EXPERTS NETWORKS AND INSCRIPTIONS IN THE FABRICATION OF ACCOUNTING IMAGES

Description
This ethrqraphy of three Australian hospitals seeks to understand how and why new accounting systems
are “experimented” with in organintions. tatour’s sociology of translation is adapted to argue that accounting
change emerged not bccausc there was certain knowledge of positive economic outcomes hut because an
uncertain faith, fostered by expert-generated inscriptions and rhetorical strategies, was able to tie together
shifting interests in an actor network The paper also highlight5 how accounting may ironically be both real
and a simulation.

Pergamon Accountfn~ orgrmfzatfons andsocfely. Vol. 20. No. 213. pp. 11 l-145, 1995
Elsevirr Science Ltd
Printed in Great Britain
0361~%82/95 S9.W+0
EXPERTS, NETWORKS AND INSCRIPTIONS IN THE FABRICATION OF
ACCOUNTING IMAGES: A STORY OF THE REPRESENTATION OF THREE
PUBLIC HOSPITALS*
WAI FONG CHUA
Uni versi ~ of New South Wal es
Abstract
This ethrqraphy of three Australian hospitals seeks to understand how and why new accounting systems
are “experimented” with in organintions. tatour’s sociology of translation is adapted to argue that accounting
change emerged not bccausc there was certain knowledge of positive economic outcomes hut because an
uncertain faith, fostered by expert-generated inscriptions and rhetorical strategies, was able to tie together
shifting interests in an actor network The paper also highlight5 how accounting may ironically be both real
and a simulation.
In the beginning there was doubt . .
It will be protested that reality, or the world, was there “lecture furniture”: several rows ofwhite formica-topped
before any representation or human language. Of course. desks that all faced the front of the room, a flip chart, an
But conceptualizing it as reality is secondary. Fit there overhead projector, a screen, a board. The twenty-three
is the human thing the making of representations. Then people who had gathered to listen to the external costing
there was the j udgi ng of representations as real or unreal, consultant, Dr. 0, were drawn from Hospitals C (7) E
true or false, faithful or unfaithful. Finally comes the (7) and N (5) the University (2), the State Health
world, not first but second, third or fourth (Hacking, Department ( 1) and an external institute of health
1983, p. 136). (1).
Truth is a thing of this world: it is produced only by
virtue of multiple forms of constraint. And it induces
regular effects of power (Foucault, 1980, p. 131).
Dr. 0: Hospids are production systems. General Motors
Holden produces cars; hospitals produce .?
Some members of the audience: Healthy people.
Dr. 0:. The answer is intermediate and final products.
Occupied bed-days are not useful definitions of hospital
products. Nor are lab tests -we are not in the business
of producing lab tests; we are not in the business of
producing occupied bed-days. But intermediate and
final products: there are obvious definitions (of these
terms)...
I March 1989
It was a plain lecture room in one of the buildings of
the University. The room was equipped with the usual
’ The author would like to thank the following for comments on earlier drafts of this paper: Peter Armstrong Peter Booth,
Mike Briers. David Cooper, Pieter Degeling, Mark Hirst, Anthony Hopwood. Richard Laughlin, Peter Luckett, Alistair Preston,
Mike Power, Tony Tinker, and Grahame Thompson. Special thanks are also due to the anonymous reviewers of this journal
for their constructive comments. This paper and earlier drafts benefited from comments received at the Annual Meeting
of the American Accounting Association, the Boston Accounting Research Colloquim, the Critical Perspectives in Accounting
Conference and the Interdisciplinary Perspectives in Accounting Conference. Helpful criticisms and suggestions were also
received in research forums at the University of Alberta, Australian National University, University of Canterbury, Case
Western Reserve University, Melbourne University, Murdoch University, Queen’s University and Wollongong University.
Finally, this study would not have been possible without the financial support of the Australian Commonwealth Department
of Health and Community Services, and this is gratefully acknowledged.
111
112 WAI FONG CHUA
There is a 0.001% of patients who use a lot of resources
but for whom there are few benefits. The money is
wasted on these patients. It is being taken away from
other patients. Physicians are not stupid but they are too
busy to think about the resource implications of ordering
tests. They want to do as much as possible for the patients
they care for. But t hey could save the resources to use
on another patient. They should be able to make
appropriate trade offs. But they lack data to make these
decisions. Health information systems are garbage at the
moment. We make savings but these go to the State
Treasury. There is no benefit to the other patients. The
money could be used more cost effectively to provide
good health care. Don’t you expect that if this
information is available, then people will do the right
thing? It has been proven in the UK [United Kingdom]
and the US [United States of America]. PPS [Prospective
Payment System] has led to declines in length of stay,
imaging rates, number of lab tests, better prevention, etc.
Hospital Financial Accountant: But you are assuming some
romantic version that doctors will accept reductions in
costs. Surgeons are no different from businessmen. They
want to put more money in their depar t ment s, t hey believe
that quality of care is increased because more money is
pumped in. They don’t want to give money to other
departments. Doctors are in a competitive situation.
Dr. 0: Yes, but the characteristics of the problem are
the same-how do we allocate money between different
units? Competition is appropriate within reason. The
basis of the argument has to be correct. A department
should say - I am producing more products than before,
therefore my budget should be increased. We should
quantify these debates and remove from subjective and
opinionated argument things that can be quantified. To
the extent that we can quantify - quantity. We can
never eliminate some degree of judgement but at least
we can debate on the basis of evidence and data.
Opposition was anticipated . .
14 August 1989
Hospital N’s committee room was dominated by a large
table in the centre. Over the course of the morning, a
number of senior hospital administrators, accountants,
intbrmation speciahsts and quality assurance stag attended
and left a meeting with four University academics. At
one point the following conversation oc c ur r ed:
Head of the Quality Assurance Unit: We need a scientific
way of making decisions. Now decision-making is more
an art and we are dealing with science. Unfortunately
QA [Quality Assurance] is seen as part of administration.
There is a them v. us among clinicians and administrators
in this hospital. Clinicians dominate the Management
Committee, this is really a clinician-run hospital.
We are trying to tell clinicians that ICD9CM
[International Classilication of D&eases, 9th Edition -
Clinical Modification) codes are more useful than local
databases. We are ving to develop UR [utilixation
reports] reports based on ICD9-CM codes. DRGs
[Diagnosis-Related-Groups], however, are seen as a
nebulous threat - connected with funding by DRGs.
We need to sell them ICD9CM coding. We have got to
sell information that is u&it1 and which is not seen as
a big stick
University Professor E: Why is the DRG connection to
funding seen as a big stick? The other side to it is to say
that those producers who are doing better ought to be
reimbursed as efficient producers?
Director of Health Information Systems: Clinicians are
afraid of programme cost-cutting .
No, the product to be sold is ICD9CM. Try and keep
DRGs out as long as possible. We are heading towards
clinical costing and cost models. We’ll need 1 or 2 bites
at it before we get it right. You need 1 or 2 bites, you
know, in order to massage the information using cost
models to get it believable. Presentation of DRG-based
data seems premature. We should try to have ICD9CM
credibility [ first]. . . There are a number of very powerful
and infhtmtial people here who are great debunkers. If
they wish to blow it up they will debunk it and we will
never get it up again. They are very critical of shot&y
data_ It’s part of our culture. It’s the way this place works.
But there were allies . . .
16 August 1989
Hospital E was in the midst of major organizational
change. There was much talk of decentralization and of
restructuring medical services into clinical modules
based on existing medical relationships. It was the
general manager’s intention that each cost centre and
module would eventually receive its own budget and
set of financial reports.
Three University academics were meeting with the
Hospital Accountant in his oflice. Talking generally about
the restructuring plans, he said: “Nurses will not work
under a doctor. We could have major industrial action
about that. Nurses will not give up their power base.
They till become the managers of clinical modules”.
In a later meeting on the same day, the University
academics met with two hospital administrators, a
medical superintendent and the divisional deputy
director of nursing.
Deputy Director of Nursing: “. Our problems with
clinical costing push us to the way of DRGs. We realized
that we were going the wrong way and we want to go
FABRICATION OF ACCOUNTING IMAGES 113
the way of DRGs . . We know clinicians are very against
DRGs. We try not to use that term. Talk about clinical
costing People felt it was unethicai to talk about
resource constraints.
The new structure? . . Nursing budgets will always
be controlled by nurses. If anybody wants to control my
budget, I’ll break their legs. They won’t be able to walk
out of this room.
The fact-building network survives.
28 May 1991
The Costing Subcommittee of the Consortium Casemix
Project met in the Board Room at Hospital S. The minutes
of that meeting record that a report detailing the DRG
costs from the lirst run of the Yale Cost Model had been
distributed to all participating hospitals and more copies
could be obtained from the University. The Consortium
was also preparing itse!lf for the second i n of the Cost
Model.
Yes, but with what consequences and power-
effects?
PROLOGUE
This is a story about a relatively small group
of people “doing accounting”. Its principal aim
is to explore the social linkages and practices
that (a) enabled this group of account-fabricators
to begin the construction of “new” accounting
numbers, (b) sustained their activity over
several years, and (c) persuaded them to
“consent” to the numbers finally created. In the
process, the paper focuses on the catalytic
capacities of knowledge experts (see also
Bloomfield & Best, 1992) the socialized
decision-ladenness ( Knorr-Cetina, 198 1) of
accounting numbers and the manner in which
they are a constitutive, transformative text in
organizations. The “economic reality” that
accounting was to reflect did not come first
ready-made. Instead it emerged through a long
process of fabrication ( Knorr-Cetina, 1981;
Latour, 1987) by a network of enrolled fact
builders and software. The collective labour of
fabrication, however, did not undermine the
labourers’ belief in the accounting images
of reality produced. On the contrary, past
compromises merely set some parameters to
further decision-making and strengthened the
networks resolve to search for better windows
on their organizations. By being a figurative
practice that promised more than it could
deliver and a set of inscriptions that was both
reality and simulation, a certain self-preserving
accounting was created, one which could
continually be made to supplement itself, thus
ensuring that the economistic truth which it had
helped to build would be perpetuated.
My narrative is constructed upon many
interpreted audio-visual observations of this
network of men and women. In my memory, I
still see them sitting in University and hospital
rooms, learning how to use a piece of costing
software, reporting back on data problems
encountered in the hospitals, strategizing on
responses to government agencies, and shuffling
bits of paper on which were written formulae,
tables, spreadsheets and graphs which yielded
new identities for their hospitals and patients. I
do not claim nor do I attempt to objectively
recount/mirror this empirical doing of account-
ing. This paper is not an asubjective, indepen-
dent writing-up of results (see Nencel & Pels,
1991 and Clifford & Marcus, 1986, for critiques
of this notion of ethnography) but the reflexive,
constructed outcome of a dialogical negotiation
of multiple realities: mine and those which
belonged to the fact-building actors employed
by the hospitals and my colleagues who
comprised the University Project Team. As
Clifford ( 1986) argues, an ethnography (such
as this) may properly be called fiction (from the
Latin root, JTngere) in the sense of being a
fashioned, that is, made up, translation of a sliver
of organizational life.
It is, however, fiction with a critical sub-text
and the paper reaches towards a notion of critique
in the several senses suggested by Luke ( 1989)
and Forester (forthcoming). First, this paper is
critical in that it is essentially reflective, reflexive
and ironic rather than positive, objective and
methodologically formalistic (see also Latour,
1988b). There are no allegedly rigorous repre-
sentations called statistics or mathematical
models here.
114 WAI FONG CHUA
Second, both through its rhetorical structure
and its subject matter, this study criticizes the
positivistic notion of an independent reality
that can be represented unambiguously by
accounting numbers or used as an epistemic
standard to adjudicate between knowledge and
non-knowledge. Explicitly, it details the social
construction of accounting images whilst
implicitly the study questions mainstream
notions of how empirical evidence commands
scientific authority (although the issue of how
ethnography itself commands what type of
validity is beyond the scope of this paper).
Third, as critique this paper focuses on
some of the complex interrelations between
knowledge, interests and power (understood as
a transformative and influential force). Foucault
(1980, p. 93) wrote,
basically in any society, there arc manifold relations
of power which permeate, character&e and constitute
the social body, and these relations of power cannot
themselves be established, consolidated nor imple-
mented without the production, accumulation, circula-
tion and functioning of a discourse. . We are subjected
to the production of truth through power and we cannot
exercise power except through the production of truth.
The production of knowledge is thus as much
about power as it is about truth. An attempt is
made here to analyse these power-knowledge
relations through, inter al& (a) detailing the
circuitous connections between the interests of
fact-builders and the political agendas of
state agencies, (b) depicting the different and
unequal access to resources possessed by actors
within the fact-building network, (c) high-
lighting how accounting images may empower
core changes in an organization’s identity, and
(d) how notions of “reality” and “truth” are
constituted discursively and help legitimize the
authority of certain experts. Social theorists
from Marx, Habermas to Foucault have indicated
that research should seek to question prevailing
systems of truth and to ascertain the possibility
of or work towards the constitution of a different
politics of truth. This paper is motivated by this
spirit of critique. However, it does not attempt
the ambitious task of detailing a more emanci-
patory accounting; its task is more modest -
the interrogation of what which is constituted
as “rational” and “real” in a specific context. The
underlying premise here is that this initial
investigation is useful, for whilst one accounting
map of an organization may be as good as
any other since neither works because it
better represents reality, each may di@er
fundamentally in terms of its institutionalized
supporting structures and power effects. Hence
the need to ascertain wherein lies the will to a
certain truth, who seeks what with changed
accounting images, how chains are configured
differently within an emergent network and
what new forms of surveillance and discipline
result.
Why is a piece of ethnography with such aims
of general interest? Critical accounting research
has long embraced the notion that accounting is
a constitutive social construction that emerges
from and becomes entangled in complex struc-
tures, localized politics, multiple discourses and
unintended happenings (Lehman & Tinker,
1987; Hines, 1988; Hopper & Armstrong, 199 1;
Hopwood, 1987; Miller & O’Leary, 1987; Nahapiet,
1988; Covaleski & Dirsmith, 1988; Bhimani,
1993). More recently, there have been attempts to
highlight the rhetorical dimensions of accounting
(Thompson, 1991; Arrington & Schweiker, 1992)
and to focus on the exercise of governmentality
by numbers (Rose, 1991; Miller & Rose, 1990).
Yet, despite this extensive body of work, there
are relatively few detailed “micro” ethnographies
of the making up of accounting numbers in
organizational settings (compare Berry et al .,
1985; Ansari & Euske, 1987; Nahapiet, 1988;
Bloomfield, 1991; Preston et al ., 1992). Whilst
Iatour’s framework has now been used by
several accounting researchers ( Robson, 199 1,
1992; Preston et al , 1992; Bloomfield et al .,
1992) the preparers of accounts have not often
been followed in action (see Latour, 1987,
1988a; Knorr-Cetina, 1981; Latour & Woolgar,
1979; Gilbert & Mulkay, 1984) and there
remains a need for research that portrays and
comments on the tactics, tools and tribulations
of account-fabrication. This paper seeks to
contribute by focusing on the part played by
FABRICATION OF ACCOUNTING IMAGES 115
experts, actor networks, computerized software
and visual inscriptions in the struggle to change
accounting representations. Further, it begins to
investigate the diverse relations that empower
accounting numbers and give them authority.
How does accounting persuade? The question
is large and this case-study only begins to
probe the issue in a particular context. Being
exploratory, it is imperfect and incomplete
(whatever completeness might mean). Given
the territory it wishes to cover, it cannot be
anything else. The paper does not promise
to provide a comprehensive story of the
complexities of account fabrication. The
narrative, for example, concentrates on the
emergence of a fact-building network and the
parts played by legitimated knowledge experts
in this process. It offers but tentative explana-
tions as to why flawed accounting numbers
command consent, especially among insiders
who acknowledge their arbitrary nature. Finally,
it does not assert the discovery of more
generalized processes and claim to be an
exemplar of these. Fabrication may take place
quite differently in different circumstances.
However, drawing upon but also going beyond
the actor network theory proposed by Latour
and Callon, the paper does spotlight a central
issue, namely the basis of accounting’s persua-
sive power, that should be further researched.
In addition, its narrative empirically illustrates
a general mode of thinking about and investigat-
ing how the microcreation of accounting
knowledge and its practice is both constituted
by and constitutes social and power relations in
a post-industrial age characterized by a “mode
of information” (Poster, 1990).
THEORETICAL NOTES
The work of Latour and Callon (see Latour,
1987; Iatour & Woolgar, 1979; Latour 1983,
1988a; Callon, 1986; Callon et al ., 1986),
in particular their actor network theory or
sociology of translation, has already been
introduced to accounting researchers (see
Robson, 1991, 1992; Preston et al , 1992). As a
result, this framework will not be discussed in
detail here. The question, however, remains -
why use their work as a springboard to
write a piece of critical ethnography about the
fabrication of accounting knowledge? Their
work is attractive for a number of reasons.
Firstly, the making up of “new” accounting
numbers and the battle to secure their legi-
timacy may be seen as being similar in important
respects to a scientific controversy. Like these
controversies, the birth of an accounting may
change the map of organizational reality, chal-
lenge existing work traditions, and unfold battle-
like, with opposing supporters and detractors
who are intent upon vanquishing each other.
Secondly, Latour’s sociology of translation
does not begin with the simplistic, positivistic
assumption that a particular science or techno-
logy (or set of accounting numbers) is rationally
accepted because it more accurately represents
reality. Nor is science assumed to be created by
the mechanical following of the rules of scien-
tific method. Instead, Latour and his colleagues
ask how a certain notion of reality came to
be socially constructed and how and why a
fact-building network emerged and survived
Machiavellian-like - what are the human and
non-human resources, strategems, ploys and
persuasive strategies used by actors with diver-
gent interests to initiate, maintain or, in certain
cases, destroy knowledge networks? Who are
the allies who/which were mobilized, how are
competitors cut down and conquered, which
authorities are used to stack particular argu-
ments, how are stronger software or formulae
built to cover as many contexts as possible in
as succinct a manner as possible? In what ways
are appeals to truth and truthfulness mobilized
by fact-builders and to what effect? Such
questions are useful beause they focus analysis
on the processes of fact-fabrication and the
technical and sociopolitical ties that hold
together to form truth. In effect, the sociology
of translation extends constructivist concep-
tions of knowledge-creation (see Knorr-Cetina,
198 1; Gilbert & Mulkay, 1984) by providing a
systematic, well-illustrated framework that high-
lights the competitive, connected and communal
116
WAI FONG CHUA
processes of knowledge-creation and the crucial
role of networks of interest.
Thirdly, the work of Latour and Callon draws
attention to the persuasive power of non-human
resources such as visual inscriptions, academic
texts and “centres of calculation” @tour,
1988a). Paperwork such as formulae, graphs and
charts are argued to possess many rhetorical
advantages: they are mobile, immutable, recom-
binable and are perceived to be built on many
facts. Most important of all, inscriptions make
black boxes visible. It is not possible for
government bureaucrats to manage or have
control over the nation’s health or economy by
looking at “it”. The “it” is plainly too vast and
complex to be seen. But when cohorts of
enquirers, accountants, medical record specia-
lists and statisticians have collected millions of
numbers, the nation’s health can be imaged and
counted. Visualization is especially persuasive.
It appeals to the sense of sight -the most valued
sense in Western, literate cultures. As Ong
(1967, 1977) points out, in such cultures, the
truth of vision has predominated over the
evidences of sound and interlocution, of touch,
smell and taste. Seeing, after all, is associated
with believing.
Inscriptions further enable the exercise of
comparative, normalizing judgement. This in
turn permits action from a distance, enabling
people far away from the scene of activity to
ostensibly have a window on those activities and
intervene in the name of better management. One
organization’s financial health may be compared
with that of another’s, conclusions drawn about
relative efficiencies and effectiveness, and action
initiated to correct evidence of ill-health.
Useful though the sociology of translation is,
it is not without ambiguity or weakness. For
example, Latour’s fourth rule of method argues
that society should not be separate fkom science.
By this, he appears to mean that the develop-
ment or spread of a body of knowledge should
not be causally explained by reference to the
interests of particular groups. This is because:
Analysts who use groups endowed with interests in order
to explain how an idea spreads. a theory is accepted, or
a machine rejected, are not aware that the very groups,
the very interests that they use as causes in their
explanations are the consequence of an artihcial
extraction and purification of a handf’ul of links from
these ideas, theories or machines. Social determinism
courageously fights against technical determinism,
whereas neither exist except in the fanciful description
proposed by the dilkion model (Iatour, 1987, p. 14 1).
According to Latour, interests and other social
factors cannot be used as causal explanations
because they are the consequences of negotia-
tion and the effect of settled disputes. Latour
uses his account of the translation of interests
through talk and negotiation as proof of the
validity of his argument. Shapin ( 1988) and
Barnes ( 1981) counter by pointing out that
accounts of interest are theoretically distinct
from interests; and each may change in a way
that is independent of the other. Also, Latour’s
and Callon’s own empirical work (see, for
example, Callon, 1986) do not depict actors as
people devoid of intentions and they too use a
concept of goal-directed action to help explain
why actors relate to a set of constructed facts
in particular ways. This apparent inconsistency
leads Shapin (1988, p. 544) to comment that
“Latour has, to all appearances, banned interests
by treating them as the same as interest-
accounts, while re-introducing the instrumental
character of technoscientific work by the back
door, in the form of ‘goals’ “. An alternative (and
more sympathetic) reading might, however,
conclude that Latour has not “banned” the
concept of collective interests per se but was
over-concerned to argue for a dynamic, more
contingent conceptualization which did not see
interests as being theoretically predetermined
by the class structure of capitalistic societies.
In this paper, interests are a crucial starting
point of analysis because people engage in
making up accounting numbers for a reason(s).
Their purposes may not be explicitly articulated
or only vaguely defined. They may change over
space-time and differ fundamentally between
individuals and groups. But accounting, like
science, is assumed to be an instrumental
activity. Thus, to begin following accounting in
the making, it is lirst necessary to identify core
FABRICATION OF ACCOUNTI NG IMAGES 117
individuals/groups who are either directly or
indirectly involved in the fact-building process
and to ask why they are or why and how they
become interested in the construction of
accounting numbers.
Interests, however, are not theorized as
atomistic, randomly distributed matters that
have no history and no possible connection with
other interests. As Scott graphically illustrates
( 1991, p. 30) in her critique of the Latour and
Callon framework, in order for researchers
to know whether a laboratory or power-
knowledge can “raise the world”, it is necessary
to “look beyond the boundaries of the labora-
tory or the laboratory’s influence to the larger
society to consider the forces that encourage
laboratories or provide counterweights to sink
them”. Current fact-building networks, there-
fore, need to be considered in terms of larger,
older networks. But interests are also not objec-
tive matters that can be rigidly read off class,
ethnic, gender, or rank differences. Instead
people’s interests are argued to be gene-like
possibilities - partly grounded in the social,
economic, political and cultural institutions of
a society at a point in time and partly possessing
the agency to transform in quite unexpected
ways, thereby injecting an element of unpredict-
ability in the relationship between interest and
outcome.
Further, unlike the work of Latour and Callon,
this paper does not present inanimate objects
such as computer software or hardware as actors
which are identical to human agents (compare
Callon, 1986). To do so reifies machines and
technologies in a way which detracts from the
purposive activities of their designers. It is
people who make up accounting numbers in
specific ways to try and achieve certain objec-
tives. Software, by contrast, has neither interest
nor agency.
ABOUT THE STORY
My story revolves round attempts to define
“the” output or product of a hospital via its
casemix. This notion is similar to the manu-
facturing concept of volume mix. Casemix seeks
to measure a hospital’s output in a defined
period in terms of the number and type of
cases (patients) treated. A number of casemix
measures exist, the one used in this study is a
set of 467 diagnosis-related groups (DRGs)
which measure output via the types of inpatients
discharged from acute care hospitals. DRG 103,
for example. refers to heart transplants while
DRG 242 identifies patients diagnosed as suffer-
ing from septic arthritis. DRGs, as an iso-
resource measure of hospital output, were
devised by a Yale research team set up in the
late 1960s. The team was primarily concerned
with hospital management, utilization review
and an evaluation of clinical performance (see
McMahon, 1987). With these early purposes in
mind, the designers used length of stay as the
proxy measure of resource consumption and
DRGs are dilTerentiated statistically on that basis.
A discourse about DRG-based accounting, that
is, the allocation of accounting costs to indivi-
dual DRGs for internal budgetary and managerial
control purposes, did not arise until the late
1970s and 1980s (see Fetter et&, 1976, 1977;
Thompson et al., 1978,1979; Fetter & Freeman,
1986; Freeman eta!, 1986; McMahon et a!, 1986;
Wickings, 1987). It was only then that specific
software, such as that embodied in the Yale Cost
Model (YCM ), became more widely known. Thus,
it is possible for researchers who are familiar with
the statistical generation of DRGs and their use
for utilization review purposes to be unfamiliar
with the technicality of DRG-based accounting.
To guide the reader, the following bare chrono-
logy lists the following “facts”. In the second
half of 1988 a team of four University academics
(the University Project Team) came together to
make a submission to the Commonwealth
Casemix Development Programme. Reflecting
the federal government’s interest in measuring
hospital output, this programme had been set
up in 1988 to fund casemix “development”
initiatives throughout Australia. Over a five-year
period, it would have at its disposal 525 million.
In its first year of operation (198889) the
programme funded some twenty-nine projects
costing approximately 15.9 million.
118 Whl FONG CHUA
The University Project Team’s final submis-
sion to the Commonwealth Programme was
made in late March 1989. By this time, the
University had become part of a consortium
which included three public teaching hospitals
(Hospitals N, C and E) and the State Department
of Health which formally co-ordinated all
submissions Iiom the State to the Commonwealth.
As a collective, the consortium applied for just
over $1 million for 1988189 to fund each hospi-
tal’s individual projects and the three projects
proposed by the University. The University’s
projects were ( 1) the development of casemix
accounting and costing systems using the YCM;
(2) the development of non-financial casemlx
management information systems and utilization
review; and (3) the investigation of organizational
change, new organizational structures which
were appropriate for casemix management, and
the development of educational packages. The
University team, which eventually included
computer specialists and research associates,
was divided into three separate subgroups, one
for each of the projects listed in the submission.
The author belonged to the subgroup invol-
ved with Project 3, in the sense that she was
actively involved with project management,
data collection and analysis. This paper, how-
ever, is essentially an ethnography of Project 1
and in particular on the making up of DRG
product costs by a subgroup of hospital per-
sonnel who were co-ordinated and guided by
University staff assigned to Project 1. In the
narrative, this subgroup is labelled the Cost
Modelling Group. Its hospital members consis-
ted of middle-level functionaries from all three
hospitals. On average, Hospital N was represented
by two accountants and one statistician, Hospital
C by two administrators (one of whom used to be
a medical records practitioner) and two accoun-
tants, Hospital E by an ex-nurse turned admini-
strator and one accountant. Project 1 staff from
the University consisted of a computer specialist,
Professor E, and a research associate. The size of
the Cost Modelling Group, however, eventually
increased by six as statf from three other hospi-
tals requested and were permitted to sit in and
learn from the consortium’s costing workshops.
Data used to prepare the case-study were
collected over the period mid- 1988 to July 1991
by (a) attendance at a series of consortium
costing workshops, regular consortium and
University project staff meetings, hospital visits,
presentations by external consultants and aca-
demics; (b) interviews of hospital personnel;
and (c) analysis of University and hospital
documents. July 1991 was chosen as a cut-off
point because a Srst run of the Model had been
completed. At this point, the results produced
were still regarded as preliminary and were not
widely circulated among doctors or nurses.
Nevertheless, the period discussed did offer
useful insights into the doing of accounting.
1988 was chosen as the start-date because that
was when the University Project Team gradually
emerged. But in another sense, the story did not
start then but much earlier, in a prior world of
“sick” governments and hospitals.
THE ANTECEDENT WORLD OF SICK
GOVERNMENTS AND HOSPITALS
Prior to the 1970s Australia was often
portrayed by its popular media, various art forms
and major literary figures as the “lucky country”.
This luck, if it did exist, apparently ran out in
the 1970s. Unpleasant economic news led to
negative social news and the community
looked to its governments (both State and
Commonwealth) for corrective action (see
Kouzmin & Scott, 1990; Painter, 1987). This was
partly because governments had grown to
penetrate many areas of Australian life - from
the provision of health, education, housing,
policing and social security to cultural and town
planning services (Painter, 1987). In tbe 1970s
however, these government agencies came to
be perceived as ineffectual and too big and
became the subject of extensive reviews. The
Bland Inquiry of 1973-1975 in Victoria spoke
of major administrative staffing deficiencies,
outmoded attitudes and a lack of initiative. The
Corbett Inquiry of 19731975 in South Australia
commented on a lack of productivity and
efficiency. The Wilenski Inquiry in 1977 into
FAEWlCATlON OF ACCOUNTING IMAGES 119
the New South Wales administrative apparatus
was equally critical. And the inIluential Coombs
Inquiry of 1976 into the Commonwealth bureau-
cracy criticized it for its aloofness from the
public, its unrepresentative composition and the
general lack of managerial accountability. These
inquiries (which spanned both conservative
Liberal and Labor administrations) enshrined a
particular definition of the malaise crippling the
Australian public sector - “inefficiency”, “little
public accountability”, “poor value for money”,
“low productivity” and “waste”.
The 1970s was not only a time of review but
also of radical reform, particularly within the
Australian health-care sector (for a more detailed
history see Crichton, 1990; Palmer & Short,
1989; Opit, 1983; Hicks, 198 1; Gardner, 1980).
Prior to 1975, Australian public hospitals were,
in general, managed by autonomous hospital
boards. Doctors, by and large, possessed an
honorary status; that is, they were not hospital
employees but were granted access to hospital
facilities for their private patients in return for
free services to public patients. As Degeling
(1992) points out, both hospital boards and
doctors were keen to maintain this type of arms-
length relationship between themselves and
with State and Commonwealth governments.
Such relationships changed, however, in 1975
when a reformist Labor Commonwealth govern-
ment created Medibank, a universal health
coverage scheme. Medibank was funded entirely
from Commonwealth revenue and administered
by a central Health Insurance Commission. This
mode of funding and administration changed
somewhat in later years as subsequent Liberal
and Labor Commonwealth administrations tin-
kered with the scheme’s precise financing
format. But universal, state-financed, health
coverage exists today, modified as the Medicare
system. The creation of Medibank and its
variations has had two important effects on the
Australian health care sector - (a) both
Commonwealth and State governments became
increasingly concerned with their respective
roles in the financing and management of
hospitals, which consume the largest share of
recurrent expenditure; and (b) the medical
profession could no longer keep governments
at arms length. Instead the profession now had
to continually negotiate (often bitterly) with
state agencies over matters such as govemment-
negotiated payment schedules and private prac-
tice rights in public hospitals (see Daniel, 1990;
Rees & Gibbons, 1986). The medical profession
also had to deal with government and com-
munity concerns about overservicing, inappro-
priate use of technology and fraud. These
worries arose partly as a result of the following
types of reports:
Between 1967-68 and 1976 there was a 20 per cent
increase in the number of medical services per head,
but, more importantly, there was a 66 per cent increase
in specialist care. Apart from more referrals, there
was a definite increase in the number of other doctor-
initiated services in the early 197Os, particularly
diagnostic investigations, including radiology and
especially pathology (Hicks, 1981, pp. 4647).
Given such concerns which, in turn, were
related to the increasing involvement of state
agencies in the funding of health care, the
proportion of the total health dollar consumed
by hospitals and the general climate of surveil-
lance and review within the public sector, it was
unsurprising that the hospital sector also became
the target for change. When Medibank was being
set up, the Commonwealth created a Hospital
and Health Services Commission which took a
direct interest in hospital management. This
Commission conducted a number of enquiries
and in 1974/75 argued that there was a need to
(a) devise funding methods that would be based
upon robust evaluations of hospital capital
expenditure and operating costs, and (b)
strengthen hospital management structures and
systems. In addition, the 1974 Sax Report and
the 1981 Jamieson Report reviewed the perfor-
mance of hospitals nationally whilst the 1983
Sax Report enquired into South Australian
hospitals. According to Palmer (1987) these
reports highlighted a number of major problems
within the Australian hospital sector. It was said
to suffer from (1) maldistribution, (2) over-
capacity and overutilization of hospital beds, ( 3)
overemphasis on inpatient care, (4 ) a lack of
120
WAI FONG CHUA
co-ordination of hospital activities, (5) absence
of statistical and management information systems
that measured hospital output and efficiency,
and (6) inadequate allocative mechanisms for
distributing funds to States and hospitals. In
short, hospitals were depicted as suffering from
the same illness afnicting the Australian public
sector more generally.
This definition of “what was wrong with hos-
pitals” in the 1970s and early 1980s was asso-
ciated with yet another shift in Commonwealth
health policy. The government that had created
Medibank was replaced in 1975 by a Liberal admi-
nistration committed to reducing Commonwealth
health expenditure as part of an attempt to
reduce inflation and reverse the downturn in
Australia’s domestic and international economic
activity. During 1975-1983, five attempts were
made by the Commonwealth to operationalize
a “user-pay” principle by reintroducing mini-
mum fees payable by patients or their private
insurers (Degeling. 1992). More importantly,
the Commonwealth adopted a recommendation
from the 1981 Jamieson Report and abandoned
existing open-ended hospital cost-sharing agree-
ments. It substituted a system of identified health
grants to States. Such grants were calculated on
a per-capita basis with adjustments for factors
such as age, sex, population size, and standar-
dized mortality. The information required to
make these adjustments was supposed to emerge
from better management information systems,
which the Report also recommended should be
established within hospitals and State health
departments. This system of fixed grants, in
effect, shifted the bulk of financial responsibility
for hospital costs from the Commonwealth back
to the States. Rut for this financing strategy to
work, the sickness of government had to be
fundamentally alleviated, if not completely
cured, in other ways.
if the 1970s was the era of problem-definition,
the 1980s was a period during which solutions
of a certain kind were implemented throughout
the Australian public sector. Caiden (1990)
Chapman ( 1990) Yeatman (1987) and Pusey
( 199 1) detail this economically rational, mana-
gerialist solution which was the increased
penetration of private business sector ideas and
practices into the public sector. To the 1970s
criticism that the public sector was not sufficiently
businesslike came the following answer:
. . . governments sought the best business advice they
could get. They employed leading business consultants
and multinational corporations expert in business and
accounting methods to investigate public organ&&ions
and produce management plans and strategies. They
brought business managers into the higher echelons of
the public bureaucracy for policy formulation and
managerial direction. They insisted that public organisa-
tions adopt and adapt business methods in their
operations. They also redesigned and restructured the
machinery of government more on the lines of the best
practices of multinational corporations and leading
Australian companies. They required their senior officials
to be more rhan policy advisers, . . . They demanded that
they be, as well, managers skilled in management and
the administrative sciences, speaking a management
language, practising management skiffs, thinking in
managerial terms and working according to managerial
models (Caiden, 1990, p. 45).
Caiden goes on to point out that this manage-
rialism of the 1980s was distinctive in being not
only associated with neoconservative econo-
mics and the predominance of economic criteria
in public policy-making but with the implemen-
tation of cost-cutting strategies, management by
objectives, financial management improvement
plans and management through information
technology. Managerialism also changed the
language of public administration. Since the
1980s words such as privatization, performance
indicators, productivity gains, commercializa-
tion, contracting-out, asset-management, and
user-charging have become part of the everyday
language of public sector managers.
The rise of managerialism and the Common-
wealth’s budget reduction strategy impacted
upon the New South Wales (the focal State of
this case-study) government and its health
sector in a number of ways (see Degeling, 1992,
for more details). In 1982, a rationalizing
State Labor government abolished its Health
Commission and health was brought directly
under ministerial control with the establishment
of the State Health Department. In the same
year, the State’s Parliamentary Public Accounts
FABRlCA-IION OF ACCOUNTlNG WAGES 121
Committee recommended a series of significant
changes aimed at tightening the financial control
of hospital accounting systems: the introduction
of a form of global budgeting, the provision
of standardized statements of sources and
applications of funds in hospital annual reports,
the requirement that hospital auditors report to
the Auditor-General and the institution of State
controls over hospitals which exceeded certain
budget allocations. In addition, in the name of
efficiency and equity, the newly established
State Health Department recommended hospital
consolidation. For the city of this study, this
meant the closure of three inner-city hospitals
in order that bed capacity might be increased
in outer suburbs which had been experiencing
fast population growth. In 1983, despite sustained
opposition from various stakeholder groups and
the media, two of the three hospitals were
closed. The third, a large teaching hospital, had
its bed capacity as well as its clinical role
substantially reduced. The State Department
further instructed hospital administrators and
boards to stay within budget or face dismissal.
No longer would the State “cough up” at year-
end in the face of hospital overspending and
“shroud waving”. As Degeling (1992) points
out, the State’s message that the rules of
the game had changed was very effectively
communicated when, in 1983, it sacked the
board and some senior staff of one public
hospital for alleged financial mismanagement.
Strict limits were also placed on total hospital
expenditures and these effectively capped expen-
diture growth in real terms. As Palmer & Short
( 1989, p. 86) point out, the total expenditure
of Australian public hospitals has not increased
in constant price terms per head of population
since 1978. In part, this outcome was achieved
through the State’s promotion of a new career
structure in medical administration (which
resulted in the growth of doctors as a distinctive
managerial group) and the increased employ-
ment of doctors (staff specialists) by hospitals.
Both processes strengthened the hand of the
State Department as it sought to more directly
control the hospitals which it was indirectly
financing. Further, in 1986 individual hospital
boards were abolished and public hospitals
came under the direct control of Area Health
Boards controlled by the State. In 1988, a newly
elected conservative Liberal State government
further strengthened moves to turn the New
South Wales public sector into NSW Inc. The
number of Area Health Boards was reduced to
ten and these were headed by chief executive
officers (CEOs) who held Iixed-term (seven
year) contracts with the State’s Senior Executive
Service. These CEOs reported directly to the
State Minister for Health, who made it clear that
continual reappointment upon the expiry of
contracts would depend on demonstrated perfor-
mance to stay within budget and improve the
efficiency and effectiveness of their Areas.
Change was also afoot at the operational and
technological level. In the late 1980s the State
Department of Health, like its Commonwealth
counterpart, was concerned to develop a measure
of hospital output that would enable inter-
hospital/Area comparison and guide resource
allocation. To this end, it followed the example
of health authorities in the mother country
(Britain) and began constructing a Resource
Allocation Formula. Under this formula, funding
to Areas would be based upon factors such as
population size, age distribution, number of
teaching hospital referrals and hospital casemix.
The State Department also introduced a form of
specialty costing and was exploring a major
overhaul of its computerized hospital informa-
tion data-gathering systems. More generally, the
State’s Premier (who held an MBA Tom Harvard)
had directed that public sector organizations,
like their private sector counterparts, should
learn to manage “all their assets and liabilities”
and change to an accrual accounting system by
given deadlines.
As the above historical sketch indicates, the
hospitals studied from 1988 onwards emerge
from a particular antecedent world in which
state agencies (both State and Commonwealth)
now funded the bulk of the nation’s health
services. As the financial burden this entailed
came to be seen as excessive in tight economic
conditions, both Liberal and Labor administra-
tions sought to shift it among themselves (the
122
WAl FONG CHUA
Commonwealth to States and vice versa) and to
others (consumers, private insurance agencies).
At the same time, increased dissatisfaction with
the quality of public administration led state
agencies to seek higher levels of surveillance,
control and accountability. This was attempted
partly through organizational restructuring and
partly through investment in “rational”, private
sector financial and information technologies.
The Commonwealth Casemix Development
Programme was one part of this portfolio of
managerial technologies.
THE TRANSLATION OF INTERESTS OR
HOW DID THE FACT-BUILDING
NETWORK EMERGE?
Developing casemix management informa-
tion systems in hospitals, however, was not
going to be easy. As the field quotes above
illustrate, it was not altogether clear that in 1989
a form of DRG-based hospital product costing
would be viable in the hospital world under
study. There were “hawks” and “doves”. The
central question of this section is: who com-
prised the fact-building network and how did it
emerge? Did the actors come together because
of a common perception that there was an
“obvious” need for DRG-based accounting
information in hospitals? Or was this need
fabricated?
The network consisted of four major groups
of actors - academics, hospital personnel,
State Department officials and Commonwealth
bureaucrats. Let us begin with the University
Project Team, for in the narrative it is made to
act as the “obligatory passage point” (Callon,
1986), the channel through which all other
network interests must pass. The Team consists
of four principals - two internationally known
DRG advocates, a political scientist and myself.
The DRG experts were well connected to the
Australian communities of health economists
and medical record practitioners and to the
American designers of DRGs. Their careers were
rooted in DRG research and prior to 1988 they
had received state funding to investigate the
applicability of DRGs for use in Australian
hospitals. Only one of the experts, a health
economist, had some knowledge of DRG-based
accounting but even he had had no hands-on
experience of installing the YCM in organiza-
tions. These two DRG experts might be described
as believers, being convinced that DRGs could
form the basis of a more efficient and equitable
system of resource allocation to public hospitals.
As they saw it, present allocative mechanisms
were overly influenced by interest group poli-
tics and in need of a rational, scientific basis. At
the very least, a more objective formula would
help identify when, how and whose politics
influenced particular outcomes. The other two
members of the Team were not familiar with
DRGs per se, but wished to study change
processes in organizations. I, in particular, was
more concerned to investigate than to develop
DRG-based accounting.
In a proposal dated 2 March 1989, the Team
wrote an early draft proposal for “the develop-
ment of integrated casemix management infor-
mation systems” that did not have a DRG
accounting subproject. Instead, it spoke of
developing casemix reports which would group
all hospital discharges by DRG and for each DRG,
show the numbers and proportions of all
discharges and of all bed days, the mean and
median lengths of stay and the coefficients of
variation of lengths of stay. Given the Team’s
expertise in health economics, medical record
practice and organizational analysis, this was
perhaps unsurprising. However, by the end of
March 1989, the final submission listed casemix
accounting as Project 1 (Projects 2 and 3 were
as originally proposed in the draft submission
of 2 March 1989). Why the change? Primarily
because the Team had been persuaded by one
of its DRG experts that hospital participation
and Commonwealth support would be more
difIicult to secure in the absence of concrete
benefit. In order that organizational change
might be observed and studied over time,
extensive access would be required. That
privilege to merely observe was unlikely to be
given and supported unless administrators and
funders received useful managerial information
FABRICATION OF ACCOUNTING IMAGES
123
in return. In addition, applied or action research
was common in health-care research and had a
certain academic legitimacy. Finally, it was
pointed out that without enabling hospitals to
develop new types of accounting information,
there might be no change to observe for
sometime.
The generation of product cost information
thus became a key “interessement” device,
which would tie together the Team’s interests
in development and research and also enrol
strategic others. Such data, for example, would
be of interest to a managerialist Commonwealth,
which as pointed out earlier, was keen to
allocate resources rationally, reward efficient
producers, eliminate waste and generally manage
its budget deficit partly through holding hospital
expenditures in check. Product cost information
would also interest the newly appointed Area
CEOs and general managers of Hospitals C, N
and E. All three Areas faced budgetary pressure
in the form of state-capped budgets, constant
state pressure not to overspend but to make
productivity gains, and inter-Area and inter-
hospital competition for scarce resources. In
1988/89, these CEOs had recently been placed
on short-term performance-based contracts.
They had witnessed the very public sacking of
hospital administrators for financial mismanage-
ment and the acrimonious closure of inner-city
hospitals much like their own. Decreasing
lengths of stay, changed population growth
patterns and associated electorate pressures had
convinced State officials that city hospitals
should be rationalized, their bed capacity
reduced and resources diverted elsewhere in
the city. Knowing more accurately what their
products were and how much they cost could
help these hospital administrators to manage
with less, and possibly get more from the
government.
However, while the University’s projects
appeared useful in principle, particular financial
and organizational pressures meant that in
practice the level of initial support among the
three hospitals was quite uneven.
Hospital E was in the throes of a major
redevelopment, with one of its premier teaching
and research hospitals being threatened with
relocation to another site. The University’s
proposals appeared incidental as administrators
at the hospital and Area level constantly
composed and recomposed capital castings
during long negotiations with a State Department
of Health focused on rationalization. However,
Hospital E also had a general manager who
intended to come in on budget, who needed to
close approximately 100 beds, to cost the
savings that these closures would bring, and who
wished to decentralize his organization by
setting up clinical divisions (modules) which
would eventually hold clinical budgets and be
held financially accountable. The University’s
projects thus appeared to be of some relevance
as they promised to reveal a picture of output
and associated costs. Also, Hospital E was one
of the main hospitals used by the University for
teaching and research purposes and thus was
tied to the University Team through other
networks.
If support from Hospital E was ambivalent,
that from Hospital N was even more so. Unlike
either E and C, N was not embarking on major
structural change. It was not introducing matrix
organizational structures nor the concept of
general management and N was managed
through a traditional management committee
comprised of directors of medical, nursing and
administrative services. However, like the other
Areas, its CEO was keen to come in on budget
as the Area had begun life “ $13 million in the
red”. It was also seeking to make department
heads financially accountable and had recently
sought to “overhaul” its financial reporting
system. Hence, when approached, the hospital
was in the process of evaluating, buying and
implementing expensive, integrated hospital
information systems which, inter alia, would
enable the provision of better financial, casemix
and quality assurance information. Prima facie,
then, Hospital N appeared to be one step ahead
of the University Team, who, in any case, was
proposing a concept that was politically “too
sensitive” (see field quotes above).
Interest in the University’s projects, however,
was strong in Hospital C, whose Area CEO was
124
WAI FONG CHUA
familiar with and supportive of DRGs. Also, he
knew the University’s DRG experts well. Further,
similar to the other two hospitals, C faced
budgetary pressures and had closed beds,
although it was not in a state of IinanciaI crisis.
Nevertheless, the Area CEO knew that given the
State Department’s policy of “shifting beds to
the west”, Hospital C would need to radically
rethink its role in the Area and the city. Recently,
to alleviate internal resource battles, the hospital
had embarked on a policy of divisional devolu-
tion. Effectively, this meant that clinical specia-
lities were divided into divisions with designated
resources (principally nurses and beds). These
divisions, however, did not hold their own
clinical budgets and the hospital continued to
produce only cost centre reports. These were
considered “rudimentary” at best and were
“useless” for the purpose of divisional manage-
ment. The University’s projects were therefore
of interest as they might “throw some light” on
the DRGs produced by each division and
associated costs. In addition, a greater know-
ledge of the hospitals costs might assist the
hospital to accurately cost its tertiary services
which formed part of the resource allocation
formula used by the State Department. Further,
product cost information could potentially help
senior administrators to “rein in the doctors”
and make them “more cost-conscious”. C thus
became enrolled, hopeful that the University
projects could be lodged within the specific
managerial agendas of senior administrators.
I believe that acceptance by C was instru-
mental in consolidating support from the
other two hospitals. To begin, traditional inter-
hospital rivalry helped create a sense of “it is
important to keep up with the competition
and not be left out of new, innovative develop-
ments”. Hospital C had a long historical tradition
and in the past had often been regarded within
hospital circles as “the jewel in the crown”. For
Hospitals N and E, their symbolic capital would
at a minimum be maintained by working in a
select group that included Hospital C and a
well-known University team.
In addition, the University offered to help
individual hospitals make submissions to the
Casemix Programme. This meant that hospital
participation in the University’s projects not
only cost little in resource terms but also had
the potential to raise extra funds at a time of
budgetary constraint. Further, a multi-hospital
and University consortium was attract&e because
(a) it lessened the degree of inter-hospital
competition for funding, (b) the University team
possessed a level of expertise about DRGs which
was not easily found within the hospitals; and
(c) the University team was trusted as a neutral
mediator who would ensure that sensitive
hospital-specific information would either be
kept confidential or be fairly traded among the
three hospitals. The two DRG experts, in parti-
cular, were well known to stat3 from all three
hospitals. Finally, at the very least, the University’s
projects would help hospital administrators
keep abreast of Commonwealth attempts to
control health expenditure via the use of
casemix-based management tools. Given the
rapid changes of the recent past at both State
and Commonwealth levels, administrators most
likely thought it politic to be in the know.
Hospital administrators were not the only
enrolled actors within the fact-building network
whose commitment was initially lukewarm.
The State Department of Health also had
to be persuaded and Commonwealth4tate
sensitivities “on who was doing what in whose
territory” had to be clarified to the mutual
satisfaction of both levels of bureaucracy (I was
not present at such discussions). Eventually, the
State Department came to be seen as having a
legitimate role in the co-ordination, evaluation
and monitoring of all successful State submis-
sions to the Commonwealth programme. This
process of role definition took time partly
because the State Department initially lacked
enthusiasm for and technical expertise on the
issue of casemix. Also, the Commonwealth’s
initiative coincided with a period of extensive
rationalization and change within the State
Health Department, during which time staff
experienced considerable stress and ambiguity.
Finally, the Commonwealth bureaucrats -
what did they want for their investment? The
provision of monies for development as distinct
FABRICATION OF ACCOUNTING IMAGE-5 125
from research and the pattern of funding that
did eventuate indicate that they essentially
sponsored enabling projects by believers, that
is, work that would aid the implementation of
a certain type of casemix control in the future.
Thus they gave money for the hospital world to
be prepared, for potential problems to be
identified and solved and for transplanted
knowledge (American DRGs) to be modified
to local conditions. Direct funding of develop-
mental work also circumvented unconvinced
State bureaucrats wary of Commonwealth inter-
ference and would help constitute a new
knowledge and commercialized reality within
the hospital world.
Inferring such a purpose, however, does not
ailow a iiuther inference that this group of actors
was confident of success or that they knew
precisely how and when their aim might be
achieved. The above account highlighted how
university-located knowledge experts, who were
convinced of the utility of a technology, could
become connected to the policy agendas of state
agencies. In this case, the practice of govem-
ment was mediated, assisted and ultimately
legitimated by the authority of external experts.
But for a new technology of surveillance to be
institutionalized in organizations, that was only
the beginning of a long battle. To achieve
victory, new converts would need to be formed,
critics silenced, competitors overcome, sceptics
convinced and the technology shown to work
in many, diverse workplaces. As Latour (1987)
points out, the fate of a technology does not lie
in the hands of designers or initial supporters
but with those who come after - actors, who
are often possessed of different interests and
subject to different pressures. Thus, I believe
that while some Commonwealth bureaucrats
wanted to develop an Australian form of casemix
management and had initial expert allies, in
1988, the widespread adoption of such mana-
gerial technologies in the foreseeable future was
by no means certain.
In essence, this section argues that the fact-
building network had to be carefully built up.
Clearly, there were actors who did want DRG-
based accounting information - a group of
Commonwealth officials, and they had money
to tie in advocates who believed in the
technology. But there was also variety in
interests and levels of expertise within the
network. The Commonwealth wanted know-
ledge about costs calculated on a casemix basis.
The University team wanted funding for both
research and development purposes and inclu-
ded experts who believed in the utility of DRGs
but who were not masters of the YCM. The
hospital administrators and State officers wanted
better financial information (this interest was
itself largely shaped by State and Commonwealth
policies in the 1970s and 1980s) but there was
no precise master-plan to achieve this better
state. In addition, each hospital and the State
Health Department faced particular pressures
on their resources; in two instances major
reorganizations were in process and all hospitals
reported budgetary constraints. Also, as the field
quotes illustrate, some hospital personnel were
not convinced that DRGs could be sold at that
particular time (especially to doctors). And
casemix itself was a notion that was relatively
new to the hospital world.
The origins of the network were thus complex
and diverse; much was happening besides the
possibility of changing internal accounting
systems within a group of hospitals. Some of
these processes acted as catalysts for accounting
change whilst others served to distract. The
allies and the doubtful had to be corralled
and their divergent concerns and levels of
commitment translated into Projects 1, 2 and 3.
The University Team had to convince itself that
Project 1 was necessary in order to enrol the
Commonwealth and hospital administrators.
The latter had to be persuaded that their interest
in obtaining more resources, decentralized
management, efficient financial control and
coming in on budget could be met by co-
producing a knowledge about DRG-based costs.
And State bureaucrats had to accept that
Commonwealth interest in the casemix manage-
ment of hospitals, which were located on their
turfwas a fai t accompl i . The Casemix Programme
was not going to go away, it had resources
independent of the State and hence the sooner
126 WAI FONG CHUA
they (the State) assumed some degree of control
the “better”. In short, the reality of inadequate
DRG-based accounting was not uppermost in
the minds of the majority of network members.
Reality came later. First, there were a number
of diverse “conditions of possibility and exis-
tence” - discontinuous state involvement in
the financing and administration of health care;
shifting macroeconomic conditions which were
perceived as more and more adverse; an
increasingly managerialist state bureaucracy at
the Commonwealth and State levels; changing
patterns of population growth and distribution
of voters; emergent professional groups (medi-
cal administrators and nurses) keen to maintain
or extend their professional territories his-ci-vis
older professional groups (clinicians) and finally,
academic experts with varying levels of interest
and expertise in DRG-based accounting.
MAINTAINING CHAINS
Latour points out that “interessement
devices” are those that are interposed between
the network members and all other competing
entities that threaten to weaken or break
network linkages. The single most important
interessement device in this case-study was the
Commonwealth’s approval of the consortium
submission. Once the money came, the network
was secured at least for two years. More
specifically, the chains between the consortium
and the Commonwealth and between the
consortium and the State Health Department
became strengthened by accountability rela-
tionships. Now the consortium had to report
back to government agencies on how they had
spent the money given and what was achieved.
The grant not only strengthened existing links.
It created new relays that in effect made the
consortium larger. The grant enabled the
hospitals and the University to employ people
who specifically worked on the project and to
buy project-specific hardware and software. In
addition, subgroups for specific projects and
committees began to form and meet regularly.
Times had to be set for these gatherings, rooms
found to house them and the new project staff.
By early 1989, the consortium projects had not
only a material presence but also a temporal
rhythm in both the University and the hospitals.
Further, word travelled round to other hospitals
within the city and at least three other hospitals
formally requested to join the fact-building
network.
Links were also extended nationally and
overseas. By virtue of the grant, the consortium
became a part of an Australian casemix commu-
nity, that included all other project workers
supported by the Casemix Programme. All
funded researchers met once a year, to present
reports of progress or results of completed
projects. In addition, DRG experts in other
Australian states were invited to present their
experience with DRGs generally and the YCM
specifically. Further, because several overseas
countries had been using DRGs for a variety of
reimbursement and internal hospital control
purposes, numerous overseas experts were
invited by the University team to discuss the
impact of using DRG-based information systems.
Those who visited the consortium included Bob
Fetter (the retired Yale professor who helped
invent DRGs), several of his colleagues, the
chairman of the United States Prospective
Payment Assessment Commission, an American
professor of nursing and several British DRG
experts. The hospitals also Independently invited
other DRG and costing experts to assist them
with their specific projects.
Money and allies were Important aids in the
battle to create product costs but as Callon
( 1986) points out, interessement devices do not
necessarily lead to actual enrolment of network
members. To describe enrolment is to describe
the group of “multilateral negotiations, trials of
strength and tricks” (Callon, 1986, p. 2 11) that
accompany interessement devices and enable
people to view the fact-buiiding exercise as a
success. For the network in question, one of the
key issues was how the non-tangible concept of
DRG-based costs could be transformed into a
series of statements that could be perceived as
more certain: DRG-based product costs coulcc
be obtained without great expenditure of
FABRICATION OF ACCOUNTING IMAGES 127
resources; the information W& reliable and
coul d be used by people such as hospital
administrators, State and Commonwealth bureau-
crats; it woul d stand up to the scrutiny of
potential debunkers such as doctors, and their
expected opposition could be overcome through
appropriate educational activities and the support
of other groups (such as nurses).
In order to obtain these types of certainty
statement, the most crucial battle was to enrol
the YCM, which was stored within a suite of
computer programmes. If the Model was to be
successfully enrolled, it had to be shown to
work. But this successful working was not easy
to achieve partly because at the beginning of
the project, even the University researchers who
were perceived to be DRG experts did not know
the Model in detail. More importantly, the Model
presumed a world that did not exist in any of
the trial hospitals. To negotiate with the Model
meant negotiating the difference, the gaps and
empty spaces between the two worlds. In addi-
tion, once an isomorphism had been achieved,
it had to be stable and capable of being achieved
repeatedly. The Model had to be a consistent
re-presentative. The following sections detail
how the Cost Modelling Group generated a
particular level of certainty about DRG costs
using the YCM. The first section describes the
technical structure of the Model itself and its
potential to change modes of seeing, thinking
and talking about hospitals and patients. The
second section focuses on the activities of the
Group as they seek to make the HOSPITAL
embedded in the Model isomorphic with
Hospitals C, N and E.
Fern many to few - i nscri pti ons wi th
t-e-presentati ve potenti al
To the end-user, the YCM presented on the
video display unit of a personal computer as a
series of screens that asked for information. The
user could not access the logic of the Model
which was located within the source code of
the computer language in which the pro-
grammes were written. But this logic could be
diagrammatically depicted, shown at costing
workshops and placed in texts (see Fig. 1).
Allocate Overhead Costs to Patient Care Cost Cenhes
Inpatient Fractions
Remove Non-Inpatient Costs
1
Allocate Remaining Costs -
Patient Data
to Inpatients
Teaching
Research
- Other Non-Inpatients
Service Consumption
Weights
1
DRG 470
Fig. 1. Yale cosf model: the core working principles.
128 WN FONG CHUA
This visual representation was important for
a number of reasons. First, there was only one
actor within the fact-building network who
could understand the computer language and
translate between it and English, the computer
specialist employed by the University. In order
to persuade fellow fact-builders (for example,
medical records practitioners, hospital accoun-
tants), the Model could not remain a total black
box because if it did, actors would not quickly
understand why they were doing what they
were doing. They would be unable to see how
adding up a set of numbers in a certain way and
discarding others fitted into the big picture.
Second, seeing and knowing that the software
embodied a certain logic paved the way for an
end-user to begin to talk with the personal com-
puter. For that conversation required the end-
user to initially play the role of a provider of
answers. It w oul d have been difficult to start and
maintain a conversation with a machine which
continually asked unless one was confident that
there was a rationale to the questions.
The input data required by the YCM were
collected in five’ core files that required many
thousands of inscriptions to be supplied. These
files are: CCDEFS, STATDEFS, ALLOC, COSTS
and INPAT. A more detailed description of the
meaning of the inscriptions as they appeared
onscreen is presented below:
CCDEFS_ 1 1 Maintenance
10 4 Housekeeping
101 2 Nursing Administration
202 Surgery A
301 99 Supply
Interpretation: The iile Cost Centre Defini-
tions (CCDEFS) asked for information on the
total number of cost centres that a hospital had.
The tist column was a code number for the cost
centre. The second was the code number of an
allocation statistic, essentially a cost driver that
was used to allocate the costs of an overhead
cost centre to a final cost centre. A blank in the
second column denoted that a particular cost
centre was a iinal cost centre whilst a 99 denoted
a descriptive header for a group of centres. The
third column contained the names of the cost
centres.
STATDEFS-1 Total Admissions
2 Total Number of Full-time
Equivalent Staff
3 Total Operating Costs per Cost
Centre
4 Total Beddays
Interpretation: This iile required a user to give
a definition of each
codes.
of the allocation statistic
ALLoc:_ 12 3 4
1 Maintenance 0 29 260,980 0
10 Housekeeping 0 35 105,800 0
10 Nursing Admin 0 10 87,368 0
202 Surgery A 100 12 168,000 2300
Interpretation: This iile was a matrix of all the
cost centres by the total value of their cost
drivers. For instance, reading across row 1 told
the reader that there were 0 admissions, the total
number of full-time equivalent staff was 29,
the total operating costs of the Maintenance
Department per the general ledger was
S260,980 and there were no beddays Incurred.
COSTS _ Labour SuppliesTotals
1 Maintenance 205,789 55,171 260,980
10 Housekeeping 67,348 38,452 105,800
101 Nursing Admin 45,631 41,737 87,368
202 Surgery A 115,850 52,150 168,000
Interpretation: This file was also a matrix of
cost centres by operating costs. The above
example shows three types of costs - labour,
supplies and total operating costs by cost centre.
INPAT:
187 Medical A
198 Radiology B
212 Nuclear Medicine
3 10 Speech Pathology
1 .oo
0.78
0.45
0.23
’ There is a sixth file which enables the allocation of cost-centre costs to individual DRGs via the application of a series
of service weights. In the first run, this file was not manipulated by hospital personnel and hence is left out of the description
here. Its workings, however, are di sc ussed in the paper.
FABBICATION OF ACCOUNTlNG IMAGES 129
Interpretation: This file asked the user
to set down the in-patient fraction, that is, the
proportion of service provided by each final cost
centre that related to inpatient activity only.
These five files are listed in some detail to
show’ how the HOSPITAL embedded in the YCM
has a distinctive architecture. It was made of
three types of cost centresjpools - overhead,
final and those classified as irrelevant/the too-
hard basket. Only cost centres of the Iirst two
types concerned the HOSPITAL. These were
conceived of as producing three main types of
products - in-patient care (which had 467
product lines called DRGs), out-patient care,
and others such as research and education. The
Model’s HOSPITAL was only designed to analyse
in-patient care production. As a result, all other
types of production had to be removed from
consideration.
The process of fabricating the 467 in-patient
care product lines was as follows. First, overhead
cost centre costs were allocated to final cost
centres based on estimates of resource con-
sumption. Next, estimates were made of the
proportion of final cost centre costs that related
to the production of in-patient care. Finally, the
in-patient services of final cost centres were
divided into several types - medical, nursing,
pharmaceutical, pathological, etc., weighted and
costed by product line. For instance, assume that
Final Cost Centre 85 produced only DRGs 1, 2,
and 3 and had incurred 5660,000 of nursing costs.
A set of American service weights indicated that
DRGs 1, 2 and 3 consumed nursing resources
in the ratio 1 : 2 : 1. The $60,000 of nursing costs
would then be allocated to DRGs l-3 in
accordance with this ratio. And by manipulating
the Model, one could then obtain a series of
cost reports of, for example, the average cost of
each DRG or the average cost of each cost
centre.
Not only did the HOSPITAL possess a parti-
cular structure and set of production processes,
it embodied a specific set of images of patients
and hospitals. A patient was not pictured as an
individual who had been diagnosed as suffering
from a particular sickness. Instead he/she was
viewed as an exemplar of a class of diagnoses
(say, fractured femurs). Since this class consti-
tuted a product line (DRG 325) the patient
became a product of this type. By extension,
hospitals were depicted as multiproduct Iirms
and health-care provision as a process of
manufacturing 467 product lines.
This architecture, set of production processes
and culture made the YCM into a potentially
powerful rhetorical and representational device.
First, the Model was extremely mobile and
relatively immutable. It fitted on a single
computer disk which operated on a standard
personal computer. Also the licence fee for using
the software was modest. Thus, copies of the
Model could be reproduced cheaply and quickly
installed in all trial sites. There was no need to
purchase expensive hardware.
Second, the YCM had the ability to reduce
the confusion and complexity of human acti-
vities within a hospital into a set of finite, visual
inscriptions in two-dimensional space. Patients,
doctors, nurses, cooks, cleaners and their
diverse activities (operating, writing care plans,
freezing meals, mopping floors) were now
re-presented as a set of inscriptions that
collectively made up the Model’s HOSPITAL.
This re-presentation was achieved not by a
cumbersome manipulation of three-dimensional
objects but by handwork, by using the accoun-
tant’s two-dimensional images of the hospitals.
That is, the YCM took the hundreds of existing
accounts in the hospitals’ general ledgers and
recombined them into a new mobile called the
HOSPITAL. There was also little need to collect
additional information that was not already
contained within existing hospital databases.
Third, owing to the use of number, formulae
and standardized units of quantification (for
example, the dollar value of resources expended )
to represent the activities of hospitals, the YCM
could be used to re-present all three trial
hospitals. Indeed, any hospital anywhere in the
world at any time could, theoretically, be
simplified and made in the image of the
hospital. The software was immutable and non-
perishable.
Fourth, the YCM offered a concept of
products and product costs that was stacked and
130
WAI FONG CHUA
supported (both in effort and money terms)
by many prestigious individuals, groups and
institutions. The group of 467 DRGs used by the
Model had a distinguished lineage, being a
refinement of an earlier group of 383 created in
the 1970s. Both groups of DRGs were manu-
factured at Yale University via a complicated
computerized statistical procedure (see Health
Care Financing Administration, 1983, for more
details). Their creation involved a stratified
sample of 1.4 million medical records chosen
from 325 American hospitals, the set of 10,171
ICD-9-CM diagnosis codes, panels of doctors and
the development of a specialized, computerized
partitioning algorithm called AUTOGRP. DRGs
were also supported by auxiliary software that
had been purpose-built by multinationals to
enable swift and accurate encodification of
medical record inscriptions into a DRG number
(a version of this software had been distributed
to all trial hospitals). Further, since 1983, DRGs
had formed the heart of a Prospective Payment
system in the world’s most technologically
advanced and expensive health-care system (the
United States). While the evidence about the
effects of that reimbursement system were
ambiguous( Chua & Degeling, 1993) they could
be read as indicating success (defined as
effective cost control). In short, pri ma faci e
DRGs appeared as the end-point of a cascade of
inscriptions all carefully combined. Many traces
had been expertly made into few. And to
question them would have required mounting
a battle against a complex system of spokes-
persons, both Australian and overseas, who
ranged from academics, consultants, hospital
professionals to government bureaucrats, etc.
Fifth, the YCM had the ability to generate
summarized comparisons in the form of visual
tables and reports such as Table 1.
Table 1 shows a sample tabulation of twenty
DRGs by different sites. This single piece of
paper attested to the potential persuasiveness
of the YCM at several levels. It showed that the
Model would not overload the reader with too
much information. The centre of calculation
(the Model) had already processed many
diverse inscriptions over numerous cycles to
produce space-time compressed summari es.
One number spoke for and represented whole
distributions and frequency counts of many
different elements - the patient’s age, length of
hospital stay, type of treatment, etc. And the
table seemingly stood at the end of a torrent of
prior inscriptions and appeared as the tip
of a mountain built of many facts. Such
visualization further enabled comparisons
across enormous stretches of space and time.
Now, not only could the trial hospitals be placed
on a single horizon, so could a sample of South
Australian and American hospitals. And this
vision did not necessitate travel to South
Australia or the United States. All one needed
were computer printouts of DRG costs from
hospitals in these faraway places and a YCM that
permitted cycles of accumulation. Information
could also be produced not only for a single
year but for many. And perhaps most important
of all, the evidence was quantitative not quali-
tative; it was not based on opinion, prejudice
or subjective judgement but consisted of hard,
visible numbers.
Finally and perhaps most important of all, the
image of patients, hospitals and health-care
provision embedded within the HOSPITAL
appeared to complement the new economic
rationalism of the 1980s and the organizational
structures being implemented by administrators
at Hospitals C and E. As pointed out earlier, both
hospitals were in the process of setting up
clinical modules/directorates. These organiza-
tional units were easily translated into the
language of the HOSPITAL. Essentially direc-
torates were productline divisions that would
be managed by product managers (clinical
directors) who were financially accountable to
the hospitals’ general managers. The financial
information required by each directorate for
budgetary and control purposes could easily
take the form of DRG variance reports which
would compare the efficiency of the same type
of clinical directorates in different hospitals or
the same division over time.
Numerous though these advantages were, the
YCM had yet to work in practice. The Cost
Modelling Group faced two tasks. First, it had
T
A
B
L
E

I
.

C
o
s
t

p
e
r

d
i
s
c
h
a
r
g
e

b
y

D
R
G

f
o
r

c
o
n
s
o
r
t
i
u
m

t
e
a
c
h
i
n
g

h
o
s
p
i
t
a
l
s
,

S
o
u
t
h

A
u
s
t
r
a
l
i
a
n

h
o
s
p
i
t
a
l
s

a
n
d

U
.
S
.

M
e
d
i
c
a
r
e
’

D
r
g

N
O

D
R
G

“
2
l
l
W

H
t
n
p
.

A

i
l
o
s
p
.

R

1
9
-
9

J
u
W
-
D
r
c
B
9

T
O
V

C
O
W

T
o
V

C
O
S
V

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

t
l
o
s
p
.

c

H
o
s
p
.

D

H
o
s
p
.

E

H
o
s
p
.

F

H
o
s
p
.

G

S
.
A
.

H
a
s
p
s

U
.
S
.
M
e
d
i
c
a
r
e

J
u
l
B
P
-
D
r
c
R
9

1
9
B
R
-
8
9

1
9
8
S
9
0

l
9
B
9
-
9
0

1
9
B
!
w
w

1
9
B
B
-
B
9

1
9
0
9

T
o
t
/

c
o
s
t
/

T
o
V

C
O
W

T
O
V

C
O
W

T
o
t
I

c
o
s
v

T
o
V

C
O
S
W

T
o
V

C
o
s
W

C
O
W

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
x
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
c
h

d
i
s
e
h

d
l
s
c
h

“
0
.

s

“
0
.

s

“
0
.

s

n
o
.

s

“
“
.

S

“
C
L

S

“
0
.

S

“
0
.

s

W
.
S
.

2
0
0

1
.

C
R
N
I
O
T
,
A
>
l
7
-
T
R

2
.

C
R
N
I
O
T
,
A
>

1
7
+
T
R

3
.

C
R
A
N
I
O
T
O
M
Y
A
C

1
8

.
I

S
P
I
N
A
L

P
R
O
C
E
D
R
E
S

5

X
T
R
A
C
R
N
L

V
A
S
C

P
R

6
.

C
A
R
P
L

T
U
N
N
E
L

R
L
5

7
.

P
R
P
W
C
R

N
R
+
O
T
+
C
C

8
.

P
R
J
’
H
K
R

N
R
+
O
T
-
C
C

9
.

S
P
I
N
A
L

D
L
Y
I
N
J

1
0
.

N
R
V
S

S
Y

N
E
O
P
L
+
C
C

1
1
.

N
R
V
S

W

N
E
O
P
L
-
C
C

1
2
.

D
E
G
E
N
R

N
E
R
V
S

D
I
S

1
3
.

M
T

S
C
L
E
W
C
R
B
L

A
T

1
4
.

S
P
E
C

C
R
B
R
V
S
C

D
I
S

I
S
.

T
R
A
N
S

I
S
C
H
E
M

A
T
T

1
6
.

N
O
N
S
P

C
B
V

D
I
S
+
C
C

1
7
.

N
O
N
S
P

C
B
V

D
&
C
C

1
R
.
C
R
A
N
V
P
R
P
H

D
S
+
C
C

1
9
.

C
R
A
I
W
P
R
P
H

D
S
.
C
C

7
5

2
5

4
9

7
2

4
R

2
4

1
0
1

4
1

1
9

5
6

6
8

2
1

3
1
3

1
5
6

7

1
3

7

R
R

1
I
.
0
3
5

6
4

1
0
.
1
6
1

1
5
.
4
6
0

1
4

1
5
,
7
4
3

1
3
,
5
9
0

2

1
2
.
B
6
2

I
6
,
0
6
4

I
I

7
.
9
O
R

6
.
0
2
3

2
5

5
.
7
8
7

I
.
5
0
1

1
9

1
.
9
0
4

I
R
,
O
6
4

6

9
,
6
0
5

2
,
7
9
0

I
7

3
,
1
8
1

9
,
8
4
5

5

4
,
0
6
5

6
,
5
0
7

7

4
,
4
0
5

2
.
7
2
2

5

2
,
2
2
4

4
.
7
0
3

2
5

4
.
3
6
4

2
.
7
6
5

7

3
,
2
1
5

6
.
0
2
5

8
3

5
,
4
7
0

2
.
0
6
6

3
3

2
.
7
8
7

4
.
1
1
6

5

4
,
2
0
6

2
,
7
6
6

2

3
,
3
6
1

4
,
9
1
7

1
2

6
.
0
1
2

l
f
l

7
,
5
8
4

3
0
1

I
O
.
1
3
4

3

1
1
,
1
6
9

3

2
5
.
7
2
9

5
8

1
5
,
6
0
2

3

1
7
,
2
0
9

2

9
.
B
6
3

2
2

1
3
,
4
2
1

0

0

6

4
.
3
3
9

2
7

6
,
4
2
0

I

4
,
6
4
9

3

6
,
4
4
2

5
1

5
,
0
4
0

2

3
.
9
1
6

2
1

1
.
3
7
5

2
2

1
,
7
2
0

I
6

2
.
0
0
9

0

0

I
8

9
,
4
7
2

2

3
7
.
3
4
7

I
I

2
,
5
0
0

8
1

2
,
3
9
4

2
7

3
,
5
7
4

3

4
,
1
7
1

5

5
.
9
9
4

1

1
,
4
2
2

4

3
,
2
2
5

4
0

3
.
6
6
B

1
4

5
.
4
2
5

4

1
.
3
B
8

6
0

3
,
1
9
6

3
1

2
,
6
2
6

1
8

3
,
3
5
5

1
5
4

3
,
4
9
3

1
7

4
,
4
2
2

0

0

6
0

3
,
B
O
3

2

4
.
7
9
0

1
2
6

4
,
3
7
6

2
7
5

5
,
7
6
9

1
9
2

6
.
1
7
3

3
2

1
,
9
0
4

7
8

4
,
6
2
1

3
8

2
.
6
5
6

0

0

5

5
,
0
9
5

5

6
,
3
5
0

2

1
,
6
5
0

I
I

2
,
7
0
3

0

0

3

3
,
2
4
0

2
5

5
,
7
1
7

4

2
,
6
4
2

3
.
8
8
1

3
7

2
,
3
2
0

1
2

N
B
:

D
a
t
a

a
r
e

e
x
p
e
r
i
m
e
n
t
a
l

a
n
d

a
r
e

s
u
b
j
e
c
t

t
o

e
r
r
o
r
s

a
n
d

o
m
i
s
s
i
o
n
s
.

2
0
5

1
0
.
9
8
1

4

2
1

1
5
.
4
6
5

0

1
4

B
.
9
2
2

1
3
5

3
R

1
2
.
4
8
4

2
7

5
9

6
,
9
6
4

0

5
1

2
.
3
0
1

0

I
O

6
8
6
8

4

I
I
7

3
,
l
I
R

4
9

4
9

2
1
.
9
0
9

I
5

R

4
,
4
5
7

I
3

1
3

3
,
2
0
5

6
9

6
4

5
.
0
7
9

3
2

5
1

3
.
4
9
5

2

1
1
0

8
,
1
3
0

I
4

3
3

2
,
0
4
H

0

I

4
.
8
%

I

4

3
,
5
8
9

0

1
2

3
,
4
3
0

0

9
1

2
,
6
3
8

9

5
,
5
5
2

3
1
3

I
$
%
7

1
1
.
8
2
9

0

1
0
5

1
7
,
5
3
6

1
4
,
0
4
5

6
,
4
4
4

2
4

I

1
,
4
8
5

9
P
9
9

6
.
3
6
3

8
3

7
.
8
%

9
,
1
0
3

0

8
7

5
,
1
4
5

5
.
2
B
6

0

1
6
6

8
9
8

1
,
5
2
5

7
,
7
6
9

6
6

6
.
1
9
3

9
,
6
4
4

1
,
8
3
0

2
1
3

I
.
9
8
0

2
.
5
2
1

I

.
A
2
0

3
4

6
.
9
4
0

4
.
3
6
1

1
.
0
3
4

6
4

3
.
6
2
3

4
,
2
2
1

1
.
3
8
2

6
3

1
,
8
7
0

2
,
6
6
3

1
.
2
1
3

2
3
7

3
.
6
6
e

3
.
1
5
3

2
.
6
5
0

6
1

4
,
6
6
9

3
.
1
4
8

3
,
6
9
1

1
.
1
5
1

3
.
R
R
9

4
.
1
8
R

0

3
4
7

1
,
6
7
6

2
.
1
4
8

I
.
3
8
5

2
4

2
,
7
8
7

3
,
5
6
6

0

3
4

1
.
7
1
8

2
,
1
3
8

0

1
1
8

2
,
6
3
8

3
,
2
5
1

3
,
3
6
7

3
0
5

9
3
4

2
,
0
6
4

l
.
5
3
R

2
7
3

3
,
2
2
2

1
6

3
.
5
1
6

132
WAJ FONG CHUA
to learn how to be competent constructors of
the HOSPITAL. Producing numbers, however,
was easy. A second, more demanding job would
be the production of tables like Table 2 that
were convincing and credible to others. TO
accomplish this task, the Group had to first
convince themselves that the numbers were
credible. More specifically, the hospital-based
actors within the Group had to learn to be as
faithful as the University’s DRG experts. As
indicated above, the emergence of this hospital-
located group of believers is essential, for a
representational device, no matter how astutely
designed, requires human support. A piece of
software is weak if left to speak for itself. New
inscriptions need spokespersons - people
who genuinely believe in their utility and final,
comparative advantage over competitors;
advocates who would be able and willing to
continually make the inscriptions pass tests
posed by human and non-human enemy forces.
This proposition in no way questions the
integrity or ability of believers. Niether is it
meant to imply that faith is blind or that
believers strategically lie about the eficacy
of a new technology or that those who are
persuaded are dupes. It merely points out that
the willingness of someone to do battle on behalf
of an untrialled technology has ultimately to be
based on faith because one does not know in
advance that one will eventually win.
Following the faithful: persuading and
converting
How did the Cost Modelling Group seek to
overcome the many enemies that threatened the
workability and credibility of the YCM? What
were the strategies used and where were the
victories, compromises or defeats to be found?
One of the Group’s earliest advantages was
that it began life as a protected species. Common-
sense had dictated that its membership should
only include hospital-based stafI who were
familiar with the financial and patient informa-
tion systems of their hospitals. As a result, there
were no sceptical debunkers in the Group; only
a core of middle-level labourers assigned the
task of fabrication by their senior managers.
Also, the Group (and in particular the hospital
personnel) perceived themselves to be novices
and constituted their activities as a tentative
first run, trial, or developmental exercise. As a
result, they came together, expecting teething
problems and errors but encouraged by one
another, resolved to deal practically with those
issues that were feasible and most urgent. In one
of the earliest costing workshops held in
November 1989, the Cost Modelling Group was
encouraged by an external costing consultant R
(who had been invited by the University) in the
following manner:
First time round, we just want to learn about it [the
Model], understand it, see how it works. Or do you want
to be upfront and use the data (straightaway)? Our
hospitals wanted to go gently, gently. We wanted to get
the clinicians onside and the hospitals were right.
Have the first stage [run] done quickly, get hands-on
experience within 3-4 weeks. Then generate some
preliminary reports, then have another look at the data
and the allocation statistics. Get it done quickly
otherwise you will have to start from scratch again.
Just start playing with the Model, you can fine-tune it
later.
You have the ability to go back and massage the data
again. .So whether you put things in this account or not
is not material at this stage because the numbers are so
small. In the final washout it really isn’t going to matter.
In effect, different concepts of time were
being evoked by the consultant to persuade.
Emphasis was placed on learning to play
“quickly”, with the ability to “return” to the
Model, with dealing with the “most urgent”
problems. Such talk constructed two notions of
time. Time as a scarce commodity did not exist
within the HOSPITAL, the Model was always
there, ready to give the “right” answers if
“correct” data were appropriately entered.
There were no restrictions at the HOSPITAL, it
could be constantly revisited. Time within the
Group, however, was limited; there were
project deadlines, government agencies and
senior managers to account to, other work roles
to be accomplished. Besides, this was not the
only difference between the world of the
HOSPITAL and those of the hospitals. There
FABRICATION OF ACCOUNTING IMAGES 133
were numerous other gaps. As one accountant
pointed out, “our cost centres were not set up
to do cost modelling”. Hospital E, for example,
initially had two sets of cost centres that did not
match; there was one set for salaries and wages
and another for supplies. Also, across the
hospitals, different labels were being attached
to the same cost centre or the same label was
being used for different cost centres. Further,
the Model HOSPITAL initially could only pro-
cess input data from a total of 186 cost centres.
This limitation meant that trial hospitals (which
had, on average, 300400 cost centres) were
encouraged to combine cost centres, parti-
cularly overhead centres that were allocated to
final cost centres using identical cost drivers. In
addition, information with respect to appro-
priate cost drivers also varied widely between
cost centres and hospitals.
How were these gaps to be filled? Which
should be attacked first? In answering these
questions, the Group evolved a range of simple,
everyday decision rules. First, it fixed iixable
problems. The HOSPITAL was redesigned so
that it was capable of processing information
from more than 186 cost centres. Second, Group
members considered the materiality of the gap.
As Consultant R argued above, errors might not
matter in the “final washout”. Materiality,
however, was not a constant notion and was
defined in at least two different ways. An
item was considered immaterial if it did not
constitute a large proportion of the hospital’s
revenue or expenditure. Trust funds which were
used to finance in-patient care production were
thus excluded partly on this basis. Besides, each
hospital had several hundreds of these trust
accounts and their analysis was considered not
“cost-effective”. Materiality. however, was at
times subordinated to another aim - acting
practically. This happened on at least two
occasions. Capital expenditures were not im-
material items. They made up a sizeable
proportion of a hospital’s expenses. However,
the Group excluded such expenses from
consideration - primarily because there was
no method by which they could easily trace
consumption of capital assets by individual
product lines. And thus like the designers of the
American Prospective Payment System, the
Group decided to put all items labelled “capital
expense” in the “material but too-hard basket”.
A similar compromise was made with respect
to service weights - the conversion factors
used to allocate the costs of final cost centres
to DRGs. These factors were to represent the
proportion of the total value of the resources of
each cost centre that were consumed by each
DRG. There were, however, no Australian service
weights. As a consequence, American (the so-
called Maryland) service weights were substi-
tuted. This strategy implied that an Australian
product line (say, DRG 302) consumed the same
relative amounts of resources as an American
DRG 302. The Group accepted that such an
assumption of identical relative resource con-
sumption might not be valid since differing
treatment protocols could exist between the
two countries. Professor E had further reserva-
tions about the American weights, “. . the way
Nursing is charged in the States is very arbitrary.
It is not a very logical basis. The ICC (Intensive
Coronary Care) weights are also very inaccurate.
And there are only eight service weights” (Cost
Modelling Group Meeting, 18 March 1991).
Despite these misgivings, the Group accepted
that there was no alternative but to use these
American weights. Not to have done so would
have meant abandoning the YCM and the Group
was not ready to surrender.
A third decision rule was to seek consistency
both within and across hospitals. Thus, where a
hospital had dissimilar lists of cost centres. these
were redrawn so that the hospital had one
consistent map of its centres. Efforts were also
made to standardize cost centre definitions
across the hospitals, to collect data for the same
time period and to use the same version of
auxiliary software. Further, as indicated above,
collective decisions were made to derive consis-
tent rules for the representation of controversial
objects, such as trust funds, capital expenditure,
hospital teaching and research costs (these were
factored out because the HOSPITAL only pro-
duced in-patient care), treatment of supply, drug,
nursing and medical officer costs, etc.
134
WAI FONG CHUA
Consistency, however, like materiality was a
slippery notion. It did not only mean that a
Group decision rule was consistently applied
for the same activity in all hospitals. By the end
of the first run, the concept had become asso-
ciated with credibility because credibility came
to be defined as generating valid comparative data:
Meeting of the Cost Modelling Group, 18 March
1991
Professor E: We are interested in the credibility of the
numbers. It is not helpful to have different methods of
calculation. We could be criticised for variety.
Hospital administrator: There is always some degree
of uncertainty. The numbers are always subject to some
limitations, always. But these guesstimates from clini-
cians When are we getting some standards from
Canberra for product costing in hospitals? What about
standard cost centre definitions. I can’t wait till we get
some standard from on-high.
Comparative cost reports have already been used to set
one hospital against another. They are being used by the
Department (State Health) to define relative differences..
Another hospital administrator: But how comparable is
the data from the first run? Our costing methodologies
were so different. How will we know how to interpret
the numbers? We need to trace back -what’s in the
cost centre? Separate out components that go into Wards
(meaning ward costs). If we know the principles used
by each hospital, then .
Professor E: Before we can undertake these kinds of
comparability, we need basic comparability about how
costs are assigned
University computer specialist: But ifwe need consistency
at every level of the trace back, we need a heroic achieve-
ment. I think it is important that totals are consistent
But by late May, even the computer specialist
felt consistency at every level was important:
Meeting of the Cost Modelling Group, 28 May
1991
Computing specialist: We need some standardization of
the data in the numbers. What are the elements included
in the OT (Operating Theatres) costs? Porters? Are they
in OT or Transport? What about clerk costs? Are C.%D
(Sterile Supplies) charged back to the surgeon? Are
protheses charged to OT or wards? I need a list of how
people have done it.
As the field quotes indicate, these concerns
about credibility arose because (a) state officials
were beginning to interpret the comparative
numbers in particular ways, and (b) the Group
knew that despite their attempt at consistent
fabrication, the output of the first run contained
numerous inconsistent treatments. Each hos-
pital, for example, calculated their in-patient
fractions differently. Hospital E essentially relied
on estimates provided by professional stti,
Hospital C used a particular formula for weighting
some of their staff estimates while Hospital N
used a different formula again for their calcula-
tions. Drugs at Hospital N were costed to
individual wards and thus their cost centre
labelled “Pharmacy” did not contain material
costs. This practice was not always adopted by
the other hospitals (Minutes of the Cost
Modelling Group, 18 March 1991). There were
also daerences in the ways that hospitals rolled
up and combined cost centres (Minutes of the
Cost Modelling Group, 18 March 1991). These
gaps continued to trouble the Group:
Cost Model&q Meeting, 18 March 1991
Professor E: Isn’t the debate about the type of activity
that is going on Rheumatology may be a ditferent
activity or Rehab (Rehabilitation). Rehab in spinal units
is very different and this matters when hospitals roll (that
is, combine cost centres with the same cost drivers)
them up differently
Hospital administrator: Yes, like our Rheumatology. We
really don’t have Rheumatology because it is done at the
(Xyz) Home and we put Rheumatology into General
Medicine. Our Rheumatology would be very d8erent
to your (Rheumatology) department.
Cost Modelling Meeting, 24 April 1391
Professor E: We should have finer distinctions as the
same cost centre may actually do di@erent things. For
example, Pharmacy, it may provide tertiary or secondary
services. We just have one Pharmacy cost centre, but
they provide essentially different services.
Hospital administrator: At our hospital we have a
different method of calculating theatre costs. Some costs
get allocated back to medical or surgical specialty by
surgeon, also some VMO (visiting medical officer) costs,
etc.
FABRICATION OF ACCOUNTING IMAGES 135
Computer specialist: You should let me know now what
your different methods mean so that when I’m rolling
up the cost centres, I know what I’m rolling together.
Hospital administrator: Some porters are just dogsbodies.
Portering could be cleaning. A patient trolleyman is a
porter. A general dogsbody is a porter. Our porters clean.
Women can’t clean for mnre than 6 hours. so our porters
do it. We put porters in Cleaning. In the hrst run, they
got applied the Nursing Weight. Now it appears they
may be doing some non-inpatient work as well. And
nurses, they should get multi-rated. Like we cost them
to Ward 2 but they work 6 hours on Ward 5 but Ward
5 doesn’t get that costed to them.
At the end of the first run then, the labourers
had made a number of compromises of material
items, differences remained and the data
were inconsistent and inaccurate in numerous
aspects. Why was the Group not sufficiently
concerned to consider abandoning the Model?
Why did they continue to negotiate with it?
I am not entirely sure of the answer to this
important question. Some easy ones suggest
themselves. The Group was largely composed
of lower-level organizational members with
little iniluence. Given that they had been
instructed by their superiors to help develop
a set of numbers and their hospitals had
participated voluntarily in the consortium, they
would have persisted with the Model whatever
the difficulties encountered. Or the Group did
not really understand the arithmetic or the
statistics involved and was content to be led by
university-accredited experts.
Giving superior pressure as an explanation is
inadequate for a number of reasons. First,
it assumes that the lower-level hospital-based
members of the Cost Modelling Group were
incapable of autonomous judgement (a proble-
matic assumption) and knew that their superiors
wanted DRG-based accounting from the start.
Earlier, I had indicated that the interest of
administrators in the IJniversity’s projects was
ambivalent and differed between administrators
and over time. Administrative commitment,
therefore, was by no means a foregone conclu-
sion. This is not to deny that the continual
support of senior administrators after the release
of preliminary data did help to sustain the
network. But administrators were only one link
in the chain and the same question could be
asked at this level -why did they not abandon
the Model? Additionally, why was superior
pressure not outweighed by the anticipated
opposition of doctors? Also, while hospital-
based personnel were clearly unfamiliar with
the intricacies of the YCM when they began,
they learnt much during the iirst run and became
quite critical of other costing consultants
“peddling quick and dirty applications of DRG-
based accounting”. So, the answer(s) to “why
no surrender” needs to be more complex and
there appear several aspects to it.
First, nobody else within the Commonwealth
Casemix programme was considering surrender.
Not the Commonwealth, nor the many other
workers within the casemix community located
in universities, research institutes, hospitals and
departments of health all over the country.
Indeed, Commonwealth bureaucrats seemed
more determined and there was talk of financing
projects that would yield standardized know-
ledge; for example, a common version of the
YCIM that was more user-friendly. There was also
the possibility of several future developments
-standardized cost centre definitions, common
formulae for the calculation of in-patient frac-
tions and Australian cost weights. As a conse-
quence, the Group’s errors seemed temporary,
justified on the basis that “one had to learn to
walk before one could climb mountains”. The
actors were therefore content, able to defend
their compromises along the following lines:
they had struggled with the inadequacies of their
hospitals, been guided by well-known experts
and done what was best given constraints. The
future could only bring continuous improve-
ment, better quality data, more standardization
and fewer errors. In the meantime, their
preliminary data would not be used to actively
manage their hospitals and the first run had
constituted a useful learning experience.
There was also a sense in which the collective
labour of producing the numbers instituted
a certain solidarity and loyalty. The Cost
Modelling Group had battled against numerous
“enemies” (inappropriate database structures,
136
WAI FONG CHUA
software limitations, uncommitted others),
acquired a new language and explored the
intricacies of a new territory. This it had
achieved together. Now they too were experts,
who took pride in their work, who wanted to
get it right and who were prepared to collec-
tively defend their output against new enemies
(“sloppy” consultants offering quick fixes, unco-
operative doctors, etc). However, although they
had acquired a level of expertise, the Group was
much aware of the limitations of their first run.
It freely acknowledged that their data were
tentative and subject to a number of measure-
ment and sampling errors. However, not only
was this public confession of inadequacy framed
within the context of a better, more consistent,
future, it became a source of competitive
strength. In one of the Cost Modelling Group
presentations, the University Project 1 team
rhetorically asked “What is the true cost of a
DRG?“. To this they replied that true costs were
“myths”. Even if they did exist, it would not be
cost-effective to discover them. True costs were
thus “unknowable” and the YCM only attempted
to “simulate reality” and generate estimates of
costs that were useful for specific and defined
purposes. Hence, the goal of the Group was not
to “reveal” true costs but consistent, good
enough approximations. Such an aim aEorded a
comparative advantage against rivals such as
clinical costing, which was “relatively expensive
to implement”, and required the setting up of
computerized feeder systems which produced
a “greatly increased volume of activity and cost
data about individual patients”. Also, in places
where clinical costing had been attempted,
“there would normally be a delay of several
years before the required data are generated and
cost estimates by DRG can be produced”
(liniversity document, 1990). By contrast, the
YCM “did not require the collection of addi-
tional service utilization data about individual
patients” (University document, 1990). Compe-
titors were thus found wanting. In any event,
they were seldom serious threats as the Cost
Modelling Group did not calculate costs using
different costing technologies and then com-
pared results. In short, from the viewpoint of
the Group there appeared few viable alternative
forms of casemix accounting.
Furthermore, the types of reports (see Table
1) generated by the Cost Modelling Group
placed the three hospitals and others in a
hitherto unseen matrix of visibility. Listed
down the left-hand column were lists of
products which could be ranked in order of
relative costliness and volume of output. Across
the top were hospitals that could similarly
be ranked in order of relative cost efficiency.
These montages (Amann & Knorr-Cetina, 1988)
included items of “substantial cost diEerences”
that were first highlighted by the University
team:
DRG 49, Major head and neck procedures, Hospital E.
There is no apparent reason why the cost in this hospital
is higher than in the others.
DRG 90, Simple pneumonia and pleurisy, age > 17,
without CCs (Complicatitms and Cnmorbidities), Hospital
E. There is no apparent reason for the high cost in this
hospital relative to the remainders
DRG 172, Digestive malignancy with CCs, Hospital E.
The average length of stay for the DRG is substantially
higher in this hospital.
DRG 370, Cesarean section with CCs, Hospital D. There
is no apparent reason for the higher cost for this hospital
(University document, 1991).
Such types of comparative statements had not
been produced before. They seemed to be
saying something new, not about the HOSPITAL
but each hospital. The centre of calculation had
taken the specific detail of each hospital, laid it
out in an ordered field in the generalizable
HOSPITAL and produced specific comment
on the past activities of each hospital. This
particularistic narrative appeared useful to the
Cost Modelling Group, for at the very least it
promoted debate, instigated some form of
investigation and promised interventionist
action from a distance.
Finally, the Group’s resolve to persist was
probably strengthened when credentialed
experts told them that the YCM was built
on powerful knowledges that had defeated
competitors:
FABRICATION OF ACCOUNTING IMAGES 137
It is all a bit confusing (now) but the mathematics of
the YCM is impeccable (Professor E).
The inpatient fractions are clearly important. . . But the
Model works by averaging out errors, both internally
and across hospitals. Including errors with inpatient
tractions. The Model is very robust (Professor E).
The State Department of Health (pause) Due to
ignorance, it didn’t understand how difficult it is to try
and design a patient classification system. took, if 1
say “I believe that I can design a better system than DRGs.”
That’s a stupid belief. There is no data available. The
practical system is DRGs. Everybody is using it and it is
pretty good. (And discussing alternative measures of
casemix, the speaker said) Disease Staging is not used
in the US and interest is dying out. And the Patient
Management Categories, it is a very difficult piece of
software and there is no comparable data. DRGs is
better than all the other garbage (Costing Consultant
Dr. 0).
The data is not hard to get. How much time did you
spend on research? How much on teaching? People can
give the answers. The problem (of obtaining input data)
is psychological not technical. Crude estimates wiil do.
Pretty easy to do as long as people are committed to
knowing the answers (Costing Consultant Dr. 0).
(Individual patient costing relies) upon the production
of a greatly increased volume of activity and cost data
about individual patients. (It is) relatively expensive to
implement from scratch and there would normally be a
delay of several years before the required data are
generated and cost estimates by DRG produced (University
document explaining DRGs).
Here is a book (Russell, 1989) that is produced
independently by a prestigious research institution
that shows that the US Medicare scheme (which is
reimbursed prospectively using DRGS) is working very
well (Professor R).
These results [statistical correlations of Australian with
American data] are very reassuring in that despite all the
data problems and the other sources of error discussed
above, the DRG cost estimates display a pattern of
consistency between each hospitai (University document,
1991).
Thus, experts legitimated and contributed
additional degrees of credibility to the account-
ing numbers produced by the YCM. They
correlated the sets of DRG costs from the
different hospitals with each other and with
DRG costs from South Australian and American
hospitals and found extremely high positive
correlations (ranging from 0.74 to 0.94) which
were all statistically significant. To the experts,
these consistent patterns of inscriptions were
comforting for they implied that there was
something out there which had been hidden but
which was now being measured at least approxi-
mately. Again the future promised improve-
ment. With better measuring instruments, this
something would be captured with greater
precision and clarity.
And so, by mid-1991, the Group was prepar-
ing itself for a second run with the YCM. But
the impending battle seemed easier, they had
gone through the territory once before and
many lessons had been learnt.
CRITICAL REFLECTION
What matters of more general, theoretical
interest may be gleaned from this specific story
of account fabrication? At one level, it lends
further support to existing arguments (Burchell
et al., 1980; Nahapiet, 1988) that accounting is
a constitutive practice. In this case, accounting
did not just passively reflect an unproblematic
economic reality. Instead, it actively transformed
existing representations of health organizations
and their activities. Within the HOSPITAL,
patients became products which consumed
scarce resources, doctors - resource managers
who determined resource demand, medical
specialties - clinical directorates/product divi-
sions, hospitals - multiproduct firms, and
health-care provision - a manufacturing pro-
cess. Administering a public HOSPITAL was seen
as identical with managing a private enterprise.
In addition, the Mode1 made certain relation-
ships possible and these appeared to cross all
kinds of physical, space-time boundaries. With-
out the YCM and its accounting lens, there was
no obvious connection between a Mr Packer in
Sydney who suffered a massive heart attack, the
unseen cook who prepared his meal, the medical
records clerk who coded his file, the accountant
who did numerous tabulations and the politician
in Canberra who allocated health-care funds.
With the Model Mr Packer became an example
of a class of hospital products who consumed
138
WAI FONG CHUA
scarce resources that reduced the “national
cake”. Further, hospitals and even doctors could
be placed on a single metric and compared in
terms of their relative efficiency and effective-
ness. Such equivalences and interpretations
were relatively unknown in the trial hospitals
prior to the advent of Project 1. They became
more widely disseminated as the Cost Modelling
Group was constituted, public lectures were
held to explain DRGs and the YCM and specialist
casemix conferences organized. In effect, the
Model acted as a centre of calculation that
di rected the many types of information that had
to be brought back to the centre and reduced.
Fact-builders had to be instructed to look for
certain cost objects and relationships, to treat
certain activities as located in final cost centres
and others as residing in intermediate cost
centres; they had to feed in the data required
in particular, preset formats and to make do
with existing facilities when information was
imperfect. The YCM did not passively reflect the
trial hospitals, it actively reconstructed them in
the image of the HOSPITAL.
At a second level. this case-study explores
how accounting numbers persuade and command
consent. Traditional management accounting
texts have long argued that certain cost alloca-
tions (for example, of joint/common costs)
possess an inherently arbitrary dimension. Why
then do they persist in practice? Are decision-
makers consistently uninformed as to their
arbitrary nature? The answer has to be “yes” if
the assumption is made that reality comes first
and accounting numbers are used when they
faithfully represent a pre-existing economic
reality. However, if one dispenses with that
assumption. cost allocations may be explained
as the outcome of interests that are tied
together. In my story, accounting numbers did
not become usable facts because they faithfully
represented reality. Indeed, fact-builders expli-
citly acknowledged them to be flawed approxi-
mations. Nature or reality was not the cause of
the settlement of debate within the Cost
Modelling Group nor of people’s willingness to
persist with the Model (see Latour, 1987).
Instead controversy and perseverance was
partly a function of historical interests which
were managed, i nter al i a, by appealing to
socially defined notions of credibility, mate-
riality, practicality, consistency, etc. These terms
were not theoretical principles that were known
beforehand in determinate form. Instead, their
meanings were cut out in practice. People per-
sisted with the Model not because they knew with
great certainty that, compared with rival tech-
nologies, it gave closer approximations to reality,
but because they decided that the numbers gene-
rated were consistent/factual enough to hold
together diverse purposes. The Commonwealth
and State officials desired a rational, efficient
method of resource allocation and an unspeci-
fied level of certainty about DRG cost informa-
tion; hospital personnel wanted to come in on
budget, better financial management, higher
levels of cost consciousness among clinicians
and cost information that was credible enough
given resource constraints. And the University
team? - it believed in DRGs, was prepared to
improve the system and sought an ill-defined
amount of academic credibility. All parties
began with differing degrees of belief in the
technology and subsequent problems required
numerous decisions be made to stretch the
limits of materiality and credibility to distances
that could be traversed by the YCM. Despite
these compromises, the end results were judged
good enough for a first run. Nobody else was
giving up on the Model, there did not seem many
viable alternatives, the future promised only
improvement. In short, account-fabrication was
a decision-laden process rooted in faith and shot
through with the Social. Reality did not come
first but after socialized processes of making and
judging representations.
The rules which were used in this process
were no different from the everyday, economic
rules of managing social life. Judgements were
made as to which problems could be fixed
quickest, which would yield more benefits for
the same resource investment, which were too
hard and would have to wait till the next run.
Similarly, the persuasive strategies pursued
essentially amounted to commonsensical efforts
to marshal1 allies and defeat competitors in
FABIUCATION OF ACCOUNTING IMAGES 139
order to defend and extend a conceptual
territory. Sympathetic experts or interest groups
(administrators, certain ranks of nurses) were
called upon to attest (or educate) either to the
necessity of informed resource management in
general or to the adoption of DRG costing in
particular. Sohware was redesigned to meet the
demands of local users and hence work in more
specific settings. At the same time these settings
with their localized concerns were translated
into habitats that were more conducive for the
YCM. There was thus a process of mutual
adjustment of both software and context such
that numbers could be created by using
everyday decision rules that explained by tying
together as many settings as possible to as few
elements (for example, cost drivers) as possible
through as few representations as possible
(see Latour, 1988b). Expertise thus appeared
inextricably intertwined with commonsense or
perhaps there was not much difference between
expert and lay knowledge in the first place.
Third, the case-study supports recent argu-
ments (Hopwood, 1987; Bhimani, 1993) that
organizational accounting has diverse, contin-
gent, extra-organizational origins. Apart from a
handful of Commonwealth bureaucrats and
academics, the need for accounting change did
not present itself to network members as a
glaring problem requiring immediate redress.
Instead, the University’s Project 1 emerged due
to a particular complex of historically located
interests, persuasive strategies, available experts
and bodies of knowledge at a certain stage of
development. Changing state involvement in
health-care provision, economic recession,
interprofessional struggles, the interests of
academic experts, technological developments
in computing and statistics and the rise of an
economistic rationality in government all contri-
buted in some measure to the emergence of an
interest in DRG-based accounting.
At the same time, the case-study also suggests
that whilst the conditions of possibility and
existence of accounting change may be many,
not all conditions are of equal strength. Some
matter more than others and are better to
develop stronger metrological (Latour, 1987)
chains. In this case, Commonwealth state
agencies played a central role in helping to place
DRG cost accounting on the agendas of hospital
administrators. It was Commonwealth officials
who possessed the will to a certain truth who,
waving the banner of rational resource alloca-
tion, ensured that money flowed to expert allies
and spawned other converts. Given that the
sample consisted of publicly funded teaching
hospitals, this observation might not be of much
theoretical interest. However, what the case
does underscore is that whilst an analysis of
government cannot be confined to the study of
diIferent administrative agencies, their interests
and funding (see Miller & Rose, 1990) such
analysis should not be down-played either. State
agencies, after all, do distribute resources and
influence many aspects of everyday life in
today’s societies. But as Miller & Rose (1990)
and Miller (1991) point out, state rule is but
one specific form of wider structures of govern-
mentality and it may not originate directly or
overtly from a single headquarters with predic-
table outcomes. The case illustrates this to the
extent that Commonwealth interest in casemix
control was uncertainly relayed through diverse
networks of expert intermediaries and norma-
lizing judges who independently developed
particular regimes of truth. The Commonwealth
did not invent DRGs or the YCM. It was not
responsible for developments in computer
technology that enabled their creation. Also, the
University DRG experts started Project 1.
already convinced that such a measure of
casemix would enable more efficient and
equitable resource allocation. They were not
persuaded by the existence of Commonwealth
interest. And within the hospitals, relatively little
was known of DRGs and casemix accounting.
Given this state of affairs, it would have been
difficult to cast these protagonists (state officials,
hospital personnel, academics) as acting out an
explicit conspiratorial policy agenda to recast
patients as products, whose consumption of
scarce resources had to be continually and
carefully monitored. There is, therefore, a
particular contingency and arbitrariness in the
plot.
140 WM FONG CHUA
At the same time, the state is not faceless
either. It is not depicted as being devoid of
agency or as an “invisible” (Davidson, 1991)
apparatus that is shrouded by unintended,
comprehensive practices of surveillance and
control. Certain consequences of introducing
DRG-based accounting in Australian hospitals
may not have been intended or even thought of
by key fact-builders. And, it may have been quite
by chance that I became a member of the
University Project Team. But localized interests
also existed and links of a particular kind were
deliberately set up. At certain levels, there were
intentions of a kind. There were also people
with unequal amounts of diEerent resources -
expertise, money, authority, etc., and these asym
metries have begun to give casemix accounting
a larger space in Australian hospitals. But with
what possible effects?
It is speculative to talk of effects because
casemix accounting has not yet been imple-
mented in Australian hospitals. Nevertheless, I
would like to end by raising the question, for a
similar ascendency of accounting in health-care
sectors in the United Kingdom and United States
has prompted recent critical comment (see
Broadbent ef af., 1991; Chua & Degeling,
1993; Preston & Chua, 199 1; Bloomfield, 199 1;
Cooper & Rea, 1992). These writers argue that
such state-initiated attempts to manage fiscal
crises through the rational use of invisible
allocation mechanisms are fundamentally unjust,
lead to inequitable distributions of health
resources, colonize people’s lifeworld through
the perpetuation of a strategic rationality,
fragment people’s decision-making powers
because of the rise of elitist experts and further
disempower and make silent patients, who
increasingly are represented by images made by
doctor, nurse, accountant and lawyer experts.
Some of these consequences may not occur here
for the structure of the Australian health sector
differs in important respects from that of the
British or American. But the YCM does
embody a managerialist “fetish of calculation”
(Bloomfield, 1991) that helps construct health
care not as a citizen’s right but as a multi-product
manufacturing process subject to the laws
of economics. This, in turn, consolidates bureau-
cratic tendencies to view health not as a social
issue but as part of the macroeconomic problem
of managing Australia’s large current account
deficit.
Further, in an informational society (Luke,
1989) like ours, accounting numbers such as
those produced by the YCM may increasingly
help create what Baudrillard (1984) calls a
semiotic hyperreality - a reality that is both
representation and reality. A costed DRG is not
identical to a physical entity called Mr Packer
who had suffered a heart attack. It is a paper
construct - a result of a Mr Packer who had
been codified into a set of inscriptions that were
then displayed and manipulated in machines and
finally printed on a page as part of a text. But
the appearance of a number on a page allowed
the Cost Modelling Group to talk about DRGs
and their costs as though they were physical,
tangible entities called products and this in turn
made certain people (for example, patients and
possibly doctors) subject to an instrumental,
interventionist rationality. The numbers pro-
duced by the YCM enabled fact-builders to ask,
for example, why Hospital X was taking less time
to produce DRG 302 and whether this meant it
was a more cost-efficient producer. In addition,
although the results from the Rrst run were
considered preliminary and subject to errors of
various kinds, hospital stafI had begun to
consider changing their existing information
systems to accommodate the data requirements
of the Model and to more directly monitor
hitherto invisible processes; for example, time
spent on certain procedures and the relationship
between trust fund expenditure and cost centre
productivity.
What is of concern here is that through the
processes of quantification, visualization, and
normalization a certain amnesia sets in when
accounting information is used in organizations,
Reports and tables, although titled as “subject
to errors and omissions” come to be seen as
windows (albeit small) on a hidden reality. The
HOSPITAL is assumed to have been embedded
within the hospitals all along, not created by the
application of a certain accounting lens and the
FABRICATION OF ACCOUNTING IMAGES 141
putting in place of a whole series of relays,
equivalences and spokespersons. What does this
assumption obscure? The active manner in
which the accounting technology di rected the
types of information that would be provided and
manipulated in certain ways; the faith that
supported the initial belief that the YCM is a
simulation or good enough reproduction of a
pre-existing reality; the fact that this reality
is itself yet another set of representations
(accounting costs extracted from the hospital’s
general ledger, patient data extracted from
medical records) which are translated differ-
ently. In effect, representations (DRG costs) that
refer to other representations (ledger accounts,
medical records) are now accepted as approxi-
mations to the real thing. And experts gain
greater confidence when one set of representa-
tions (Australian DRG costs) vary in a statisti-
cally consistent manner with another set of
representations (American DRG costs). But
what is so real about signs that make reference
only to other signs? Is the Group not merely
dealing with multiple sets of images that are
repackaged under different classificatory rules,
with a reality that is an already reproduced
reality? Yet it is this fabricated hyperreality that
is ironically being regarded as the hidden reality
that is being brought to light. Luke ( 1989)points
out that in democratic politics, a simulated
hyperreality of public life emerges from public
opinion polls, whose mathematical indices are
substituted in practice for “the public” itself or
its “silent majorities”. With the YCM, one
appears to have a similar simulacrum of orga&a-
tional processes, a hyperreality which in
simulating patients and hospitals becomes a
reproduced and reproducible substitute.
At issue here is not merely some minor
confusion between representation and reality or
loose talk on the part of a small group of fact-
builders but the manner in which discursive
appeals to reality can set the parameters of
debate (in economic and accounting terms) and
make one set of hyperreal images more taken-
for-granted and legitimate than another. For
once it is accepted that a certain semiotic
montage rationally, objectively approximates
the real, that closer, more consistent approxima-
tions can be delivered in the future, effort is
directed at supplementing the technology by,
for example, constructing a set of Australian
cost weights, standardizing definitions of cost
centres, and developing similar classification
systems for ambulatory care. Attention is
thereby deflected from questioning the very
basis on which that debate was first initiated.
As Chua & Degeling (1993) point out, the
reproduced reality of average accounting costs,
efficiency, etc., is only one of a number of
possible realms of social life - there are also
moral and aesthetic realities. What of them?
What are the moral and aesthetic consequences
of conceptualizing patients as costed products,
hospitals as factories and doctors as product-
line managers? When these other realities are
considered, the results may be quite ironical.
Chua and Degeling, for example, argue that
although casemix accounting was implemented
in the United States in the name of instrumental
goals, it was not possible to determine whether
these had been achieved. Instead accounting
appeared to succeed as a moral mediator (by
helping to change the moral basis of health-care
debates) and as a power-knowledge that could
potentially institute new forms of panopticon
discipline upon doctors and patients (Foucault,
1977). An appeal to rational science, thus, did
not appear to banish the hard questions. Not
only was it not possible to answer the instru-
mental questions - have costs been decreased
without affecting the quality of care, who is an
efficient or inefficient producer? - issues of
ethics remained - how much health care
should be provided to whom at whose expense,
who should be paid how much for providing
what types of service and how can patients
represent themselves in debates that ultimately
influence what types of care they receive and
how long they stay in hospitals?
EPILOGUE
The case-study finished at the end of the first
run of the Model. At that point, the casemix data
142 WAL FONG CHUA
generated had not been widely circulated to
doctors or nurses and none of the hospitals used
it for internal management purposes. The data
had essentially been presented to some senior
administrators. A second run of the Model was
actively being planned. In mid-1992, the three
Ilniversities’ projects formally finished. At pre-
sent, the Commonwealth Casemix Development
Programme continues to fund numerous projects
and the University team* remains actively invol-
ved. In addition, one of the Big 6 accounting firms
has been sponsored to develop a set of Australian
service weights. It is widely expected that the
Commonwealth will implement some form of
casemix funding in the near future. Nurses (at
least those within the original consortium
hospitals) appear to support casemix manage-
ment systems believing that they will enable
them to highlight their contribution and better
procure resources (Degeling, 1992). Doctors,
while resistant to the.use of casemix reports to
focus on their modes of practice, are also not
campaigning actively against some form of
casemix control. They are supportive of the
ideas of clinical directorates and the “freedom
to manage their own resources”. They too
perceive that casemix reports would assist in
resource acquisition (see Degeling, 1992).
BIBLIOGRAPHY
Amann. K. & KnorrCetina, K., The Fixation of (Visual) Evidence, in Lynch, M. 81 Woolgar, S. (eds),
Representation in Scient~@ Practice, pp. 85-l 2 1 (Cambridge, MA: MIT Press, 1988).
Ansari, S. & Euske, K J., Rational, Rationalizing and Reifying Uses of Accounting Data in Organizations,
Accounting Orgunizutions and Society ( 1987) pp. 549-570.
Arrington, C. E. & Schweiker, W., The Rhetoric and Rationality of Accounting Research, Accounting,
Otgunizutions and Sociefy (1992) pp. 51 l-533.
Barnes. B., On the “Hews” and “Whys” of Cultural Change, Social Studies of Science ( 1981) pp. 481-498.
Baudrillard. J., The Structural Law of Value and the Order of Simulacra, in Fekete, J. (ed.), The Structural
Allegory Reconstructive Encounters with the New French Thou@ (Minneapolis: University of
Minnesota Press, 1984).
Berr), A. J.. Capps. T., Cooper, D., Ferguson, P., Hopper, T. 81 Lowe, E. A., Management Control in an
Area of the NCB: Rationales of Accounting Practices in a Public Enterprise, Accounting, Orgunfzutions
and Society ( 1985) pp. -3-28.
Bhimani. A., Indeterminacy and the Specificity of Accounting Change: Renault 1898-1938. Accounting,
Orgunizutions and Society ( 1993) pp. 1-39.
Bloomfield, B. P., The Role of Information Systems in the U.K. National Health Service: Action at a Distance
and the Fetish of Calculation. Sociui Sfudies of Science (1991) pp. 701-734.
Bloomfield. B. P. & Best, A., Management Consultants: Systems Development. Power and The Translation
of Problems, Sociological Reoiew (1992) pp. 53.3-560.
Bloomfield, B. P., Coombs. R., Cooper, D. J. & Rea, D., Machines and Manoeuvres: Responsibility Accounting
and the Construction of Hospital Information Systems, Working paper (July 1992).
Broadbent, J., Laughlin, R. & Read, S., Recent Financial and Administrative Changes in the NHS: A Critical
Theory Analysis, Criticul Perspectives in Accounting ( 1991) pp. l-29.
Burchell. S., Clubb, C., Hopwood, A.. Hughes, J. & Nahapiet. J., The Roles of Accounting in Organizations
and Society, Accounting, Orgunizutions and Society ( 1980) pp. 5-27.
Caiden. G., Australia’s Changing Administrative Ethos: An Exploration, in Kouzmin. A. & Scott. N. (eds),
Qnamics in Australian Public Management. Selected Essays ( 1990) pp. 29-49.
Callon. M.. Some Elements of a Sociology of Translation, in Law, J. (ed.), Pow, Action and Befief A
New Sociology of Knowledge (London: Routledge & Kegan Paul, 1986) pp. 196-233.
Callon, M.. Law. J. & Rip. A. (eds). Mapping the Dynamics of Science and Technology: Sociology r>f
Science in fbe Real World (London: Macmillan. 1986).
‘The author was not part of this team.
FABRICATlON OF ACCOUNTING IMAGES 143
Chapman, R.. Simple Answers, Complex Problems: Administrative Reform in Australia, 1970-85, in
Kouzmin. A. & Scott, N. (eds). Dynamics in Australian Public Management: Selected Essqvs. pp.
200-22 1 (Melbourne: Macmillan, 1990).
Chua, W. F. & Degeling, P., Interrogating An Accounting-based Intervention on Three Axes: Instrumental,
Moral and Aesthetic, Accounting, Organizations and Society, ( 1993) pp. 291-318.
Clifford, J.. Introduction: Partial Truths, in Clifford, J. 81 Marcus G. E. (eds). Writing Culture, pp. l-26
(Berkeley: University of California Press, 1986).
Clifford, J. 81 Marcus, G. E. (eds), Writing Culture (Berkeley: University of California Press, 1986).
Cooper, D. J. & Rea, D., Struggling with Accounting and Reason: Resource Management in the National
Health Service, Working paper, University of Alberta ( 1992).
Covaleski, M. A. & Dirsmith, M. W.. The Use of Budgetary Symbols in the Political Arena: An Historically
Informed Field Study, Accounting Oqanirations and Society ( 1988) pp. l-24.
Cricbton. A., Slou~ly Taking Control. 7 - Australian Government and Health Care Provision,
1788-1988 (Sydney: Allen & Unwin. 1990).
Daniel, A.. Medicine and tbe State: Professional Autonomy and Public Accountability (Sydney: Allen
& Unwin, 1990).
Davidson. A., The invisible State. Tbe Formation of the Australian State 1788-1901 (Cambridge:
Cambridge University Press, 1991 ).
Degeling, P., Policy as the Accomplishment of an Implementation Structure. Working paper. tlniversity
of New South Wales ( 1992).
Fetter, R. B. & Freeman, J. L., Diagnosis Related Groups: Product Line Management Within Hospitals,
Acudemy of Management Reviav (1986) pp. 41-54.
Fetter, R. B., Mills, R. B.. Riedel. D. C. & Thompson, J. D.. The Application of Diagnostic Specific Cost
Profiles to Cost and Reimbursement Control in Hospitals, Journal of Medical $]~stems ( 1977 ) pp.
1.37-l 49.
Fetter, R. B.. Thompson, J. D. & Mills, R. E.. A System for Cost and Reimbursement Control in Hospitals,
Yale J ournal of Biology & Medicine ( 1976) pp. 123136.
Foucault. M., Discipline and Punishment. The Birtb of the Prison, Sheridan, A. (transl.) (London: Penguin
Books, 107).
Foucault. M.. in Gordon, C. (ed.). PowerlKnowledge. Selected I ntervieus and Other Writings 19-rZ-l 3”.
(Brighton: Harvester Press, 1980).
Forester, J.. On Fieldwork in a Habermasian Way: Critical Ethnography and the Extraordinary Character
of Ordinary Professional Work. in Alvesson, M. & Willmott, H. ( eds), Critical Management Studies
(London: Sage. forthcoming).
Freeman. J. L.. Fetter, R. B.. Newbold, R. C., Rodrigues. J-M. & Gautier, D.. Development and Adaptation
of a Hospital Cost and Budgeting Method for Cross-National Use, J ournal of Management in Medicine
(1986) pp. 38-5’.
Gardner, H. (ed. ) The Politics of Health: Tbe Austrualiun Experience (Melbourne: Churchill Livingstone.
19AO ).
Gilbert. G. N. & Mulkay. M., Opening Pandora’s Box. A Sociological Anu!ysis of Scientist’s Discourse
(Cambridge: Cambridge University Press, 1984).
Hacking. 1.. Representing and I ntervening: I ntroductory Topics in the Philosopb of Nutural Science
(Cambridge: Cambridge University Press, 1983).
Health Care Financing Administration, Grunts and Contracts Report; The New I CD-9-G+f Diagtzosis-
Related Groups Class@cution Scheme (Baltimore: United States Department of Health and Human
Service, 1983 ).
Hicks. R.. Rum Regulations and Riches: The Evolution of the Australian Health Care .(‘ystem (Sydney:
The Australian Hospitals Associations, 1981 ).
Hines. R 1) Fmancial Accounting: In Communicating Reality. We Construct Reality. Accounting,
Orgunizations and Sociev ( 19HW ) pp. 25 l-26 1.
Hopper. 1’ CL Armstrong, P.. Cost Accounting, Controlling labour and the Rise of Conglomerates,
Accounting. Organizutions and Society ( 1991 ) pp. 405-f38.
Hopwood. A. G.. The Archaeology of Accounting Systems. Accounting Orgunizutions undSociep ( 19R7)
pp. LO’-23-r.
Knorr-Retina. K. The Manufacture of Knowledge (Oxford: Pergamon. 1981 ).
Kouzmin. A & Scott, N. (eds). Qytzumics in Australian Public Management: Selected Essqs (Melbourne:
Macmillan. 1990 1.
144 WAI FONG CHUA
Latour, B., Give me a Laboratory and I Will Raise The World, in Knorr-Cetina, K D. & Mulkay, M. (eds),
Science Observed Perspectives on the So&l Study of Science, pp. 14 l-l 70 (London: Sage, 1983).
Latour, B., Science in Action How to Follow Scientists and Engfneers Thug6 Socfety (Cambridge, MA:
Harvard University Press, 1987).
latour, B., Drawing Things Together, in Lynch, M. & Woolgar, S. (eds), A Repnssentatfon in Scfentf@c
Practice, pp. 19-68 (Cambridge, MA: MIT Press, 19881).
Iatour, B., The Politics of Explanation: An Alternative, in Woo@, S. (ed.), Knowledge and Refloxivfry.
New Frontfers in tbe Sociology of Knowledge, pp. 155-176 (London: Sage, 1988b).
Latour, B. &Woo@, S..L.aboratoryLlfe: TbeSocfal Constructfon ofScienti$cFacts(Iondon, Sage, 1979).
Lehman, C. 81 Tinker, T., The “Real” CuItural Significance of Accounts, Accountfng Organfiatlons and
Society ( 1987) pp. 503522.
Luke, T.W., Screens of Power. I d ogy, Domfnation and Resisrance fn fnfotmatfonal Society (Urbana:
University of IUinois l%zss5zI .
McMahon. L., Fetter, R. B., Freeman, J. L & Thompson, J. D., Hospital Matrix Management and DRG-Based
Prospective Payment, Hospital & Health Servfces Admfnfstratfon ( 1986) pp. 62-74.
Miller. P., Accounting Innovation Beyond the Enterprise: Problematizing Investment Decisions and
Programming Economic Growth in the UK in the 1960s. Accountfng, Organfzatfons and Socfety
(1991) pp. 733762.
Miller, P. & O’Leary, T., Accounting and the Construction of the Governable Person, Accountfng
Organizations and Society ( 1987) pp. 235-266.
Miller. P. & Rose, N., Governing Economic Life, Economy and Society ( 1990) pp. l-31.
Nahapiet, J., The Rhetoric and Reality of an Accounting Change, Accounffng, Organfzatfons andsociety
( 1988) pp. 333358.
Nencel, L. & Pels, P. (eds), Conshucting Knowledge (London: Sage, 1991).
Ong. W. J.. The Presence of the Word (New Haven, CT: Yale University Press, 1967).
Ong, W. J., I nterfaces of the Word (Ithaca, New York: Cornell University Press, 1977).
Opit. L. Wheeling, Healing and Dealing: The Political Economy of Health Care in AustraIia, Commlmf~
Health Studies ( 1983) pp. 23G-246.
Painter, M.. Steering the Modern State (Sydney: Sydney University Press, 1987).
Palmer, G. R., The Economics and Financing of Hospitals in Australia, Australfan Studfes in Healtb
Administration ( 1987) pp. l-21.
Palmer, G. R. & Short, S., Health Care and Public Policy: An Australian Analysis (Melbourne: Macmillan,
1989).
Poster. M., The Mode of I nformation Poststructuralism and Social Contest (Chicago: University of
Chicago Press, 1990).
Preston. A. & Chua, W. F., The Diagnosis Related Group Prospective Payment System and the Tragic
Choice of Rationalising Health Care to the Elderly, Paper presented at the Third Interdisciplinary
Perspectives in Accounting Conference, Manchester (8-10 July 1991).
Preston, A. M.. Cooper, D. J. & Coombs, R. W., Fabricating Budgets: A Study of the Production of
Management Budgeting in the National Health Service, Accounting, Organizations andSociety ( 1992)
pp. 561-593.
Pusey, M., Economic Rationalism in Canberra (Cambridge: Cambridge University Press, 1991).
Rers. S. J. & Gibbons, L., A Brutal Game: Patients and the Doctor&Dispute (Sydney: Angus & Robertson,
1986).
Robson. K.. On the Arenas of Accounting Change: The Process of Translation, Accountfng Organfzatfons
and Society ( 1991) pp. 547-570.
Robson. K. Accounting Numbers as “Inscription”: Action at a Distance and the Development ofAccounting,
Accounting Organizations and Society ( 1992) pp. 685-708.
Roes. S. J. & Gibbons, L., A Brutal Game: Patients and Doctors’ Dispute (Nonh Ryde: Angus & Robertson,
1986).
Rose, N.. Governing by Numbers: Figuring out Democracy, Accounting, Organfzatfons and Society
(1991) pp. 673-692.
Russell. L, Medicares’ New Hospital Payment System. I S it Working? (Washington, DC: The Brookings
Institution. 1989).
Scott, P.. Levers and Counterweights: A Iaboratory that Failed to Raise the World,SoclalS&dfes ofScfence
(1991)pp. 7-35.
Shapin, S.. Following Scientists Around, Socthl Studies of Science ( 1988) pp. 533550.
FABRlCATlON OF ACCOUNTING IMAGES 145
Thompson, G.. Is Accounting Rhetorical, Luca Pacioli and Printing? Accounting Orgunfzutions and
Soctety ( 199 1) pp. 572-599.
Thompson, J. D., Averill, R. F. 81 Fetter, R. B., Planning, Budgeting and Controlling - One Look at the
Future: Case-mix Cost Accounting, Heultb Serufces Reseurcb (Summer 1979) pp. 111-125.
Thompson, J. D.. Fetter, R. B. & Shin, Y., One Strategy for Controlling Costs in University Teaching
Hospitals, Journal of Medical Educatfoon (1978) pp. 167-175.
Wickings, I., Planning Clinical Budgets Using DRGs, in Bard&y, M., Cokes, J. &Jenkins, L. (eds), DRGS
and Health Cam, 2nd Edn, pp. 147-159 (London: King Edward’s Hospital Fund for London, 1987).
Yeatman, A., Public Management and the Australian State in the 1980s Austruliun J ournul of Public
Administrution (1987) pp. 339-356.

doc_500639465.pdf