, from the audi-
tor’s point of view, is that it provides the first
occasion to look beyond the tax return. Audi-
tors use this opportunity to begin forming judg-
ments about the taxpayer.
Determi ni ng the scope. As part of starting an
audit, the agent must also decide what issues
will be examined. Determining the scope of
the audit is a crucial step and the agent has a
fair amount of discretion here. Certain issues
are mandatory, but other issues are chosen by
the agent based on data gathered during the
initial interview and throughout the audit pro
cess. These decisions require the agent to
voluntarily take on considerable stress and
responsibility:
To be perfectly honest, I could make it less stressful by
just going for the tiny little adjustments I’d never
have to work hard, the taxpayer would get away with
all the really big stuff, and it’d all be real easygoing. But I
don’t work that way. I like to go after the bigger issues,
the tougher ones, and so I think I put a lot of stress on
myself.
Revenue agents are operating on a schedule
and must be careful about the time applied to
each case. Group managers are under pressure
to “move the inventory”, and unless there is
adequate justification, revenue agents are
expected to minimize the “time on case”. Mini-
mizing time requires taxpayer cooperation:
“you want to close cases as quickly as possi-
ble, so you have to be able to get the taxpayer
to cooperate.” Two kinds of cooperation are
needed. Throughout the audit, the agents
must get the taxpayer to provide the necessary
information to do the audit. Then, as the audit
nears completion, the agent must get the tax-
payer to willingly comply with the “adjust-
ments”. For this reason, dealing with the
taxpayer is a critical practical constraint in
the audit process.
For revenue agents working under these
practical constraints, justice is best served
through a give and take process that some
label as “horse trading”. Forcing adjustments
on the taxpayer through the formal application
of tax rules and IRS regulations often leads to
belligerence, non-agreements and appeals. A
revenue agent in the Midwest summed up her
thoughts concerning this issue in this way:
Some agents really want to collect the taxes owed. For
myself, I basically don’t have any unagreed cases. I can
give it [revenue] away just as easily as appeals can [a
statement about the inability of the tax courts to reach
agreements as well]. I’d rather do it myself because that
way I know what’s really being given away.
It is intriguing to note that a revenue agent
can be aware of additional taxes due and
choose to ignore them in the interest of com-
pleting an audit. The practice is reminiscent of
“plea bargaining” in criminal cases; in effect,
the revenue agent is telling the taxpayer,
“agree to this issue and we won’t go into that
one.”
One could view this practice as a rational
response to an overload of cases, but there is
TAX AUDITING 279
more to it than that. The audit game involves
objectives and payoffs for the auditor that are
distinctly non-monetary. Like taxpayers, audi-
tors find their work highly stressful. Agents
must be careful to maintain the appropriate
front for extended periods of time, as this
agent explains:
Spending the entire day with a taxpayer is very stressful
because you can never let your guard down. You have
to be real polite and be careful what you say the whole
day, with no breaks.
Worse yet, revenue agents are often required
to audit individuals who are already under-
going financial hardships and have cheated on
their taxes to help tide them over (e.g. cases
Involving divorce, medical disabilities, and
personal bankruptcy are common in auditor’s
discussion of their work). Not everyone can
handle these kinds of audits, as this manager
explains.
We know you’re really part of the IRS team when you
can go out on an audit where the person has taken all
kinds of questionable deductions so they can afford to
have their kids teeth fixed. The guy starts screaming or
crying, because he’s got-ma owe maybe 10% of his
income, and he needs that money. To do your job,
you have to go out there and explain it to him,
empathize with his problem, maybe let him cry on
your shoulder, and still get that adjustment. If you can
do that and still live witb yourself the next day, then you
arereallypartoftheIRSteam.
These examples also suggest some moves
that Goffman did not envision that are specific
to the auditing game, such as “giving it away”.
The problem of defining the scope of an audit
has been treated in formal models (e.g. Felling-
ham & Newman, 1985). In terms of Goffman’s
framework, it would entail the decision not to
pursue certain uncovering moves. These audi-
tor’s comments suggest, however, that limiting
the scope of an audit and giving away revenue
are not always undertaken as part of an opti-
mal, revenue maximization strategy. gather,
they may be undertaken as part of a personal
self-preservation strategy on the part of the
auditor.
Gatheri ng the data. As the audit progresses,
the agent may make subsequent visits, request
documents from the taxpayer, and so on. This
is where the expression game can be seen in its
most elaborate form. The process of gathering
and analyzing data is interactive; the data
needed depends on what has already been pro
vided. At their discretion, the agent can
broaden the scope of the audit to past years,
or bring in related returns (of family members,
business associates, or investors). In some
cases, particularly those where fraud is sus-
pected, an agent may also gather data from
third party sources. This may mean tracking
down documentation from customers and sup
pliers, getting appraisals of merchandise or
gifts, and so on, as in this description of a
“hard audit” by a revenue agent:
Another one that was wild involved the overvaluation of
livestock. What happens is that party A sells it to party
B, who sells it to party C. They collude to create
invoices that show certain prices, while in fact they
pay much less. This allows them to overvalue the items
for tax purposes, thereby deducting more, and they
have valid-looking documentation, and so forth. We
had to bring ln a specialist who says, “No way is that
worth that much .” and then trace it back to party A,
and so on. It can take a lot of detective work.
Often, the revenue agent is dependent on
what he or she can learn from the taxpayer
(or accountant or attorney representing the
taxpayer) - there being no other sufficient
source of information. An agent in the Midwest
provides a unique metaphor for this aspect of
the interaction and explains what can happen
if the interaction goes badly:
Dealing with the taxpayer is llke walking a tightrope.
You have to pick every word you use so carefully. Some-
times they do things to deliberately delay things that
really get you angry. Once, I was doing an audit where
I needed some information from the taxpayer and
wanted to call him. The accountant said be didn’t
have the phone number, so he’d have to get back to
me. Later on, something else came up and the accoun-
tant picks up the phone and dials the taxpayer, just like
that. Now obviously be was lying before, but I can’t say
anything, even though what I really want to do is wring
his neck. And once something like that happens, it
280 B. T. PENTLAND and P. CARLILE
changes the whole situation. I mean if you trust some-
one you can accept a lot of credible oral testimony even
lf they don’t have all the right supporting documenta-
tion. But once they start pulling that kind of thing, you
aren’t going to belleve anything they say.
Once a taxpayer loses credibility on one
issue, the auditor is far less likely to accept
their word for anything. In terms of our
model, when one coveting move fails, it
decreases the likelihood of subsequent moves
succeeding Thus, it makes sense that many
agents suggest that “we are in the business of
reading people.” In this sense, revenue agents
really are “auditing the taxpayer”, searching
for clues as to whether the taxpayer’s state-
ments can be treated as factual. One agent
put it this way: ‘you’ve got to be inquisitive,
and don’t take things at face value. Evaluate
the person you are talking to - is he a high
roller or whatever? They tell you what they do,
but you can’t be sure if it’s the truth. You
almost need detective skills.”
The problem revenue agents face, of course,
is how far to go in searching for the facts of the
case. revenue agents have significant power to
force taxpayers to disclose information, but
they do not always want to apply it, as this
agent explains:
for next month. She rescheduled, but it basically cost
her a day when she had nothing else to work on. When
she came back to the office to tell me about it, I sat her
down and told her that she should never have let the
guy get away with that without at least dressing him
down but good. I made her memorize Section 10.2 of
the regulations governing the practice of representa-
tives before the IRS, according to which he has the
duty to provide access to records promptly upon
request. The next time someone tries to pull that
stuff, she can quote those regs and give the guy an
earful about how lf he keeps this up, we’ll report hlm
to hls licensing board, how it costs the taxpayers time,
how the government has a job to do, and so on. It won’t
help for that day, but he’ll be less likely to do it to some
other agent, and he’ll never do it to you again.
In this example, the manager identifies the
interdependence between audits. By giving the
attorney an “earful” and threatening punitive
action, the revenue agent may not advance the
progress of the present audit, but may help
with future audits. In this way, the strategy
and demeanor of the auditor in one situation
can spill over and influence the outcome of
other audits.
Stalling tactics are common; they represent a
kind of covering move that can help swing the
audit in the taxpayer’s favor by making it more
diicult for the revenue agent to get informa-
tion that is necessary to complete the audit.
The maln question is, when do you want to engage in
battle, and when do you want to cut your losses? I have
dropped out of audits because it wasn’t worthwhile to
do battle. You have some real signllicant power, if you
use it when it’s appropriate. You can summons records,
stat notice [“statutory” computation of taxes without
the taxpayer’s cooperation], and so on, but you don’t
want to do that ln every case.
Agents here get a lot of Back from reps. . It’s a stalling
tactic; people try to keep us from doing our work
because its to their advantage to hold us off. Some
agents may eventually just give up. In any event, older
cases are harder to close. The trail can be harder to
follow in some cases. Also, if an agent has to go on
detail, that puts it on hold or may force them to transfer
the inventory, all of which makes it harder to really do a
thorough job. So stalling can be useful.
A group manager tells the following story
about how a new revenue agent was “jacked
around” by an attorney who wanted to delay
an audit for a client:
These kinds of tactics can be taken to
extremes by some taxpayers, as the following
example illustrates.
The biggest and most important part of the job is deal- It’s an audit of a couple of attorneys who are represent-
ing with taxpayers and reps, and deciding where you ing themselves and are giving us the hardest time you
draw the line. Even if the taxpayer is the customer, as can lmaglne, the most childish klnd of things. And since
they like to say these days, you can’t let them jack you they are the taxpayer, Section 10.2 doesn’t apply. This is
around. I had an agent ln one of my tralnlng groups who the klnd of thing they would do. [Picks up a piece of
went out to an attorney’s office for an audit and was
paper to demonstrate ] Our agent asks for a docu-
told that he wasn’t ready yet, and could she reschedule ment, and the attorney says, “Here it is.” [Holds paper
TAX AUDITING 281
on desk, not moving hands off of it.] Our agent asks,
94ay I please MC it? *’ and the attorney says, “OK, you
can see it.” [Holds paper up near hia head, across the
table about 6 feet away, 80 you couldn’t MC it without
binocukuu.] Or maybe they’ll hold it out like this and
when the agent reaches for it, they’ll pug it away. Au
the while keeping a calm, slow, deliberate demeanour.
So WC served them with a summons, which they fought
in court. We’ve had two judges dismiss tbemsehres from
the case because these people were so bad. . It’s just
crazy, but people treat us that way sometimes.
Whether it is crazy or not, this example
illustrates the extent to which some taxpayers
may go to prevent or forestall disclosure of tax
information to IRS. In the face of such opposi-
tion, it would be easy for revenue agents to
become highly aggressive. It is important,
however, that revenue agents not be ovenea-
lous in their pursuit of tax dollars, as this
manager explains.
If we go out there looking for a yield, we’re going to
destroy morale and the IRS’s reputation. And we’d prob
ably provoke people into cheating as much as possible,
which is exactly the opposite of what we want to do.
Part of my job as manager is to keep agents from being
too aggressive, from taking on the yield mentality. Just
the other day, we had an agent get thrown out of a
[taxpayer] rep’s office for being too aggressive. He
was demanding all kinds of things, generally overstep
ping what is normally accepted. He’s a new guy and was
letting it ail go to his head.
These stories provide classic examples of
control moves and uncovering moves. They
also reveal the exceedingly dynamic character
of the game within a given audit. These fea-
tures are consistent with the expectations
one might derive from a formal audit. These
features are consistent with the expectations
one might derive from a formal model. But
we also find a pecuIiar divergence from the
formal model here, in the example of an audi-
tor who was reprimanded for an aggressive
audit strategy that went beyond what was
“normally accepted”. As Goffman (1969)
notes, there are cultural constraints on the
moves in a particular situation. By overstep
ping normal expectations, revenue agents can
damage their ability to obtain the necessary
cooperation from taxpayers.
Research on t&e tax law. In more complex
audits, the agent may need to do research on
the applicable tax law. The tax code is the
body of formal knowledge required for the
job, and a large part of the training the IRS
provides revenue agents focuses on tax law
and legal research. Agents may request help
from managers, but are encouraged to work
alone. Legal research can be a significant task;
a typical tax library in an IRS district office fills
a rather large room. As with issue selection,
agents have some discretion over the depth
of the analysis.
Performing legal research diverges from the
basic face-to-face expression game by introdu-
cing “back-stage” activities (Goffman, 1959)
that occur during the course of the game.
This reinforces the observation that the audit
game can be played in multiple settings and by
multiple players. This is especially true in the
case of larger audits, where teams of auditors
are often involved. Nonetheless, legal research
is like an uncovering move, because it can pro
vide the revenue agent with justification for
reinterpreting the taxpayer’s actions or
records in a new light.
Computing the adjustments. Once the
appropriate data are assembled and the applic-
able tax law determined, the agent prepares
the proposed adjustments to the return. This
process may be tedious, especially for large,
complex cases like insurance companies. Prior
to 1988, when the IRS began to provide rev-
enue agents with portable computers (Pent-
land, 1989) all of the analysis would be done
on ledger pads with mechanical pencils, or on
“pro-forma” worksheets, where the appropri-
ate columns and factors were already laid out.
The resulting computation is written into the
“revenue agent Report” or RAIL In addition to
simply computing the tax, agents must also
document the process in their “workpapers”.
Workpapers show how each analysis was con-
ducted; for a smaller audit they may be thirty
pages, but for a large one they may be several
hundred pages.
282 B. T. PENTLAND and P. CARLILE
As the IRS began to automate the audit pro-
cess, revenue agents began to express concern
over the possibility that taxpayers (or the
accountants who represent them) would react
strategically. One agent likened the situation to
the French defenses against German invasion
prior to World War II:
Iftheexamin ation system was too standardized, then it
would be like the Maglnot line of software - the
accountant will go over it, around it, under it, any
way they can. Remember, accountants can get [the soft-
ware] through the freedom of information act, so they
can Bnd the best method of beating the system. It’s an
adversarial relation.
These comments suggest the use of counter-
uncovering moves on the part of taxpayers.
Knowing that the IRS looks for certain things
allows taxpayers to manipulate their transac-
tions and their records appropriately.
“Sel l i ng” the adj ustments. After computing
the new tax, interest, and penalties, the agent
must discuss the adjustments with the taxpayer
(or representative). The objective of this meet-
ing is to get the taxpayer to agree to the adjust-
ments, since the appeals process is expensive
and time consuming. As one manager pointed
out, “It’s important to settle the case at the
lowest level we can.” Taxpayers can contest
the results of an audit (resulting in a consider-
able amount of additional work for the revenue
agent). This leads agents to talk in terms of
“selling the adjustments” to the taxpayer,
rather than just making the adjustment.
In the typical audit scenario, the revenue
agent looks over the taxpayer’s records and
recomputes the amount of tax due. If this is
different than what the taxpayer computed,
the difference is called an “adjustment”.
Adjustments can go in either direction, and
amazing though it seems, a small percentage
of audits result in money being returned to
the taxpayer (Pentland, 1989). Of course, in
the more common case of an additional assess-
ment, taxpayers (and their representatives)
may be unhappy when presented with the
bill and may even become belligerent, as this
revenue agent describes:
There’s one accountant who was a screamer, one of the
worst. You’d go in them, and he’d yell, bang on his
desk, the whole works. I learned to just sit there and
smile real nice like this @utting on a huge, gratuitous
grin) the whole time. When he got tired of screaming,
he’d say, “Hey, you’re always so nice, you sit there and
smile.” And then he’s sign the adjustments. It worked
like charm.
Patience and a pleasant demeanor are an
important part of selling the adjustments. But
getting a taxpayer to agree to a revised tax
assessment also depends on establishing cred-
ibility as a fair, competent auditor, as this man-
ager explained:
You see, when you’re out there doing an audit and it
comes time to close the case, what you are really doing
it selling your skills and image as a credible, competent
examiner. Would you sign an agreement form prepared
by someone you didn’t trust to do it right? Probably not.
In addition to seeming competent, it is criti-
cal that the revenue agent be perceived as fair.
This is important not only for getting agree-
ment in the current audit, but because an audit
that is perceived as punitive might cause a tax-
payer to cheat in future years, undermining the
objective of voluntary compliance:
We want to be perceived as being fair. We want to turn
bad taxpayers into good ones. On one case there were
so many things wrong that they would never agree to
the adjustments, and they’d hate us forever. So, I just
stopped looking, wrapped things up, and suggested
they keep better records in the future.
Once again, we see that auditors sometimes
overlook issues in the interest of closing cases.
To the extent that they do so, they are collud-
ing with the taxpayer’s claim to be honest and
compliant, thereby helping the taxpayer to
save face in the interaction. This kind of beha-
vior is a natural part of an expression game
(Goffman, 1967).
The other practical reason to strive for tax-
payer agreement is that “unagreed cases”
involve a lot of additional work for the rev-
enue agent and additional expense for IRS.
The RAR and workpapers are much longer,
and there may be follow-up work needed to
TAX AUDITING 283
support the appeals process. In general, agents
prefer closing the case with taxpayer agree-
ment as early as possible, because it saves
both the IRS and the agent’s time, prevents
escalation to other IRS functions, and demon-
strates the agents competence and skill in
meeting and dealing with the taxpayer. For
these practical reasons, agreements are better
than non-agreements.
It should be clear from these examples that
successful completion of an audit requires a
great deal of impression management on the
part of the revenue agent. The revenue agent
must maintain an image of fairness and compe-
tence or the taxpayer will be reluctant to agree
to the results and the audit cannot be com-
pleted. Hence, the revenue agent is engaging
in a simultaneous expression game with the
taxpayer. Because the situation is usually
framed so strongly as “scrutinizing the tax-
payer” the reverse dynamics can be difficult
to tease out, but they are there nonetheless.
Wrappi ng up the casefi l e. The Iinal product
of the revenue agent’s work is the case file,
which consists of a summary sheet (called a
4318), the workpapers, the RAR, and any sup
porting documents. This case tile is turned in to
the group manager for review and processing.
Once the case tile is complete, it may be
reviewed for completeness and quality. Man-
agers review a certain percentage of cases,
and a specialized review staff also looks at a
sample. If a case is sent back, it may involve
recontacting the taxpayer, and it takes a lot of
extm time. If the case passes review (and
doesn’t involve an appeal or other subsequent
litigation), then the revenue agent’s job is
complete.
Once again, we find an example of a move
that is peculiar to the audit situation. By giving
his or her consent to a revised tax assessment,
the taxpayer clears the way for a whole new set
of IRS representatives to take over the case: the
Collection Division. While an analysis of the
activities of the Collection Division is beyond
the scope of this paper, suffice it to say that the
rules of that game are very different. It is no
longer an expression game, because the facts
are no longer in contention. Once an assess-
ment is made and deemed valid, the IRS has
very broad powers to collect what they are
owed. The game is essentially over.
DISCUSSION
This analysis suggests that both taxpayers
and revenue agents are engaged in impression
management that is vital for the completion of
the audit. There are at least two games going
on at once because each party to the audit is
attempting to maintain a front (Goffman,
1959). The payoffs in these games are not
merely economic, because, in addition to the
facts of the case, the identity of the participants
is at stake. Revenue agents need to live with
themselves and keep their jobs, which means
that they cannot allow their identity as fair,
competent individuals to be undermined. Tax-
payers, whether honest or not, have similar
needs that extend beyond their pecuniary inter-
est in the outcome of the tax computation. If
we accept these objectives as a significant part
of the game, it is difficult to imagine computing
an optimal strategy for either party. The moves
in an expression game are difficult to predict
and the payoff from any given move is difficult
to compute. To reduce the complexity of the
analysis, one could ignore the objectives of the
participants, but doing so would lead us to
overlook the process through which audit
results are actually arrived at. Contrary to the
simplifying assumption needed to make formal
games tractable, the selection of a return to be
audited is where the real audit game begins. In
addition to providing a rich vocabulary in
which to describe the dynamics of the audit
game, the expression game framework pro
vides insight into some additional issues that
we would like to turn to at this point.
Rati onal i zati on of audi t work through
formal rul es
In order to preserve the appearance of objec-
tivity and fairness, the IRS is rule dominated,
rationalistic and aggressively self-monitoring.
284 B. T. PENTLAND and P. CARLILE
The prescribed role of tax collecting involves
the objective appllcatlon of rules to the facts of
the case. These rules and the methods for their
appllcatlon comprise the formal knowledge or
by-the-book procedures to collect taxes. The
formalization of rules and procedures is a nat-
ural outgrowth of the IRS peculiar position in
society and its status as an enormous bureau-
ClXCy.
The data presented here suggest that these
formal rules and their application are only part
of the knowledge required to complete an
audit. The expression game framework helps
draw attention to the practice-based or work-
ing knowledge that revenue agents draw upon
in the material and social constraints of the
audit situation. In this respect, it is similar to
other perspectives that stress the importance
of analyzing actual practices to explain how
knowledge is structured and used (Bourdieu,
1977, 1990; Giddens, 1984; Lave, 1988; Brown
& Duguid, 1991; Pentland, 1992). Expression
games reflect the material, social and historical
influences - social praxis - that generate and
sustain action and competent behavior (Bour-
dieu & Wacquant, 1992, p. 15).
As expression games, audits embody a con-
test over the facts; until this contest is resolved,
application of formal rules is impossible. As we
have seen, this contest over the facts takes
place in a situation which may encourage rev-
enue agents to overlook formal rules, which
they rationalize by saying that they are trying
to “encourage voluntary compliance” in subse-
quent years (i.e. turn bad taxpayers into good
ones). The informal knowledge governing the
decision of “where to draw the line” is gener-
ated and embodied in the practice of tax audit-
ing, rather than being prescribed by rules
contained in the formal or canonical proce-
dures of the IRS. Likewise, the particular
moves used by revenue agents and taxpayers
are constrained and enabled by the audit
situation.
The practical constraints of time, resources,
information asymmetries, and power all help to
create a situation where the best outcome for
the revenue agent, as well as the IRS, 1s to let
people walk away from some portion of their
tax burden even after an audlt. While thls is
surely not part of the Internal Revenue Code,
it is seen by revenue agents as a way of getting
the job done, and hopefully, encouraging
voluntary compliance ln the future. If the IRS
were not under pressure to perform efficiently,
auditors might make different trade-offs in their
pursuit of adjustments. In particular, they
would be less concerned with delaying tac-
tics, since there would presumably be no
need to close cases quickly. Nonetheless,
even under conditions of unlimited resources,
they would be subject to information asymme-
trles and the need to enlist taxpayer coopera-
tion to accomplish their goal. This outcome
corresponds, in qualitative terms, to Becker’s
(1968) normative analysis of “optimal” enfor-
cement levels, but there is no claim of optim-
ality. It is simply what is done.
The expression game framework offers an
interesting alternative to the problem of objec-
tivity in auditing, as well. In much of the litera-
ture on auditing, there is an opposition
between objectivity and subjectivity (e.g.
Sylph, 1988). Objectivity is achieved through
the uniform application of formal accounting
rules. Subjectivity is conceived of as individual
error or bias in the application of those rules
(Plumlee, 1985). But strictly speaking, the
results of a tax audit are neither objective or
subjective. Rather, as the product of an expres
sion game, the revised tax assessment is merely
good enough for all practical purposes. The
interpretation and application of the tax code
is highly contingent upon the character of the
interaction between agent and taxpayer; the
facts that are negotiated subject to the practical
constraints that exist. The tax audit interaction
generates a pragmatic sculpture that is unique
to the properties and constraints of each audit.
Practi cal constrai nts: worki ng knowl edge
and formal knowl edge
Perhaps the most critical issue raised in this
paper concerns our understanding of the role
TAX AUDITING 285
of knowledge, and what counts as knowl-
edge,
in this much feared occupational
group. From one perspective, (that of the
national office in Washington), revenue
agents are engaged in the bureaucratic appli-
cation of the tax code to a sample of tax
returns, according to procedures spelled out
in the IRS Manual. In Orr’s (1990) terms, rev-
enue agents are responsible for applying a
body of “canonical knowledge” to each
return they audit. The National Office (and
IRS management in general) places a great
deal of emphasis on following the rules
because of the necessity of malntaining the
Impression of fairness and impartiality.
As we have seen, however, the practice of
tax auditing draws on the tax code as a
resource, but is largely constituted out of a
series of practical contingencies that would
remain largely the same regardless of the spe-
clfic formulation of the tax code. Even if all
deductions were ellmlnated, there would still
be the matter of determining income; waiters
and taxl-drlvers could still skim their tips. The
motto of the field audltor tells us where the
action is: “audit the taxpayer, not the return.”
If a revenue agent sticks to the numbers on the
page, he or she will likely miss the really big
issues, like the sailboat sitting ln the driveway.
The real tax audit game involves knowing how
to find issues, and that is not part of the cano-
nical knowledge base as embodied in the tax
code. Nor is it part of the standard game-
theoretic treatment of auditing, which
assumes that auditing produces accurate
results (Graetz et al ., 1986).
Bather, the tax audit game is a product of
the situation in which taxpayers and revenue
agents Iind themselves. Taxpayers generally
know where they have cut a few comers,
and they almost certainly know when they
have committed fraud. But the auditor does
not; he or she can only suspect problems
and then attempt to ferret them out. As a
result, the outcome of an audit cannot be
determined by simply applying the tax code
to the “facts” because the facts are the
product of the expression game. And because
auditors and taxpayers differ in their skill,
resources, and perseverance at this game, the
outcome of any particular expression game is
uncertain. The taxpayer may or may not sustain
the impression of credibility. The auditor may
or may not sustain the impression of fairness
and competence.
CONCLUSION
Auditing is a classic example of an expres-
sion game. The auditor and the taxpayer are
locked in a contest over the “truth” of the
taxpayer’s return, the outcome of which has
considerable consequences for both parties
and the public at large. Thls is, indeed, where
the rubber hits the road. Most analyses of this
important problem, however, have hovered
safely above the ground, thereby avoiding the
complexity that lies ln the details. Formal eco-
nomic analyses, in particular, assume away the
very contest that makes up the game. The sim-
plifications needed to “solve” games of this
klnd by computing normatlvely optimal strate-
gies would seem to obscure our understanding
of how the game is played, what moves are
possible, and what the practical outcomes
might be.
Tax auditing is not the only example of an
expression game, of course, as Goffman’s
(1967, 1969) analyses make clear. Conse-
quently, many of the observations we have
made here are applicable to any situation
where facts are contested and information is
asymmetric. Any situation that can be inter-
preted in principle-agent terms, for example,
has this quality. In short, expression games
are ubiquitous in organizational life and in orga-
nizational research. Owners, managers, audi-
tors, customers, and researchers are all in the
position of searching for “facts” that are, at
best, the products of expression games. What
our analysis suggests, we think, is that the
“facts” are considerably more tenuous in these
situations than we would like to believe. In
286 B. T. PENTLAND and P. CARLILE
these situations, the ideals of objectivity and practicality and informal judgments. Nothing
formal rules must be set aside in favor of more is possible, and nothing less will do.
BIBLIOGRAPHY
Allingham, M. G. & Sandmo, A., Income Tax Evasion: A Theoretical Analysis, J ournal of Publ i c Econom-
i cs (1972) pp. 323-338.
Becker, G. S., Crime and Punishment: An Economic Approach, The J ournal of Pol i ti cal Economy
(February 1968) pp. 169-217.
Bourdieu, P. & Wacquant, L., An I nui tati on to Refexi ve Soci ol ogy (Chi cago, IL: The University of
Chicago Press, 1992).
Bourdieu, P., Outl fne of a Theory of Practi ce (Cambridge, U.K.: Cambridge University Press, 1977).
Bourdieu, P., The Logfc of Practi ce (Cambridge, U.K.: Cambridge University Press, 19%).
Brown, J, S. & Duguid, P., Organizational Learning and Communitiesof-Practice, Organization Scfence
@umary 1991) pp. 40-57.
Carroll, J. S., Compliance with the Law: a Decision-making Approach to Taxpaying, Law & Human
Eebavfor (December 1987) pp. 319-335.
Chommie, J. C., The I nternal Revenue Servi ce (New York Praeger, 1970).
Diogenes, The Apri l Game: Secrets of an I nternal revenue agent (Chicago: Playboy Press, 1973).
Fellingham, J. C. & Newman, D. P., Strategic Considerations in Auditing, The Accounting Revi ew (1985)
pp. 634-650.
Giddens, A., me Consti tutfon of Socfev (Berkeley, CA: University of California Press, 1984).
Goffman, E., Asyl ums Essays on tbe Soci al Si tuati on of Mental Pati ents and other I nmates (Garden
City, NY: Doubleday, 1961).
Goffman, E., InteracNon Ri tual (New York: Pantheon Books, 1967).
Goffman, E., Strategi c I nteracti on (Phildelphia, PA: University of Pennsylvania Press, 1%9).
Goffman, E., 76e Presentati on of Serf in Every&y Li fe (Garden City, NY: Doubleday, 1959).
Graetz, M. J., Reinganum, J. &Wilde, L. L., The Tax Compliance Game: Toward an Interactive Theory of
Law Enforcement, J ournal of Law, Economi cs, and Organi zati on, I I , (1986) pp. l-32.
J ournal of Accountancy, Auditing interpretation: Quantifying Departures from Generally Accepted
Accounting Principles (April 1986) pp. 156-157.
Klepper, S. & Nagin, D., Tax Compliance and Perceptions of the Risks of Detection and Criminal
Prosecution, Law C Soci ety Revi ew (1989) pp. 209-240.
Kusterer, K. C., Knot&bow on the J ob: The I mportant Worki ng Knowl edge of “Unski l l ed” Workers
(Boulder, CO: Westview Press, 1978).
lave, J., Cognftfon in Practice: Mi nd, Mathemati cs, and Cul ture i n Everyday Li fe (Cambridge, U.K.:
Cambridge University Press, 1988).
Lohse, D., Tips to Follow for Preventing an lRS Audit, Wal l Street J ournal (21 February 1995)
p. c-1.
Mead, G. H., Mi nd, Self and Soci ety, Morris, C. (ed.) Chicago: University of Chicago Press, 1934).
Morton, S., Strategic Auditing for Fraud, me Accounti ng Revi ew (1993) pp. 825-839.
Orr, J., Tal ki ng about Machi nes: An Etbnograpby of a Modern J ob, Unpublished PhD dissertation,
Cornell University, 1990.
Pentland, B. T., Organizing Moves in Software Support Hotlines, Admfni stratfve Sci ence Quarterl y
(1992) pp. 527-548.
Pentland, B. T., Use and Productivity in Personal Computing: An Empirical Test, Proceedfngs of tbe
I nternati onal Conference on I nformati on Systems (Boston, MA: ICIS, 1989) pp. 211-222.
Plumlee, D., The Standard of Objectivity for Internal Auditors: Memory and Bias Effects. J ournal of
Accounting Research (1985) pp. 683-699.
Reinganum, J. & Wilde, L. L., A Note on Enforcement Uncertainty and Taxpayer Compliance, The
Quarterly J ournal of Economi cs (1988) pp. 793-798.
TAX AUDITING 287
Reinganum, J. % Wilde, L. L., Income Tax Compliance in a Principal-Agent Framework,Journul of Publ i c
Economi cs (February 1985) pp. 1-18.
Sanders, L., Yellow L@hts and Red Lights, F&es (27 March 1995) pp. 92-93.
SingletonGreen, B., In Pursuit of the Unfudgeable, Accountancy (November 1992) pp. 110-123.
Surface, W., Inside Intwnal Remme (New York: Coward-McCann, 1%7).
Sylph, J., Has the Objectivity Question Clouded the Real Issues? CA Mugazf~e (August 1988) p.
121.
Watterson, T., More People Fear Tax Audits Than Get Them, but That May Change, & Cbrfsffan Sci ence
Monftor (3 Au8ust 1988) p. 12.

doc_680254556.pdf