Auditing and the production of legitimacyThis essay discusses an important series of forma

Description
This essay discusses an important series of formative contributions to contextualist and critical research in auditing.
A small number of relatively recent papers question rationalized accounts of the audit process and explore the complex
‘backstage’ of practice. These papers are interpreted as contributions to our understanding of the production of
legitimacy around four substantive themes: the audit process and formal structure; auditing as a business; working
papers and image management; new audits.

Auditing and the production of legitimacy
Michael K. Power
Department of Accounting & Finance and ESRC Centre for Analysis of Risk and Regulation,
London School of Economics and Political Science, London WC2A 2AE, UK
Abstract
This essay discusses an important series of formative contributions to contextualist and critical research in auditing.
A small number of relatively recent papers question rationalized accounts of the audit process and explore the complex
‘back stage’ of practice. These papers are interpreted as contributions to our understanding of the production of
legitimacy around four substantive themes: the audit process and formal structure; auditing as a business; working
papers and image management; new audits. The papers also point to the socially constructed nature of professional
inference and suggest a fruitful basis for taking these research e?orts forward.
#2003 Elsevier Science Ltd. All rights reserved.
That accounting is more than a neutral techni-
cal practice is well established; it shapes pre-
ferences, organizational routines, and the forms of
visibility, which support and give meaning to
decision making. There is much empirical work to
support this view and the related methodological
imperative to study accounting practice ‘in con-
text’ has even become something of an orthodoxy.
An important focus of this work is the way that
accounting systems in their broadest sense func-
tion often more to legitimate individual and orga-
nizational behaviour than to support e?cient and
rational decision making.
For all the strengths and very evident in?uence
of this line of thinking, most auditing research has
been relatively untouched by it. While a great deal
is now understood about the actual functioning of
budgeting and control systems, standard costing,
ABC and other aspects of the management
accounting ?eld, very little is known about audit-
ing in practical, as opposed to experimental, set-
tings. Accordingly, it is appropriate to explore the
modest but important extension of the ‘con-
textualist’ and, one might say, ‘critical’ imagina-
tion to the ?eld of ?nancial auditing and to the
role of auditing in ‘producing legitimacy’. Such a
theme is particularly pertinent in the early 21st
century when the legitimacy of auditing is under
?re as perhaps never before, and when the much
disputed boundary between auditing and consult-
ing seems as blurred and unclear as ever (Jeppe-
sen, 1998; Power, 2000). The body of qualitative
work to be discussed is small and relatively recent
(nearly all the papers discussed here were pub-
lished in the 1990s) compared to the management
accounting corpus, but it is no less signi?cant for
that. Taken together, the papers represent an
emerging e?ort to question rationalized accounts
of the audit judgement process, and to explore the
0361-3682/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
PI I : S0361- 3682( 01) 00047- 2
Accounting, Organizations and Society 28 (2003) 379–394
www.elsevier.com/locate/aos
E-mail address: [email protected] (M.K. Power).
complex ‘back stage’ of practice in its social and
organizational context.
1
There is very little of what is now called ‘?eld
work’ in auditing. The apparent reason for this is
that professional service ?rms are reluctant to
provide research access to client data and to live
audit assignments, although this is more often
assumed than demonstrated. A close examination
of many of the studies discussed below reveals that
they have usually been conducted in the margins
of more orthodox work, research by stealth with
all the ethical issues that this raises. But the issue is
also in part one of motivation for the scienti?c
community; auditing research is dominated by the
experimental psychology tradition and, to a lesser
extent, analytical economics. It follows that ?eld
studies of auditing may be risky for research
careers regardless of the question of access. To say
this is to say that the studies discussed below are
‘brave’ studies, conducted largely against the
intellectual tide of mainstream approaches.
2
This
should always be borne in mind when considering
their apparent methodological weaknesses. For
example, some of the research is nascent anthro-
pology and, although it would unfair to judge it by
the standards of that discipline, there is evidence
of ethnographic aspirations, which are possible in
principle but severely limited in practice.
The arguments that follow attempt to explicate
a guiding theme, an unconscious agenda perhaps,
which links these papers together as an emerging
but incomplete programme. This is a programme
which is also being taken up outside the account-
ing ?eld (e.g. Strathern, 2000). The theme in
question is ‘auditing and the production of legiti-
macy’. It can be maintained that ?nancial auditing
is always engaged in a process of self-reproduction
largely because, in order to generate trust in
?nancial statements, audit practice must generate
trust in itself. Furthermore, as we shall see below,
the production of legitimacy is an intimate part of
an apparently ‘technical’ audit judgement process.
In short, this essay suggests that the legitimacy of
both auditor and auditee are co-produced.
We might begin this inquiry by asking a funda-
mental question: ‘what do audits produce?’ The
conventional answer to be found in professional
guidance is that they produce assurance or
increased con?dence in the subject matter of audit.
Thus, ?nancial statements are regarded as more
reliable than they would be without an audit and,
so the story goes, this improves the e?ciency of
capital markets. However, a more mischievous
answer to this question is to say that audits pro-
duce paper, and quite a lot of it, although, with
the advent of information technology much less than
they used to do. Indeed, the fact that audits produce
(at least in principle some) paper is central to the
theme of legitimacy production in this essay.
In the next four sections, I provide a critical re-
reading of a small number of auditing papers in
terms of their contribution to our understanding
of the production of legitimacy around four sub-
stantive themes: the audit process and formal
structure; auditing as a business; working papers
and image management; new audits. This is fol-
lowed by a discussion of the socially constructed
nature of professional inference in settings such as
auditing. Third, some methodological re?ections
on the nexus between controversy, closure and
credibility are developed which may be relevant to
the continuation of this research programme, and
which draw in themes from the sociology of
science and technology more generally.
1. Structure, judgement and the legitimacy of the
audit process
Despite programmes by ?rms and professional
institutes to standardize the audit process, di?er-
ences in the style and application of audit routines
have been evident. Early research to characterize
and explain these di?erences (e.g. Dirsmith &
McAllister, 1982) crystallises the problem in terms
of the di?erence between highly ‘structured’,
formal approaches, and those that provide more
1
No principle of selection is perfect and my choice of
papers within this review may be contested at the margins.
However, this essay is concerned thematically more with the
level of the audit process rather than with that of the auditing
profession despite, as Andrew Abbott (1988) has shown, the
important relations between these two levels understood as task
and jurisdiction, respectively.
2
For further re?ection on the politics of auditing research,
see Gendron and Bedard (2001) and Humphrey (2001).
380 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
space for individual judgement. Cushing and Loeb-
bekke (1986) provide a content survey of audit
approaches derived from the formal manuals of 12
audit ?rms. Based on this analysis of the technical
guidance, they classify ?rms according to whether
they adopt a structured methodology or not.
The structure–judgement debate in auditing
re?ected, and still re?ects, practitioner concerns
with the optimal design of technical procedures.
Dirsmith and Haskins (1991) build on Dirsmith’s
earlier work in the 1980s and on organization the-
ory to cast the debate in terms of the deep meta-
phors that organize the audit process, ways of seeing
or constructs which in?uence evidence gathering. In
e?ect, they argue that audit is not a naturally coor-
dinated series of technical steps; it is a social enter-
prise relying on deeply embedded perspectives.
Dirsmith and Haskins (1991) suggest that
underlying the structure–judgement distinction is
a more fundamental one between mechanism and
organism as root metaphors of the audit process.
‘Mechanism’ names an aspiration for an inte-
grative formal approach to audit which holds out
the promise of an algorithmic knowledge base.
‘Organism’ assumes that the whole is always
greater than the parts, and that the speci?city of
knowledge places limits on the mechanistic world
view. Dirsmith and Haskins test this conceptual
distinction in the context of inherent risk assess-
ment. Their study is based on the content analysis
of public and proprietorial materials in a number
of ?rms, and uses interviews and a survey instru-
ment comprising groupings of factors relevant for
assessing inherent risk. The analysis con?rms an
association between general audit style and the
factors which ?rms consider to be signi?cant in
inherent risk assessment. In short, ?rms with
unstructured (organic) audit approaches tend to
be more wide ranging in the factors used to make
inherent risk assessments.
3
While Dirsmith and Haskins (1991) make a
normative case for the superiority of the organic,
unstructured audit style, it is also clear that both
structured and unstructured audit approaches are
problematic ideals or programmes (Rose & Miller,
1992) which can never be fully or perfectly rea-
lized. The tension between structure and judge-
ment is multifaceted, and the predominance of
relatively more structured audit approaches
re?ects the increasing demand for legitimate and
transparent forms of standardized practice for
management control purposes. Idiosyncratic,
local, and intuitive judgements by seasoned prac-
titioners are di?cult to manage from the centre of
the ?rm. In short, structure is about legitimacy
and control, which is not necessarily consistent
with better or more e?cient auditing.
The normative case for ‘seasoned judgement’ in
the audit process is also taken up by Francis
(1994). According to him, nothing less is at stake
in the judgement–structure debate than the
‘deformity of practical reasoning’ by a form of
scientism. Francis takes issue with Cushing and
Loebbekke (1986) as partisans of structure in the
service of e?ciency and litigation protection. The
‘scientistic’ assumptions behind structured audit-
ing idealize the audit process as a logical series of
steps which can be encoded in algorithmic decision
aids. In this way, structure represents and models
the ‘universal’ audit as a system of rules. However,
these pre-determined representations of the audit
process can never really inform the auditor what
to do. For example, Francis argues that the risk
model is an empty abstraction which represents a
loss of practical reasoning and a corruption of the
audit judgement process. Similarly, Fogarty (1996)
points to the colonizing potential of structure and
formal systems, which search for ‘new areas of
human vitality to codify’.
A technical area which focuses the judgement–
structure discussion is that of statistical sampling.
Power (1992) argues that the brute economics of
auditing forced auditors to be selective before the
advent of statistical sampling ideas in the ?eld.
Practices such as amateur block testing methods
came to be rationalized and legitimized by the
language of statistical sampling, thereby making a
virtue of economic necessity and serving to align
3
In recent years, the need for wider ranging risk assessment
within the audit process and other practices has become generally
accepted (COSO, 1991; ICAEW, 1999) and the current fashion
for holistic risk management supports the ?ndings of Dirsmith
and Hasksins (1991), despite the methodological issues that their
study may have raised. And yet while the post-COSO world takes
a much broader view of inherent risk, there is equally a demand
for greater formalism in auditing and risk management.
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 381
auditing with legitimate ideals of science. Carpen-
ter and Dirsmith (1993) reinforce these conclu-
sions and argue that statistical sampling was much
more than a means for greater audit e?ciency. It
also served an important legitimizing role for a
practice increasingly self-conscious about the need
for abstract foundations and a ?rm pedagogical
basis.
Carpenter and Dirsmith (1993) argue that sam-
pling emerged from an economically driven rede-
?nition of the objectives of auditing, away from
fraud detection and towards the idea of ‘examina-
tion’. The associated rise of the concept of internal
checking and control represented a new stage in
the relationship between internal and external
monitoring, one in which senior management
responsibilities for systems of control could be
articulated. In addition, sampling helped to align
auditing with a broader culture of scientism and
formalism in the 1930s and 1940s.
Accordingly, statistical sampling in particular,
and formal structure in general, are more than
simply technical features of audit practice. They
are social phenomena, part of what Carpenter and
Dirsmith (1993), borrowing from Abbott (1988),
describe as the regeneration and modi?cation of
the audit profession’s abstract system of knowl-
edge. Even in its deepest assumptions, sampling
represents and constructs the ‘old economy’ ideal
of the mass transaction, low error bureaucratic
organization (p. 55). And just as sampling makes
the client organization ‘knowable’ and ‘auditable’
(Power, 1996), it functions simultaneously as the
basis for making visible the decisions of the prac-
titioner. Carpenter and Dirsmith (1993) suggest
that the formality of statistical sampling functions,
via documentation, to transfer power from practi-
tioners to the administrative centres of the large
?rms, and to legitimate practice with a more aca-
demic basis. Equally, statistical sampling as a
technique also disturbs the web of social expecta-
tions that surround the audit process, making it
contestable in new ways and fuelling the so-called
expectations gap.
The structure–judgement problem was, and
remains, the academic articulation of an issue
which goes to the very heart of professional prac-
tice and identity (see also Carpenter et al., 1994).
The large ?rms continually adjust their audit pro-
cesses and try to balance a formal, defendable,
economic and manageable structure for the audit
process with the autonomy of auditor judgement.
The trend towards greater structure may re?ect
the intensi?ed need for ?rms to control the quality
of what they do and provides a partial solution to
an agency problem within ?rms which threatens
their economic survival. In short, the trend for
greater structure accompanies the explicit devel-
opment of auditing as a business.
2. The business of auditing
Auditing has always been a business. However,
during the 1980s and 1990s the audit approaches
of the large ?rms changed and evolved as the eco-
nomics of auditing became more sensitive.
Accordingly, the structure–judgement issue dis-
cussed above in terms of the managerial relations
between the audit ?rm and its own members, is
also in part about cost control and the manage-
ment of audit as a business activity. McNair’s
(1991) study documents the ‘suppressed dilemma’
between cost and quality objectives which push in
di?erent directions. Individual auditors are forced
to internalize and resolve this dilemma through
time budgeting and reporting processes. Accord-
ing to McNair, auditors must learn the ‘business’
of auditing by trading cost and quality in a zone
that is rendered undiscussable, and which leads to
time under-recording. This makes auditing pro-
foundly ambivalent because the acute compro-
mises that the auditor is forced to make as an
individual are rendered invisible. The cost–quality
dilemma becomes embedded in the management
control system where all the pressures for change
are economic in form, and where the internal
legitimacy of auditors is framed in terms of sup-
porting the business of auditing.
The business imperative is also visible in the
study by Humphrey and Moizer (1990) who
interviewed practitioners about the audit planning
process. Their analysis reveals the re-fashioning of
a regulatory imperative into one of client service.
Audit planning serves technical (traditional),
ideological (legitimating) and marketing (selling)
382 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
roles simultaneously. Although planning is the
touchtone of rationality for every practical dis-
cipline, Humphrey and Moizer (1990) observe
how the fundamental issue of the su?ciency of
work is resolved in situ. The study is strongly cri-
tical of the gap between o?cial views of the audit
process and what happens in practice. For exam-
ple, they observe that o?cial guidance plays little
role in everyday audit work, but rather it legit-
imizes work in the event of disputes with clients.
In addition, sampling and con?dence levels are
often acts of faith within a crudely mechanical and
bureaucratic process. Indeed, sta? often would not
follow sampling plans correctly. Anchoring and
the search for con?rmatory evidence are also visi-
ble, a ?nding which is interestingly consistent with
a number of experimental cognitive studies of the
audit process. Overall, Humphrey and Moizer
(1990) regard the formal structure of the audit
planning process as something that is not used
other than to legitimate the authority of the audi-
tor. In-house audit manuals present a loose fra-
mework for the exercise of judgement.
In another study concerning changes to the
technical basis of audit, Fischer (1996) begins with
the question as to how audit ‘e?ciencies’ are pro-
duced at the level of the ?rm. Fischer (1996)
gathered data over a one year period and con-
ducted interviews on the back of access granted in
the context of an education programme. The
organizational environment is one where there are
pressures for cost reduction and where quality is
di?cult to observe, even for practitioners them-
selves. This means that ‘adequate’ auditing must
be constructed as part of a community project, the
community in Fischer’s (1996) case being the ?rm
itself. As in the Dirsmith and Haskins (1991)
study, there is clearly an agency problem for large
?rms wishing to control practice.
In Fischer’s (1996) study, the concern at senior
levels within the ?rm is with the problem of two
audits, one dealing with documentation and form
?lling (compliance with the traditional structured
approach) and the other more risk-oriented, oper-
ated by higher level partners (judgement sup-
ported by new technology). The perceived
problem is that sta? who are trained and inducted
into the ?rst approach are under-trained in the
latter. Accordingly, the new audit approach and
technology was initially simply added to the old,
rather than substituted for it: in the context of a
change programme the new approach was not
regarded as adequate or legitimate.
Fischer observes that the ?rm is particularly
concerned that sta? are over-focused on doc-
umentation in addition to the obvious expense of
‘two’ audits. The new technology is an attempt to
remind sta? of the importance of judgement.
However, interviews reveal a lack of in-depth
understanding of technical concepts and assump-
tions embedded in the new procedures and practi-
tioners tended to refer mechanically to the ?rm’s
approach rather than to the principles embedded
in GAAS [see also Humphrey & Moizer (1990) for
a similar view]. The required new approach
involved a computerized decision aid developed by
the national o?ce, which was intended to enhance
consensus in auditor judgement. Yet, in the words
of one partner: ‘‘its all a way of institutionalizing
less work under GAAS’’. Sold internally as a ‘re-
emphasis’ of the old approach, Fischer suggests
that actual adoption was not because it was per-
ceived as better, but because of central diktat.
The essence of the problem in Fischer’s (1996)
analysis is that the cost e?ciencies of the new
approach could not be realized until it was accep-
ted by practitioners as providing su?cient evi-
dence. In other words, techniques must be
legitimate before they can be e?cient, although we
usually think that it is the other way around, i.e.
that their technical e?ciency supports their legiti-
macy. According to Fischer (1996), the unknow-
ability of audit quality means that practice is a
social construction, i.e. that practitioner consensus
is essential before something can count as legit-
imate knowledge in the professional system. The
process of technical change requires that a new
technology must be ‘realized’, i.e. made real by the
action of the actors. As Fischer puts it, ‘‘the audi-
tor creates the bene?ts, in the form of reduced
audit costs, from using the new technology—by
reducing or eliminating procedures that had been
performed in the past, concurrent with the adop-
tion of the new technology’’.
Fischer’s (1996) analysis shows that change, and
the reinvention of the audit process, depend on a
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 383
micro-politics of legitimizing change for practi-
tioners. Old approaches must be discredited at the
same time as new approaches must be seen as
adequate by overcoming practitioner anchoring.
In this case, it is clear that the business side of
auditing drives technical innovation, but a great
deal of work must then be done in large ?rms to
convince practitioners that the technical innova-
tion works. The risk model, which legitimated
sample size reduction, itself became subject to cri-
tique and the perception that ‘over-auditing’,
simultaneously a business and epistemic kind of
failure, was persisting.
Innovation may also be less radical than is ima-
gined because of this micro-politics. According to
Humphrey and Moizer (1990), the much vaunted
risk based approach to auditing was more loosely
coupled to practice than commonly thought: risk
is more of a buzzword rather than an integrated
part of the planning process committed to the
‘thinking’ audit. Although, auditing methodology
has changed even further since their ?eld study,
Humphrey and Moizer (1990), like Fischer (1996),
see the revised planning process as a tool for cut-
ting work down in the name of e?ciency. How-
ever, the risk-based reinvention of the audit
process is ‘‘altering the traditional compliance
based nature of the audit function’’ which is no
longer concerned with credibility assessment but
with a more integrated advisory role. Planning is
not just a basis for conducting the audit but also a
platform for other services. The audit is no longer
about ?nding control errors; legitimacy with cli-
ents is essential.
Developments in audit methodologies, (e.g. Bell,
Marrs, Soloman, & Thomas, 1997) suggest that
pure planning is a myth and that the boundaries
between ‘audit planning’ and ‘knowledge acquisi-
tion’ as a basis for cross-selling other assurance
and risk management advisory services is no
longer a clear one. In the light of these develop-
ments, consulting is being designed into the audit
process itself. It is not simply that there are ‘addi-
tional’ services whose implications for indepen-
dence can be debated, but that the audit/
consulting distinction has become operationally
unclear (Jeppesen, 1998; Power, 2000). Or, as
Humphrey and Moizer (1990) observe, in the
audit planning process technique and selling go
hand in hand.
Both Fischer (1996) and McNair (1991) give the
impression that the knowledge base of auditing is
malleable according to economic imperatives, not
simply to reduce audit costs but also to provide
extra services as part of the audit planning pro-
cess. However, the economic di?erence between
the risk model and developments by large ?rms to
reengineer the audit process around ‘strategic’
modeling (Bell et al., 1997) is this: whereas the
former supports cost reduction, the latter is pre-
mised on costly investments to justify fee increa-
ses. In short, in the business of auditing a cost
emphasis is being replaced by a revenue empha-
sis.
4
Nevertheless, the general point of the two
earlier studies remains relevant. Both bring out the
complex processes of legitimization in play, the
one by requiring auditors to suppress genuinely
de-stabilizing dilemmas, the other by getting
auditors to accept new cost- and revenue-e?ective
ways of doing the audit. And if knowledge acqui-
sition, audit planning and cross-selling are di?cult
to disentangle, it is clear that the audit process
must involve elaborate forms of image manage-
ment. Auditing is an expressive, as much as a
cognitive, process.
3. Image management and working papers
Pentland’s (1993) study provides an important
contribution to our understanding of the social
construction of the audit process. In essence, the
argument is that macro-constructions of pro-
fessionalism, auditor independence, and institu-
tional trust in audit practice are accomplished,
built up from, and reproduced by micro-interac-
tions. As a team-based activity comprising ongo-
ing interactions, auditing is a collective practice
with important processual components. The study
focuses on ‘‘interaction rituals’’ (Collins, 1981) as
the foundation of the production of orderly prac-
tice, and of the social trust which grounds all
contractual activity.
4
I owe this to Ira Solomon.
384 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
To say that auditing is ritualistic is to argue that
it is a collective activity bringing order to an
uninterpreted world; it must also be understood as
an activity with signi?cance beyond the manifest
(programmatic) rational function attributed to it
in o?cial documents. In suggesting an a?ective or
emotional foundation for audit practice, Pentland
(1993) opposes an expressive model of human
(auditing) agency to that of cool deliberative cal-
culation. (The contrast with audit judgement
research could not be greater.) On the basis of this
theory and his empirical work, Pentland (1993)
argues that auditing is a process of giving ‘com-
fort’ at the micro-level, which in turn makes pos-
sible its macro systemic function of legitimation.
The micro-rituals of audit practice enable audit
team solidarity, impression management and the
progressive puri?cation of the mess of practice for
outside consumption; audit is a process of pur-
i?cation culminating mainly in the ‘clean’ audit
opinion for outside consumption. The speci?city
of organizational ?nancial systems is ?ltered
through the lens of standard techniques. Only as
organizational ‘noise’ is reduced can legitimacy be
exported from the audit process.
From this neo-anthropological point of view,
how auditing is done is central to understanding its
micro- and macro social role. Pentland (1993)
observes extensive social control over the patterns
of behaviour of the audit team which go beyond
the technical discipline of ‘structure’ discussed
above. Impression management, as the manage-
ment of the micro-public face of auditing, is cru-
cial: to be an auditor, you have to act like one
?rst. Indeed, agency theory reminds us that the
emphasis on appearances and process as quality
proxies is to be expected where observability of the
quality of the output is problematic. But much
more is at stake than the use of proxies as metho-
dological shortcuts: understood expressively,
quality proxies may embody deeply held norms of
conduct and evaluation And they may ?nd their
extremes not only in dress codes but also in skin
colour.
Pentland (1993) observes and deems signi?cant
the preponderant use of expressive and emotional
language in the audit process. For him the lan-
guage of ‘comfort’ is more than just a metaphor,
but also a signal that hunch and intuition are
formed from repeated collective interactions
within the audit team. This is not just idiosyncratic
‘gut feel’, but a process by which auditors learn to
have the intuitions appropriate to the professional
communities they inhabit. In this respect indivi-
dual judgements are products not simply of cues in
an external environment but of collectively main-
tained habits which are formed in progressive
levels of comfort, from the speci?c audit team to
the public. In short, multilateral agency problems
get overcome at the level of a collectively shared
language of comfort.
Pentland (1993) takes as given that in many cir-
cumstances auditing is the ‘certi?cation of the
unknowable’ which requires ritual procedures to
transform indeterminacy into institutionalized
order. For example, who does work is as impor-
tant as what is done, and the distinction between
insiders and outsiders is important. Even if the
former do the ‘work’ in a conventional sense,
nothing is really ‘produced’ until the external
auditors provide authorized witnessing of it.
Hence, the mere presence of an auditor as an
independent ‘ritual priest’ serves to construct the
institutional authority of the work. While cogni-
tive based research focuses more on the speci?c
tasks in the audit process, the anthropological
perspective permits a focus on the authority
structure of the audit process, a structure which
must be reproduced. ‘Signing-o?’ at all levels of
the audit process is simultaneously a signal of
work that has been satisfactorily done and an
expression of the authority of the reviewer to dis-
cipline the work process. Hunch and comfort
underlie the formality of review.
Because Pentland (1993) is trying to characterize
the interactive nature of the audit process, as
sociological studies of science and ‘laboratory life’
have done (e.g. Latour & Woolgar, 1979), aspects
of practice which seem peripheral from certain
methodological perspectives become analytically
important. For example, working long hours with
short breaks may be needed to get the work done in
some technical sense, but this also has an important
social function. The impression of audit is much
less of a practice comprised by careful and con-
sidered judgement, as by relentless application and
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 385
e?ort. In place of, or at least in addition to, cool
cognition, we see elaborate displays of hard work.
These display dimensions of audit are particularly
evident in the business of working paper produc-
tion. In a material sense, this is all that audits pro-
duce. Van Maanen and Pentland’s (1994) study of
‘‘cops and auditors’’ emphasizes the central role of
documentation in representing practice. Records
are ‘‘institutionalized forms, [and] represent the
collective wisdom of those who are trained to keep
them’’. Furthermore, they are designed implicitly or
explicitly to produce an e?ect i.e. they have a rheto-
rical role to convince a presumed audience, espe-
cially a potentially adversarial legal audience,
something that defenders of formalist and structured
approaches to auditing have always made explicit.
‘Paperwork’, even in electronic form, mediates
the front and back stage of a practice as an active
process of erasing mess and of scripting a rational,
defendable and legitimate ‘face’. Indeed, actions
and judgements can be scripted after the event,
although whether auditors (and scientists) actually
write their conclusions in advance is an empirical
question. At least, Van Maanen and Pentland
(1994) demonstrate that records such as audit
working papers have multiple uses and are not
simply an actuarial re?ection of what has hap-
pened. Records can be used to beat the system,
control others, protect oneself, save time, avoid
scrutiny, as well as document that work has been
done. Above all, they must take a culturally legit-
imate form to perform any of these roles.
Audit working papers are therefore a genre of
organizational text: in technical terms they sup-
port the conclusions of the auditor and encode
good habits of ?le maintenance. Since the initial
audience is the review team, working papers are
also a basis for evaluating and teaching sta?. The
language within working papers is policed care-
fully and the structure–judgement tension is evi-
dent in the question of personalized or impersonal
style: the ?rst person is usually deemed unscienti?c
on the one hand but expresses professional judge-
ment on the other. As part of the policing role,
working papers are often ‘cleaned up’ and edited
with a potential adversarial audience in mind:
‘‘[W]orking papers become the discursive repre-
sentation of the auditor’s deliberation, discern-
ment and judgement’’ (Francis, 1994, p. 260).
The role of litigation in structuring the self-pre-
sentation of audit practice is a very important
theme in a number of the papers discussed in this
essay. The drift towards increasing formal struc-
ture re?ects the need to produce a discourse that
can stand up to the court’s gaze. In this respect, all
the institutional incentives are on the side of
structure and audits may even become more
focused on the production of working papers than
on the ‘doing’ of an audit (Francis, 1994, p. 261).
Documentation also plays an important role in
enabling the regulation of audit itself. The role of
working papers in disciplining audit sta? has
already been mentioned, but the development of
peer review in the US as a response to quality
worries in the 1970s extended the audience for
them. Fogarty’s (1996) critical study of the peer
review process argues that it has resulted only in
‘‘symbolic conformity displays’’ by accounting
?rms and a de-coupling between peer review as
programmatically espoused and as achieved. This
regulation of practice tends to privilege systems
and de?nes quality in terms of system structure
and espoused process, a paper chase which results
in the ?nding that most audits are ‘successful’
(Fogarty, p. 251; also Power, 1999, chapter 4).
Fogarty concludes that regulation lacks teeth
because its ‘logic of con?dence’ does not deal with
the core problem; peer review focuses on the indi-
vidual (typical for US regulatory systems) rather
than the ?rm, and translates a serious public
interest problem into a normalized discourse. The
position of the SEC in regulating auditing is
necessarily ambivalent; it needs the auditing sys-
tem as a whole to work since it underpins much of
its business regulation. Hence, peer review systems
are generally under pressure to particularize audit
failure (also Power, 1993) and to support the
legitimacy of the audit process as a whole.
Despite the fact that the ?le keeping habits of
audit ?rms may have changed, and that working
papers are increasingly electronic in form, the
insights of the studies above remain relevant. Each
in di?erent ways draws attention to the legitimiz-
ing role of documentation and image management
in the audit process. The organizing concepts for
386 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
these studies are not those of e?ciency or techni-
cal e?ectiveness, but ‘order’, ‘stability’ and ‘taken-
for-grantedness’. In the attempt to get close to the
workface of practice, they strongly suggest how
the micro-production of legitimacy is an intimate
and powerful component of auditing, which also
supports its public role. Con?dence in that role
has allowed auditing, as a systemically legitimate
practice, to develop in many new areas, as we shall
now discuss.
4. Making new things auditable
The fourth and ?nal substantive theme in this
discussion concerns the extension of auditing
practices to an increasing number of areas, some-
thing which has been evident in public sector
administration in all developed countries (Power,
1994, 1999), but which is particularly conspicuous
as a ‘commonwealth’ e?ect in the UK, Australia,
New Zealand and Canada. In essence, auditing
has become a legitimate part of good management
practice in a wide variety of domains. But if there
is this tendency in certain economies towards the
‘veri?cation of everything’ (Pentland, 2000) then
there is also a need for more detailed empirical
and comparative studies on the extension of audit-
type practices into new areas, and their e?ects and
consequences.
Radcli?e (1998) provides one such useful case
study of the emergence of ‘e?ciency’ audit in
Canada and of the way in which audit becomes
inscribed with the aims and objectives of public sec-
tor reform. The argument is that problems experi-
enced in public sector management led to a radical
programme for administrative reform known as
the ‘new public management’. In essence, political
processes produced a need to know and realize
e?ciency in public services, and this in turn led to
an expansion of audit mandates. While Radcli?e
(1998) deals speci?cally with developments in the
province of Alberta, the analysis enlightens devel-
opments in other regional and state jurisdictions.
Radcli?e (1999) extends his earlier analysis to
consider more directly the speci?c operationaliza-
tion of e?ciency auditing. While e?ciency may be
clear as a high level and abstract concept, its
operational reality is less so. State auditors could
only come to know and judge e?ciency on the
basis of their infrastructure legacy in the language
and practice of attest auditing. E?ciency auditing
necessarily gets constructed in situ from available
resources because manuals and o?cial guidance
are vague and formalization incomplete, another
twist to the structure-judgement debate. In this
context, the very lack of structure, the high
dependency on tacit knowledge and the limitations
of traditional CA skills mean that making ‘e?-
ciency’ auditable is very much the art of the pos-
sible in a new normative climate with a new set of
communicative demands for the auditor. Accord-
ing to Radcli?e (1999), in the ongoing process of
matching available technologies and templates to
the work needed, auditors came to see their
reports on e?ciency as strategically important and
as a way to act on the very environment which was
making new demands upon them. This shows how
new mandates provide opportunities for profes-
sional reinvention and for a new legitimacy in the
managerial hierarchy of control agents.
The expansion of auditing mandates in the
entire area of ‘value for money’ and e?ectiveness
auditing is a special case of a more general audit-
ing phenomenon, one that is already embedded in
the practice of ?nancial auditing. Power (1996)
attempts to capture some of the general char-
acteristics of the relation between auditing and the
auditable object. Based on the cases of research eva-
luation, quality management and brand accounting,
it is argued that auditability must be constructed as a
series of acceptable and legitimate techniques. In
other words, prior to the application of audit
procedure by individual practitioners, the knowl-
edge base of auditing must be institutionalized.
Such institutionalizations are often temporary, as
the case of sampling discussed above demon-
strates, but they make possible the extension of
the audit craft into new areas and object domains.
The de?nition of the role and scope of auditing
becomes a very important part of this institutio-
nalization process, or lack of it. From this point of
viewde?nitions of auditing are dependent variables,
something which underlies the development of
environmental auditing and the emerging role of
accountants within it (Power, 1997). The de?nition
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 387
of environmental auditing is a professional stake
because it may render one body of expertise ‘rele-
vant’ at the expense of another. The analysis
begins from the evident plurality of de?nitions of
environmental auditing, re?ecting the di?erent
institutional origins and interests at stake in this
emerging ?eld. The plurality of accreditation
schemes and the ‘new managerialism’ of environ-
mental management created a competitive profes-
sional ?eld. Observable projects of closure in this
?eld include the attempt to de?ne environmental
auditing in such a way as to place it close to
existing professional competencies. Accordingly,
accountants favoured a more managerial and
abstract de?nition over one that that was more
scienti?c in content. These strategies of de?ning
and promoting similarity relations underlie pro-
fessionalization projects and claims to expertise in
new domains.
5
The loose assembly of tools and
techniques underlying ?nancial auditing could
only be extended to environmental auditing if the
legitimacy of these claims to similarity could be
accepted. In countries like Denmark this was pos-
sible, whereas in the UK it was not.
More generally it is clear that o?cial de?nitions
must be regarded as dependent variables, i.e. as
representative strategies that may legitimize parti-
cular groups. Such de?nitions always have an
expressive or aspirational dimension for the practical
?eld and are part of its ideal self-representation;
their non-use in practice is also signi?cant. High-
level de?nitions circulate in text books and o?cial
guidance documents as ideal statements of what
auditing should be. These de?nitions can be regar-
ded as ‘programmatic’ in Rose and Miller’s (1992)
sense, products of a regulatory discourse which
inscribes its pure ambitions in them. The rise of
‘assurance services’ and ‘risk management’ are fur-
ther projects of professional reinvention for audit
and auditors, in which aspects of the audit craft are
being re-assembled for di?erent purposes and mobi-
lized with di?erent vocabularies, often in the face of
competition from other specialists, such as insurers.
This section has reviewed a small number of
papers which suggest that the roles and operations
of auditing are never ?xed and self-evident. These
roles are a function of speci?c political and man-
agerial discourses which rely for their implement-
ability and legitimacy on some kind of audit
practice. From this point of view, all forms of
auditing, particularly in the state sector but also in
the corporate context, have always been more
than a simple attest function. New roles and
functional scope have always been demanded of
audit. It is as if the attest function can never be
realized and operationalized in its pure form but
has nevertheless provided an important foothold
for the new roles that audit in practice is asked to
play. New objects and practices are continually
being made auditable, and now ‘assurable’, as old
ways of doing them break down.
5. Discussion: the social construction of profes-
sional inference
Auditing is essentially an inferential practice,
but there are di?erent perspectives on the nature
and quality of the inference process. The cognitive
psychology approach, which dominates auditing
research, concerns itself primarily with how audi-
tors respond to cues in the environment, make
decisions and perform techniques. Experimental
studies investigate judgments in laboratory set-
tings and examine the sources of stability and
consensus of judgement over time and across dif-
ferent auditors. A normative aspect of this work
concerns the development of improved practice,
such as in the form of decision support tools. A
careful reading of this literature shows up the
importance of procedural e?ciency in the light of
the di?culty of observing the output of audit.
Thus, in the margins of cognitivist work the pro-
duction of legitimacy is clearly visible: good
auditing is equated with process, how work is
done becomes important (Power, 1995). More
generally, empirical and analytical studies assume
that audit quality is ‘unobservable’ and that other
5
One very interesting strategy by accountants in the envir-
onmental ?eld is the appeal to independence as a de?ning core
competence of accountants. Willmott and Sikka (1995) have
drawn speci?c attention to this strategy and it is consistent with
their observations that in recent years the accounting profes-
sions have been more active in marketing this aspect of what
they do, transforming independence from a public interest
desideratum into an advisory imperative.
388 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
empirical proxies must be used: size of audit ?rm,
years of experience of the practitioners. In part
audit quality is regarded as a function of auditor
‘e?ort’, something which it is inherently di?cult
for principals to observe.
Yet the problem of audit quality is not merely
an agency problem; there are interesting agency
problems associated with it, but they do not tell
the whole story. Even where audit e?ort can be
made transparent, audit quality is obscure to
auditors themselves. It is not analytically clear
what ‘good auditing’ really is, since outputs are
su?ciently ambiguous for auditors themselves to
be unsure. Auditors do not know if they are good
auditors or not, however much e?ort they put in
and signal to outsiders. They often cannot be sure
whether the audit has ‘failed’ or not, whatever
failure means, despite their best procedural e?orts.
And, as we have seen above, the audit process is
permeated by hunch and intuition, despite projects
to introduce more formal structure. These intui-
tions produce an order of some kind, an order
which is relied upon both by auditors themselves
and by others. There are problems from time to
time, but these emerge against a background sta-
bility and it is toward a deeper institutional
understanding of the sources of that stability, and
ideas of audit quality, that many of the studies
discussed above lead us.
Di?erent aspects of the audit quality problem
have been addressed: the formal structure of the
audit process; the economic pressures to de?ne
audit quality in a manner consistent with the need
to cut costs; the use of working papers as a basis
for solving the micro-agency problems in the audit
process (in so far as e?ort and compliance with
standards can be observed in the working papers);
the extension of audit procedures to new areas and
the problems associated with this. In this way audit
quality is de?ned procedurally rather than in terms
of the constantly asserted but elusive output of
added assurance. From this point of view, the qual-
ity of audit resides only in part in the judgement
processes of individual auditors; it is also a func-
tion of what gets accepted, stabilized and institu-
tionalized as a way of doing things. Furthermore,
a workable division of labour between the cogni-
tive and sociological orientations can now be
postulated: the former takes as given the backg-
round system of rules and processes and observes
their implementation by individual auditors; the
latter does not take this background as given. The
sociological point of view seeks to make explicit
the social and economic basis for audit practice.
So how is the communal belief that audits pro-
duce something valuable sustained and achieved?
The preceding discussion suggests that auditor
expertise and audit quality are endogenous parts
of a social system which leaves its often hidden
mark on the technical processes of audit. In dif-
ferent ways, the studies discussed above suggest
that professional inferences are social practices
which are technical in form but which also de?ne a
community of knowledge and practice; inference is
a matter of sharing assumptions that are accepted
by a group and involves learning similarities and
concept application, a cultural enterprise which
cannot be fully encoded in a formal rule. A simple
example illustrates the issue.
If we ask how to continue the series 2, 4, 6,
8. . .. . ., it may seem ‘natural’ to say: ‘10, 12, 14,
16, etc.’ i.e. to use the ‘rule’ of n+2. However,
there are other possibilities, such as: ‘2, 4, 6, 8. . .2,
4, 6, 8’ or ‘2, 4, 6, 8. . .who do we appreciate?’
Which continuation is appropriate will depend on
the community of inquiry (Collins, 1985). So we
can suggest that inference in the context of audit-
ing is a matter of knowing how to go on in a very
speci?c community. The suggestion is that in the
case of ‘2, 4, 6, 8. . .’ a procedural algorithm can-
not be determined without knowing the institu-
tional context and this is to suggest that inference
is a matter of habit and custom as much as of
deliberation. While a great deal of justi?cation for
inference can be made within the system in ques-
tion, in the end the only response to a sceptic out-
side the system is simply to say: ‘‘Well, this is just
how we do it in this system’’.
6
6
Another example is given by the classical paradox of infer-
ence implied by Lewis Caroll’s famous parable of Achilles and the
Tortoise. See Collins (1985, p.15) for a fuller discussion of the
social basis of rules. In its strongest form this suggests that
deductive connections are ‘interpretative afterthoughts’, meaning
that formal thought stands in an ex post facto relation to informal
thought. Understanding the way that formal thought is made to
appear formal is the essence of the sociological project.
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 389
For any formal rule, one must know how to
apply it and in knowing how to apply it one ful?ls
the role demanded of an institution. Rules and
roles are intertwined as part of an orderly system.
To say this is to say that there is a part of rule
following which is not cognitive or individualistic.
Rather, auditors are invoking a certain kind of
collective order when they make inferences,
negotiate (Pentland & Carlile, 1996) and learn to
know what ‘‘gives comfort’’ (Pentland, 1993). As
McNair (1991) and Fischer (1996) show, this
background order reveals itself in times of change
and economic crisis. It may protect itself via blame
allocation and particularization (Fogarty, 1996),
de-credentializing challenges (Power, 1996, 1997)
and suppression (McNair). Alternatively, new
ways of doing things may emerge (Radcli?e, 1998,
1999).
Analyzing the social construction of auditor
inference can be read as a merely critical project
which undermines the knowledge base of auditing
by relating it to speci?c contexts and local inter-
ests. Thus, bringing in the messy backstage of
audit practice seems to undermine the front stage
which auditing presents to the outside world.
However, although there is clearly an element of
challenge and criticism in the papers discussed
above, particularly a deep scepticism about the
merits of structured audit approaches, their
underlying constructivism is not necessarily cri-
tical in this sense. Drawing attention to the com-
munal component of knowledge need not make
such knowledge insecure. The conclusion is more
that auditing researchers should treat all questions
of audit knowledge as social questions, i.e. as
questions about communities, rules and authority,
and that agreement about what is e?ective is
coextensive with de?ning a community of practi-
tioners. Auditors appear to, and are able to,
engage in cognitive acts because certain practices
have become accepted and they have been
endowed with authority to do so.
6. Producing legitimacy: some research issues
The introduction above hinted at some of the
methodological issues raised by the papers
reviewed, and the politics of auditing research
within which they operate. Very few of these
papers are really ?eld studies and none is ethno-
graphic in the anthropological sense of that term,
despite Fogarty’s (1996, p. 262) call for more eth-
nographic treatment of ‘‘The exchange processes
that construct the transactional reality of reg-
ulation. . .’’. A detailed discussion of the metho-
dological strengths and weaknesses of each of
these papers would de?ect from the substantive
focus, but some obvious issues can be observed.
The traditional worries about the loss of control
and generalizability, and the problems of theory
driven data collection, can be traded o? against
the need for opportunism in an under-researched
area, and the merits of richness and interest
value. It is also important to note that many of
these studies are by-products of other projects,
a fact which re?ects the ongoing politics of
legitimate research, problems of access and of
presentation.
The studies share a commitment to explicate the
conventional nature of audit practice and to ana-
lyze the need for both individual practitioners and
the ?eld itself to remain legitimate. In this respect,
there is a family resemblance between these studies
and work in the sociology of knowledge more
generally. One of the interesting and transferable
insights from science studies [e.g. the so-called
‘strong’ programme in Bloor (1976)] is the need to
explain how credibility and legitimacy are accom-
plished, other than in terms of the truth or other-
wise of what agents maintain. That is, the studies
reviewed point to an interesting explanandum for
audit research: credibility and common sense are
to be explained, rather than truth or e?ciency. In
its extreme form one is required to follow the
methodological imperative to ‘‘treat descriptive
language as though it were about imaginary
objects’’ (Collins, 1985, p. 16).
This form of methodological bracketing pre-
vents the auditing researcher from being slave to
the concepts and categories of the audit ?eld itself;
it may enable the researcher to be in the ?eld but
not of it. In addition it reminds the researcher
than many terms of art in professional practice are
aspirational or programmatic, rather than descrip-
tive. In this way one can avoid what Bourdieu
390 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
calls the ‘‘legalistic illusion’’, i.e. the assumption
that formal and apparently well-formed concepts,
like ‘risk management’ or ‘audit’ refer to anything.
The relation between these terms and a set of
practices which such terms are regarded as
describing is something to explain rather than to
assume.
Another suggestive lesson is that the papers
reviewed contain, albeit implicitly and in an
underdeveloped form, an explanatory schema
which deserves to be made explicit and which is
useful in orienting research into accounting and
auditing change. The schema may be characterized
in terms of three analytical categories, which have
intellectual currency in the science studies litera-
ture: controversy, closure and credibility.
6.1. Controversy
The ?rst stage of the research process requires
the selection and analysis of an episode of con-
troversy or crisis in the ?eld. This could be a par-
ticular controversy such as: the need to cut audit
costs but maintain quality; auditing e?ciency
according to vague political mandates; changing
audit techniques in a climate of practitioner con-
servatism and resistance; de?ning environmental
auditing when di?erent de?nitions serve di?erent
interest groups; steering a path between structure
and judgement in the evolution of best audit
practice; regulating auditors by a peer review sys-
tem and so on. In general, the fate of any dis-
turbance to a well-established ?eld of professional
knowledge, such as auditing, is an open and com-
plex one. Collins (1985, chapter 6) argues that
controversy can be conceptualized as an event or
chain of events which:
1. will only disturb if it is di?erent or contra-
dicts what went before AND
2. will only disturb if it carries with it su?cient
allies.
Clearly principles (1) and (2) pull in di?erent
directions. While principle (1) accords with our
common sense notion of controversy, principle
(2), focuses on the social basis for something to
count as a genuine controversy, rather than to be
ignored. The more extreme the challenge to exist-
ing forms of order, the less legitimate it will appear
and the fewer allies it will have. The greater the
number of allies under principle (2), the less likely
it is that an event will generate radical change and
the more likely that some kind of consolidation or
codi?cation will take place. In short, controversies
themselves must be framed as legitimate to count
as a controversy in the ?eld.
6.2. Closure
The concept of closure is well-established in the
social sciences. In this phase of empirical work,
the substance of the analysis addresses those pro-
cesses arising from the controversy by which con-
sensus and order are re-constructed around
practice, for example in the development of new
auditing guidance for sampling, e?ciency audit-
ing, the suppression of the cost-quality problem,
audit sign-o?. As Fischer’s study shows, rather
than good technique leading to consensus in the
practical ?eld, consensus is required before the
technique can be good, even though the mechan-
isms by which controversy is resolved are often
di?cult to observe in the ?eld.
6.3. Credibility
Closure evolves from processes of negotiation at
di?erent levels characterized by allies and oppo-
nents with an interest in renormalizing practice.
Credibility building strategies by participants may
in turn implicate a wider social and economic
milieu, for example a scienti?c culture in the case
of statistical sampling in North America, the new
public management in the case of e?ciency audit-
ing, and beliefs in self-regulation in the case of
peer review. We saw above that how auditing is
written up and reported is an important part of
auditor credibility, a credibility which in turn is
reinforced by licensing, regulatory and accredita-
tion systems. Professionalization projects involve
complex articulations of expert tasks and jurisdic-
tional claims which seek to attach themselves to
the legitimizing authority of the state (Abbott,
1998; Willmott, 1986). Even at the level of exam-
ination systems and the seemingly mundane codi-
?cation of practical knowledge, legitimacy will be
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 391
constructed around speci?c rhetorics of public
interest and knowledge.
In all these cases, the advancement of credibility
and the renormalization of practical common
sense depends on eventually ‘e?acing’ the context
in which closure was contested. A great deal of
work is done by agents to ensure that what is
eventually done and agreed upon appears natural,
obvious and, temporarily at least, uncontested.
The suppression of con?ict, the production of
working papers, the development of formal audit
techniques, all involve the puri?cation of context,
authorship and interests. Against this, counter
strategies move in precisely the opposite direction,
to reopen controversy and to disturb closure by
appealing to hidden interests and the particularity,
rather than universality, of expertise (Power,
1997). It is eventually an empirical question as to
who is a quali?ed auditing expert, and what the
boundaries of certi?ed knowledge are.
To summarise: the auditing papers discussed
above suggest the relevance of three guiding
themes which could work well in a wide number of
research areas, both at the ground level analysis of
practice and at the policy development level look-
ing at committees and interest groups. These
themes do not favour one qualitative methodology
over another, nor do they exclude quantitative
analyses. Rather, this scheme provides a basis for
framing critical audit research which avoids being
hostage to professional concepts and claims to
expertise. Following, Latour’s (1987, p. 258) ?rst
rule of method, there should be no premature
black boxing of the audit process and researchers
need to be re?ectively vigilant about the inherited
stock of concepts which they use.
7. Conclusions: auditing and the production of
legitimacy
Auditing practice can be regarded as a self-reg-
ulating system whose components (training pro-
grammes, professional standards, quality control
sub-systems, disciplinary mechanisms) are an
interacting, semi-institutionalized and loosely
coupled whole, a structure that is constantly mov-
ing and subject to economic, regulatory and poli-
tical pressures for change. This assembly of
practices is pervaded by ideas about what auditing
could and should be. Even at the level of techni-
que and process addressed by many of the papers
reviewed above, practice is deeply constituted by
blueprints and de?nitions. And even though the
actual conduct of auditing practice predictably
does not conform to the blueprint, idealizations,
de?nitions and concepts are constantly invoked to
give structure to practice, to provide frames of
meaning for practitioners and crucially, to provide
the idiom through which practice is written up and
represented to the outside world. Accordingly, an
important part of being a practitioner is to create
representations of problems and solutions that are
generally regarded as legitimate. Indeed, one can
go further and say that there are not two distinct
things: practice and legitimations of practice.
Rather, practice is itself permeated and con-
stituted by strategies of representation.
So auditing is far from being a self-evident set of
techniques which require occasional improvement,
but is rather a series of hopes and aspirations
inscribed in its most mundane routines. Witness
the aspirations of internal auditors in recent years
and their e?orts to renew their legitimacy on the
back of corporate governance initiatives. The
auditing system exports and imports legitimacy,
but in a cycle of constant reform in which
change must be made legitimate itself, and in
which the legitimacy of audit is constantly
threatened by the misalignment of expectations
about and within the system. These threats lead
to pressures for the rationalization, formalization
and transparency of the audit process, a clearly
visible trend as the writing of audit, in the form of
standards and technical guidance, has intensi?ed.
Yet, despite the ‘counter-mainstream’ studies
discussed in this paper, the role of auditing in
the production of legitimacy, and the con-
sequences of this, remain under-documented and
under-researched.
Acknowledgements
This paper was originally presented at the 25th
anniversary conference for Accounting, Organiza-
392 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394
tions and Society held in Oxford, July 2000. I
would like to thank the Warden and Fellows of
All Souls College, Oxford, where I was a visiting
fellow in the Hilary and Trinity terms 2000, for
providing the supportive environment within
which this project was conceived. In addition, I
gratefully acknowledge the ?nancial support of
the Economic and Social Research Council of the
United Kingdom. I am also very grateful for the
helpful comments of Ira Solomon, Robert Libby
and two anonymous referees.
References
Abbott, A. (1988). The system of professions. Chicago: Chicago
University Press.
Bell, T., Marrs, F., Solomon, I., & Thomas, H. (1997). Auditing
organizations through a strategic systems lens. KPMG Peat
Marwick LLP.
Bloor, D. (1976). Knowledge and social imagery. London: Rou-
tledge & Kegan Paul.
Carpenter, B., & Dirsmith, M. (1993). Sampling and the
abstraction of knowledge in the auditing profession: an
extended institutional theory perspective. Accounting, Orga-
nizations and Society, 18, 41–63.
Carpenter, B., Dirsmith, M., & Gupta, P. (1994). Materiality
judgements and audit ?rm culture: socio-behavioral and
political perspectives. Accounting, Organizations and Society,
19, 355–380.
Collins, H. M. (1985). Changing order: replication and induction
in scienti?c practice. London: Sage.
Collins, R. (1991). On the microfoundations of macro-
sociology. American Journal of Sociology, 984–1014.
COSO (1991). Internal control—integrated framework.
Committee of Sponsoring Organizations of the Treadway
Commission.
Cushing, B. E., & Loebbecke, J. K. (1986). Comparison of audit
methodologies of large accounting ?rms, studies in accounting
research no. 26. Sarasota, FL: American Accounting
Association.
Dirsmith, M., & Haskins, M. (1991). Inherent risk assessment
and audit ?rm technology: a contrast in world theories.
Accounting, Organizations and Society, 16, 61–90.
Dirsmith, M. W., & McAllister, J. P. (1982). The organic versus
the mechanistic audit. Journal of Accounting, Auditing and
Finance, 214–228.
Fischer, M. J. (1996). Real-izing the bene?ts of new technolo-
gies as a source of audit evidence: an interpretive ?eld study.
Accounting, Organizations and Society, 21, 219–242.
Fogarty, T. J. (1996). The imagery and reality of peer review in
the US: insights from institutional theory. Accounting,
Organizations and Society, 21, 243–267.
Francis, J. R. (1994). Auditing, hermeneutics and subjectivity.
Accounting, Organizations and Society, 19, 235–269.
Gendron, Y., & Bedard, J. (2001). Academic auditing research:
an exploratory investigation into its usefulness. Critical Per-
spectives on Accounting, 12, 339–368.
Humphrey, C. (2001). Audit research—looking beyond North
America. Critical Perspectives on Accounting, 12, 369–376.
Humphrey, C., & Moizer, P. (1990). From techniques to ideol-
ogies: an alternative perspective on the audit function. Cri-
tical Perspectives on Accounting, 1, 217–238.
ICAEW. (1999). Internal control: guidance for the directors of
listed companies incorporated in the United Kindgom. London:
Institute of Chartered Accountants in England and Wales.
Jeppesen, K. K. (1998). Reinventing auditing, rede?ning con-
sulting and independence. European Accounting Review, 7,
517–539.
Latour, B. (1987). Science in action. Milton Keynes: Open
University Press.
Latour, B., & Woolgar, S. (1979). Laboratory life. London: Sage.
McNair, C. J. (1991). Proper compromises: the management
control dilemma in public accounting and its impact on
auditor behaviour. Accounting, Organizations and Society,
16, 635–653.
Pentland, B. T. (1993). Getting comfortable with the numbers:
auditing and micro-production of macro order. Accounting,
Organizations and Society, 18, 605–620.
Pentland, B. T. (2000). Will auditors take over the world? Pro-
gram, technique and the veri?cation of everything. Account-
ing, Organizations and Society, 25, 307–312.
Pentland, B. T., & Carlile, P. (1996). Audit the taxpayer, not
the return: tax auditing as an expression game. Accounting,
Organizations and Society, 21, 269–287.
Power, M. (1992). From common sense to expertise: re?ections
on the pre-history of audit sampling. Accounting, Organiza-
tions and Society, 17, 37–62.
Power, M. (1993). The politics of ?nancial auditing. The Poli-
tical Quarterly, 64, 272–284.
Power, M. (1994). The audit explosion. London: Demos.
Power, M. (1995). Auditing, expertise and the sociology of
technique. Critical Perspectives on Accounting, 6, 317–339.
Power, M. (1996). Making things auditable. Accounting, Orga-
nizations and Society, 21, 289–315.
Power, M. (1997). Expertise and the construction of relevance:
accountants and environmental audit. Accounting, Organiza-
tions and Society, 22, 123–146.
Power, M. (1999). The audit society: rituals of veri?cation (2nd
ed.). Oxford: Oxford University Press.
Power, M. (2000). The audit implosion: regulating risk from the
inside. London: Institute of Chartered Accountants in Eng-
land and Wales.
Radcli?e, V. S. (1999). Knowing e?ciency: the enactment of
e?ciency in e?ciency auditing. Accounting, Organizations
and Society, 24, 333–362.
Radcli?e, V. S. (1998). E?ciency audit: an assembly of ration-
alities and programmes. Accounting, Organizations and
Society, 23, 377–410.
Rose, N., & Miller, P. (1992). Political power beyond the state:
problematics of government. British journal of sociology, 43,
173–205.
M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394 393
Strathern, M. (Ed.). (2000). Auditing cultures: anthro-
pological studies in audit, ethics and the academy. London:
Routledge.
Van Maanen, J., & Pentland, B. (1994). Cops and auditors: the
rhetoric of records. In S. B. Sitkin, & R. J. Bies (Eds.), The
legalistic organization. Newbury Park: Sage.
Willmott, H. (1986). Organising the profession: a theoretical
and historical examination of the development of the major
accountancy bodies in the UK. Accounting, Organizations
and Society, 11, 555–580.
Willmott, H., & Sikka, P. (1995). The power of ‘‘indepen-
dence’’: defending and extending the jurisdiction of account-
ing in the United Kingdom. Accounting, Organizations and
Society, 20, 547–581.
394 M.K. Power / Accounting, Organizations and Society 28 (2003) 379–394

doc_860348260.pdf