You are on page 1of 44

See discussions, stats, and author profiles for this publication at: https://www.researchgate.


Enterprise Resource Planning Systems, Management Control and the

Quest for Integration

Article  in  Accounting Organizations and Society · October 2005

DOI: 10.1016/j.aos.2004.11.004

264 1,431

2 authors:

Niels Dechow Jan Mouritsen

EBS Universität für Wirtschaft und Recht Copenhagen Business School


Some of the authors of this publication are also working on these related projects:

Platform production View project

All content following this page was uploaded by Jan Mouritsen on 04 November 2017.

The user has requested enhancement of the downloaded file.

Accounting, Organizations and Society 30 (2005) 691–733

Enterprise resource planning systems,

management control and the quest for integration
a,* b
Niels Dechow , Jan Mouritsen
Said Business School, Park End Street, Oxford OX1 1HP, UK
Department of Operations Management, Copenhagen Business School, Solbjerg Plads 3, 2000 Frederiksberg, Denmark


This paper analyses how two companies pursued integration of management and control through enterprise resource
planning (ERP) systems. We illustrate how the quest for integration is an unending process and it is produced concur-
rently and episodically. Integration is not only about ÔmereÕ visibility and control at a distance. ERP systems do not
define what integration is and how it is to be developed, but they incur a techno-logic that conditions how control
can be performed through financial and non-financial representations because they distinguish between an accounting
mode and a logistics mode. A primary lesson from our cases is that control cannot be studied apart from technology
and context because one will never get to understand the underlying ÔinfrastructureÕ—the meeting point of many tech-
nologies and many types of controls. ERP systems are particularly interesting for what they make impossible, and our
cases illustrate how the two organizations in the quest for integration mobilized a number of Ôboundary objectsÕ to over-
come systems-based Ôblind spotsÕ and Ôtrading zonesÕ. The paper points out that management control in an ERP-envi-
ronment is not a property of the accounting function but a collective affair were local control issues in different parts of
the organization are used to create notions of global management.
Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: ERP; Integration; Management control; Control systems; Visibility; Representation; Actor-networks; Boundary objects;
Classifications; SAP

Prologue some from the R/3 package to do the supply

chain. Now, the question is what do we want
ÔBy now, we have implemented virtually all integration to be all about? (Informant from
modules from the R/2 package, and even TimeCorp, one of our research sites 1)

Corresponding author.
E-mail address: (N. Dechow). No. 2 during a pre-study meeting in November 98.

0361-3682/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
692 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

Introduction 2001; Hunton, McEwen, & Wier, 2002). Others re-

main sceptical (e.g. Poston & Grabski, 2001).
Enterprise wide resource planning systems There are those who argue that the new ERP
(ERP systems) attempt to integrate all corporate technologies illustrate the potential to become
information in one central database, they allow complete calculation machines governing all
information to be retrieved from many different activities and affairs of the firm. Cooper and
organizational positions, and in principle they al- Kaplan (1998) for example envision profound ef-
low any organizational object to be made visible. fects on and for management control. Such ef-
It has been suggested that such systems facilitate fects, however, have yet been difficult to sustain
unprecedented levels of organizational integration with the available empirical evidence. Surveys
(Davenport, 1995, 1998, 2000). Even if this is often (Booth, Matovsky, & Wieder, 2000; Granlund
appealing to firms, it is a formidable task. Some & Malmi, 2002; Spathis & Constatinides, 2002)
times this is justified as an investment (Deloitte suggest that ERP systemsÕ impact remains Ôvery
Consulting, 1998; KPMG Consulting, 1997; PA moderateÕ, partly because they are not typically
Consulting Group, 1999), and sometimes it is dis- designed with change in mind. They replicate
missed due to the complexity of integration (Ci- the structure of the existing systems. Others there-
borra, 2000; Hanseth, Ciborra, & Braa, 2001). fore suggest that ERP systems are enormously
Integration is surely at stake here. Is it possible powerful juggernauts that not only may be diffi-
and interesting to integrate the firmÕs activities by cult to control but that eventually also may strike
information systems? And will this enable manage- back (Ciborra, 2000; Hanseth et al., 2001). The
ment control? These are the questions that possibly disruptive effects of integrated informa-
have been suggested in the past by various ap- tion, which will cause disintegration, frighten this
proaches among which we can identify at least set of authors.
three strands. In contrast, a third strand of literature is
One strand of literature on ERP says that firms now emerging that is concerned with how ERP
implementing ERP systems (have to) go through a technologies are made to work as ÔsystemsÕ. Here
learning curve and then benefit from their invest- the system is seen in the context of numerous
ment (e.g. Ross & Vitale, 2000). This strand of lit- organizational concerns and conditions that play
erature builds on the Ôstage-maturity modelÕ themselves out in complex ways. Quattrone and
(Nolan, 1979; Hirschheim, Earl, Feeny, & Lockett, Hopper (2001) for example argue that ERP never
1988), which in spite of its criticisms (e.g. Benbas- stabilizes and that change is constant. Caglio
sat, Dexter, Drury, & Goldstein, 1984; Holland & (2003) and Scapens and Jazayeri (2003) relate the
Light, 2001; King & Kraemer, 1984) continues to impact of ERP to the transformation of the roles
have a lot of appeal and is often used as a basis of management accountants, while Lodh and Gaff-
for consultantsÕ advice on ERP implementation ikin (2003) and Newell, Huang, Galliers, and Pan
(Deloitte Consulting, 1998; KPMG Consulting, (2003) follow the implementation of ERP through
1997; PA Consulting Group, 1999). social and technical networks. In contrast to the
A second strand of literature on ERP is con- first two strands of literature, this set of authors
cerned with performance and asks whether ERP identify noteworthy and sometimes huge effects
works? Cautious tales suggesting that ERP will of ERP systems both in the process of design
have positive financial and productivity effects only and in the process of use. They also start to explain
if installed correctly (e.g. Davenport, 1995, 1998, why the surveys are limited and thus unable to
2000; Koch, 1997) often dominate this strand of capture ERP systemsÕ effects. To understand the
literature. Perhaps therefore the conclusions vary. impact of ERP requires a heightened attention to
Some find that ERP may drive general financial ef- control as practice across the firm. This strand of
fects (Poston & Grabski, 2001), divisional perfor- literature suggests that clerical accounting work
mance (e.g. OÕLeary, 2002), or even capital is shifted out of the accounting function (Caglio,
market reactions (e.g. Hayes, Hunton, & Reck, 2003; Quattrone & Hopper, 2001; Scapens &
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 693

Jazayeri, 2003). All organizational members are becoming a ÔsystemÕ through both design and use
said to become Ôhybrid-accountantsÕ (Burns & (Bowker & Star, 1999).
Baldvinsdottir, 1999), and they take on control- Our paper illustrates actors debating through
work provided the availability, accuracy and practice: how they constitute an organization;
shareability of data (Koch, 1997). The accounting how they conduct their business; and how knowl-
function is possibly not necessary for information edge of the business contributes to managing the
production. firm. Through our analysis of objects, practices
Even if the third strand of literature in contrast and arguments the paper foregrounds a debate
to the first two has started assuming non-linear that in both corporations is real-time often also
explanations that refute simplistic notions of cause performed silently and unquestioned through var-
and effect, how ERP systems work and operates ious routinised control practices.
and the processes of engaging with ERP system
needs more detailed exploration.
If the ERP system does certain things, how is it Empirical setting, research approach and case
an actor; how does it play with other actors; and analysis
how does their interaction impact and challenge
our understanding of management control? The In order to explore the questions mentioned
aim of our research is to shed more light on an- above, we draw on Actor-Network Theory (La-
swers to such questions. tour, 1987, 1999a, 1999b) to structure our analysis
Our 11/2-year engagement with two firms that and we so emphasize how ERP-related practices
had been using ERP technology—in casu SAP sys- are constituted as networks of both technology
tems—for some years persuades us that such sys- and narrative.
tems neither perform as complete calculation To structure our understanding of these net-
machines nor provide integration per se. What works, we employed Star and Griesemer (1989)
they do instead is to draw continuing organiza- Ôboundary objectÕ to identify passage points. Fol-
tional attention to the question: what integration lowing our introduction to the empirical setting
is and how it is to be achieved? in the following section we explain on our case
Drawing on aspects of Actor-Network Theory, study approach.
this paper suggests that ERP technologies become
actors in very distinct ways given organizational The empirical setting
hopes for the future, the residues of the past and
certain technological traits. By themselves, SAP Prior to our field research we participated in six
ERP systems separate strongly between an weeks of configuration training at SAP Denmark,
Ôaccounting modeÕ and a Ôlogistics modeÕ that craft in order to enable ourselves to initiate and
the locus of management control in different ways. participate in technical discussions about the
But ERP technology does not do this by itself. The ERP systems in place. The training proved to be
power of technology is circumscribed by numerous very helpful in various ways both in terms of being
other elements—some of which are various infor- able to understand and make sense of technical
mation systems (sic!) and others, which are vision- conversations and also in terms of being able to
ary objects in the form of futuristic business elaborate on the organizational consequences of
models. systems technical configurations. For example
Therefore, the effects of the ERP system are re- several of our informants made reference to a high
lated not only to the information processing facil- level and a low level in the information structure,
ity of databases, but also to how this technology and they suggested that it is possible to some
can be made interesting far beyond the mechanics extent to Ôtake care on a low level of things that
of its standardizing calculations. We thus distin- were forgotten on a high levelÕ. Such comments
guish between ERP as a technology of database may sound strange, but they describe in very accu-
modules and ERP as a system, the latter only rate ways how actors circumvented technology to
694 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

make management practices work, as we will ex- global factories were located in Denmark, and
plain later. our research focused on one business unit that
After our training we gained access to two large operated two of these 12 factories. SpaceCorp
firms that for some years had been using SAP and was not only known as the first adopter of ERP
we stayed for about 1(1/2) year. In our research we systems in Denmark but also as one of the most
refer to the two firms by the pseudonyms of Time- experienced and successful users of ERP systems.
Corp where an agenda of the future bound debates The business unit in which we conducted our inter-
about ERP, and SpaceCorp where the autonomy views had been using ERP systems for more than
and dependence on IT, logistics and accounting 10 years and had by 1995 completed their system
were at stake. upgrade to the R/3 platform. In addition to the
TimeCorp was one of the worlds leading manu- two factories operating in Denmark, a number of
facturers of high-quality pumps and pumping sys- factories operated in Eastern Europe and most of
tems, with 50 production and sales companies in them were at the time of our research in the pro-
more than 30 countries and an annual turnover cess of implementing ERP systems. We did not in-
of approximately 1.5 billion Euros. TimeCorp of- clude them in our research, as we focused on users
fered multiple product lines spread across twelve sharing the same ERP system. Unlike TimeCorp
distinct product families. About half of the work- the factories at SpaceCorp were not legal units,
force of 9000 people was employed at the main but as our study will explain in further detail, from
company (TimeCorp Inc.) founded in Denmark an ERP system point of view, it was not even obvi-
in the 1940s. TimeCorp Inc. controlled more than ous that there were two factories.
15 manufacturing companies that legally were We approached these two firms because they
subsidiaries (profit centres). In addition to the 15 were widely recognized for their success with
manufacturing units TimeCorp Inc also had na- ERP systems. Also in the words of informants at
tional and international sales companies and a the two companies, success was clear. At
number of regional distribution centres placed TimeCorp:
throughout Europe. All TimeCorp business units
‘‘Before SAP, we simply were not in a posi-
used ERP systems from the system vendor SAP
tion to do a monthly P&L and balance with
AG. The distribution centres and the sales compa-
our old systems—really. There were two
nies, operated on a different ERP platform (SAP
problems at that time. We ‘‘lost’’ money—
R/3) than the one used by the headquarters and
and we didnÕt know why, and therefore we
the manufacturing companies (SAP R/2), and
had to write it off on high scrap rates. At
these two platforms were only integrated to a lim-
the time (before SAP) our bottom-line result
ited extent. Our research did not focus on the
would come up with positive/negative vari-
interface between these ERP systems as we there-
ances of up to 2 million Euros per month.
fore focused our data collection on interviews with
This has been eliminated now.’’ 2
respondents from the central accounting function
(at headquarters) and with respondents from man- And at SpaceCorp:
ufacturing units. They were all part of and use the
‘‘You see, since we started working with
same ERP-system.
SAP, we have been able to follow our inten-
SpaceCorp was one of the largest industrial
tions. . . SAP was implemented in Õ95—some
groups in Denmark. Worldwide SpaceCorp em-
of the things that (ED: that SAP automates)
ployed more than 20,000 employees with an an-
I had been working with—you know to sep-
nual turnover of approximately 2.1 billion Euros.
arate things thoroughly (i.e. total quality
SpaceCorp was organized in relation to its three
management). These concepts were already
business segments. It operated 53 factories in 21
on the agenda by Õ93, but they had no effect
countries and 95 sales subsidiaries on a worldwide
basis and approximately one third of the global
workforce was located in Denmark. 12 of its 53 2
No. 2—April 28, 1999.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 695

before SAP. However, at the time it was only mentary documentations materials. The result is
conceptual. We had many ideas. SAP made a story of the effects of ERP systems in two
them possible—our operations flow, for situations.
example. At that time our inventories were In the following we distinguish between ERP
always a function of our production and technologies and ERP systems. Whereas ERP
not in tune with our sales. Today, we are able technologies refer to the hard- and software of
to control our inventories. . . Today, invento- SAP, we use the term ERP systems as our label
ries are the result of our decision-making. for a broader notion of interrelated concerns and
We decide how we want them to look—and problems that are open-ended at least until
then they appear like that.’’ (SpaceCorp explained.
informant 3)
‘‘. . . the Ôbig successÕ in terms of accounting is Case approach
that we can split the final result by means of
the business areas. Last year when everything Our cases are written up as a Ôcross-case-com-
was settled, we had a residual posting of only parisonÕ to illustrate that there are always a
2 Euro. Otherwise we had everything under number of viable alternatives for management
control—and that is of course because oper- control—with and/or without ERP. Our Ôcross-
ations and sales were implemented the right case analysisÕ illustrates not only how actors act
way—we can do the profit and loss accounts with ERP but also how ERP systems in local set-
and two individual balances for our own tings act on the actors, sometimes forcing them
business and the other business that to go out of their way to accomplish certain forms
still reside in our system.’’ (SpaceCorp and functions of management control and vice
Informant 4) versa.
Our case approach draws on (French) Actor-
We conducted 34 interviews—18 at Space-
Network Theory that emphasises the symmetry
Corp and 16 at TimeCorp. All interviews were
of humans and non-human actors in producing
discrecorded and transcribed in full length in
reality (Latour, 1987, 1999a, 1999b), and on the
order for us to study boundary objects, classifi-
stream within the related field of (American) Sym-
cations, movements etc. to explore integration.
bolic Interactionism (Star, 1995) that advocates
The interviews were conducted in two phases.
attention to how standards and classifications de-
In the first phase we focused our conversations
velop and become mediated in practice through
on the configuration of the ERP systems and re-
Ôboundary objectsÕ (Bowker & Star, 1999; Star &
lated background information. In the second
Griesemer, 1989).
phase we focused our conversations on the
use of these systems for various management
control purposes. In both phases we were inter- Performing a network analysis
ested in what technology did and did not do; Like others who have drawn on aspects of actor
how it aligned with organizational procedures, network theory (e.g. Briers & Chua, 2001; Chua,
responsibilities and management principles; and 1995; Hansen & Mouritsen, 1999; Munro, 1993;
how it transported, replicated and consolidated Robson, 1992), we focused on the linkages be-
data in order to initiate a conversation on man- tween elements developed by the narratives of
agement, reporting and control processes. This interviewees, documents and technologies (includ-
gave us the opportunity to elaborate on the var- ing the ERP systems).
ious themes and to retrieve and discuss supple- As Actor-Network Theory posits, the world is
neither purely social nor purely technical but al-
ways a mix of both (Latour, 1999a, 1999b). In con-
No. 53—April 15, 1999. sequence it is important to study how human
Nos. 53 and 27—April 15, 1999. actors and ERP systems happen to each other
696 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

and influence each other in the course of exploring Characterising episodes of practice
integration. What exists and what has to be made In order to explain how Ôintegration workÕ
sense of, are the connections that can be traced by worked in two large firms that use SAP systems
exploring how actors are related to other actors of we draw on the notion of Ôboundary objectsÕ. To
various kinds and swap competencies (Latour, Star and Griesemer (1989), boundary objects are
1999a, 1999b). media through which different groups of actors
In order to perform a network-based analysis connect. In the introduction by Bowker and Star
there are only two guidelines to follow. The first (1999) define boundary objects as follows:
guideline suggests to study practitioners on the
‘‘Boundary objects are those objects that
premise that while they know what they do,
both inhabit several communities of practice
we—the researchers—must make an effort to
and satisfy the informational requirements of
understand them without imposing on them a
each of them. Boundary objects are thus
world that they do not draw in. The second
both plastic enough to adapt to local needs
guideline suggests as framed by Latour, 1991,
and constraints of the several parties employ-
p. 372:
ing them, yet robust enough to maintain a
‘‘Instead of opposing the individual level to common identity across sites. They are
the mass, or agency to structure, we simply weakly structured in common use and
follow how a given element becomes strate- become strongly structured in individual
gic through the number of connections it site-use’’. (p. 296)
commands, and how it looses its importance
In other words, boundary objects can be mate-
when loosing connections’’.
rial, such as for example the ERP system, or they
The network created in this way is a semiotic can be immaterial, such as for example a Ôcorpo-
one where elements are relevant to the extent that rate visionÕ. Star and Griesemer (1989) coined four
they make a claim about (corporate) integration. types of Ôboundary objectsÕ and referred to these
We explicated the relations between the things that as: Ôstandardized formsÕ, ÔrepositoriesÕ, Ôcoinciden-
technologies in place Ôtold usÕ about integration tal boundariesÕ and Ôideal typesÕ to which may be
and the things that interviewees told us through added Ôvisionary objectsÕ as Briers and Chua
their illustrations and stories about technologies (2001) do. With the boundary object Ôstandardized
at work. formsÕ Star and Griesemer refer to the communica-
The outcome is a form of theorizing that fo- tion across dispersed groups that takes place by
cuses on the particular episodes that illuminate means of common indexes and methods. In our
the convoluted process through which subjects cases Ôstandardized formsÕ refer to ÔpureÕ illustra-
and objects co-produce practice in terms of the tions of corporate structures and functions, for
ways business, management and technology example the presentation of supply chain visibility
relate to each other. We seek to avoid any sim- in TimeCorp, and for example the production of
plistic labelling of this process that could leave operations in SpaceCorp. With the ÔrepositoriesÕ
our examples both un(der)explained and un(der)- boundary object Star and Griesemer refer to the
explored in terms of how subjects and objects piles of objects in organizations that are indexed
network and we therefore do not attempt to and organized in standardized ways. In our cases
provide any accounts of the deep roots of these piles of objects are both global and local sys-
social behaviour. Instead we have sought to tems and also the debate about systems-related
understand how ERP was performed, and so practices that illustrate the indexation of objects
we firstly seek to illustrate how integration was in action. The third boundary object type is Ôcoin-
accomplished in relation to the ERP technology, cident boundariesÕ where objects with different
and secondly we illustrate how integration content share demarcations. In our cases Ôcoinci-
impacts the reality of management control dent boundariesÕ were found when the ERP sys-
systems. tems in place offered more integration to logistics
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 697

than to accounting issues. Here the ERP systems nologiesÕ through which local communities of
created boundaries that both organizations had practice develop durable cooperation practices
to either Ôlive byÕ or actively circumvent one way with other communities. What in particular makes
or the other. We found that Star and GriesemerÕs the Ôboundary objectÕ interesting is that it seeks to
description of Ôcoincident boundariesÕ too Ôa-his- overcome the use of a dualistic notion that often
toricÕ when compared to our cases. We therefore reduces organizational action and information sys-
invented a new one in order to account for the res- tems either to the categories of either ÔglobalÕ or
idues of the past, which accounted for the accumu- ÔlocalÕ. Boundary objects makes possible that inte-
lated activities around the ERP systems in gration can represent itself beyond what is made
question and through which certain kinds of expe- present by the standards of the global ERP config-
rience were explained (Mouritsen & Dechow, uration. Boundary objects various communities of
2001). We refer to Ôpath dependenceÕ as boundary practice to overcome the anomalies of standardi-
object as a reference to the way that system config- zation without imposing a naturalization of cate-
uration and practices of management control mat- gories from one community to the other:
ters through its implications in the present and for
the future. Finally, Star and Griesemer refer to ‘‘Boundary objects, however, are not just
Ôideal typesÕ as a symbolic means of communicat- about. . . ambiguity, they are not just tempo-
ing and cooperating about something. In our cases rary solutions to disagreements about anom-
such Ôideal typesÕ illustrate themselves through the alies. Rather, they are durable arrangements
means by which both organizations created an among communities of practice. (Bowker &
outlook to the future. In TimeCorp this outlook Star, 1999—p. 307)’’
symbolically was created through the 10 projects In terms of our study the notion of boundary
launched in order to become a ÔWorld Class Man- objects allow us to study integration in wider terms
ufacturer (TimeCorp) and in SpaceCorp a similar than only with reference either to its global (based
symbolic outlook towards the future was rein- on the ERP configuration in place) or local (based
forced through the conceptual practices of Total on supplementary technologies) properties.
Quality Management, Time Based Management Boundary objects allow us to overcome an artifi-
and Total Productivity Maintenance through cial reduction of practice into a duality comfort-
which Ôcontinuous improvementÕ was materialized. ably neglecting how narratives and technologies
As mentioned, Briers and Chua (2001) have sug- network.
gested Ôvisionary objectsÕ as a new category of As Bowker and Star (1999) would argue, only
boundary objects. We have added this boundary by considering the possibility of Ôboundary objectsÕ
object to our list given that both corporations crossing-over global and local arrangements,
throughout our exploration process made refer- integration becomes possible to explore as a
ence to their visions respectively to become a ÔsystemÕ—or network—of diverse narratives and
Ôworld class manufacturerÕ (TimeCorp) and to technologies that relate and yet have different
realize Ôcontinuous improvementÕ (SpaceCorp). focuses/objectives and therefore very well can be
While this list may not be exclusive it is sugges- at tension with each other. This tension in current
tive of objects one can look for in order to identify accounting and information systems research is yet
arrangements of technologies and narratives in the both underexplored and undertheorized:
two firms at a deeper level of detail than simply
looking for the ERP systems that in this study of ‘‘If social scientists do not understand peo-
course are massive boundary objects by them- pleÕs definition of a situation, they do not
selves. As Bowker and Star (1999) suggest, the understand it at all. . . This is a much more
important thing about the notion of boundary ob- profound cut on social construction than
jects is to consider them in relation to the stan- the mere notion that people construct their
dardization that global information systems own realities. It makes no comment on where
themselves represent. Boundary objects are Ôtech- the definition of a situation may come
698 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

from—human or non-human, structure or exploration making a quick backwards contextual-

process, group or individual. It powerfully ization. First we look at where each of the two cor-
draws attention to the fact that the material- porations want to go (A). But in order to
ity of anything (action, idea, definition, ham- understand these visions we quickly turn to the
mer, gun, or school grade) is drawn from the past by studying (B) what configuration choices
consequences of its situation.’’ (Bowker & both corporation made during the implementation
Star, 1999—p. 289) of their present ERP systems and by exploring (C)
the immediate consequences of these choices for
The Ôboundary objectÕ represents the idea that
the representation of data. On this basis we turn
an important element of durable cooperation
to studying the production management (D) and
among communities of practice often is based on
the debates that accompany the various modes
a perceived need for a mending of Ôglobal stan-
and forms of management (E), after which we fi-
dardsÕ and other Ôun-naturalized classificationÕ. In
nally turn to our exploration of the ways that both
relation to our exploration of ERP, the signifi-
corporations mobilize to proceed from the present
cance is that it allows us to consider how integra-
into the future (F). The table provides an outline
tion materializes through a series of working
of our presentation and summarises the elements
arrangements across the entire scale of global
of our case study analysis in a number of headlines
and local arrangements, which in turn allows us
contrasting the situations in the two firms.
to discuss consequences for management control
Through our illustration of each of these
on a more detailed level of understanding than
boundary objects we establish a ÔbricolageÕ of in-
currently illustrated in most studies on the impact
stances that serve to contextualize each other—
of ERP on accounting.
i.e. we use the present and the past to make sense
of the visions and the ways that outlook is created
Presenting a case based on episodes of practice in each of the two organizations. However, each
To organize our presentation of the two cases, episode has to be understood in its own right
we have organized the list of boundary objects as rather than as an illustration in principle of some
follows from Table 1, below. Each of the following larger processes that cut across episodes and
sections elaborate on how corporate integration is instantiations, but which is never present in its
explored and achieved through certain arrange- own form. While we would not claim this brico-
ments through each of these Ôboundary objectsÕ. lage to be two Ôwhole case studiesÕ that will allow
In the following we introduce the reader to this us to understand all things and all events in these

Table 1
List of boundary objects
TimeCorp Boundary object SpaceCorp
Vision World class manufacturing Visionary objects Continuous improvement A
Live by 15 plants instead of 15 business areas Path dependency 1 instead of 2 plants 2 B
business areas
instead of 2 companies
(+) Logistical integration Coincidental boundaries (+) Statistical analysis C
( ) Divisional reporting ( ) Management accounting
Produce Supply chain visibility financial Standard forms Supply chain focus D
accounting performance management Technical functionality
Operations accountability
Debate Global/local system mix vs. Repositories Being systematic vs. E
One ERP system fits all Producing explanations
Outlook 10 WCM projects Ideal types From TQM over TBM to F
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 699

two firms, we think that our analysis goes beyond cases also point out, not everyone will be equally
Huberman and Miles (1998) distinction between adept at this activity.
Ôwithin-case-analysisÕ and Ôcross-case-analysisÕ. Table 2 summarises the empirical findings, and
Our comparison illustrates that integration is it shows how numerous actors participate in devel-
less an end-result and more a Ôcumulative trajec- oping the integration work performed around the
tory of messÕ (Strauss, Fagerhaugh, Suczek, & ERP system in the two firms. The table separates
Wiener, 1985) where opportunities at one point analytically between Narrative (N) and Technol-
in time are barriers at a later point in time. The ogy (T) around each of the boundary objects ex-
struggle to efface mess is performed against a hope plored in the text. Table 2 shows how some
for a future. In both organizations the ERP system human actors are better than others at the game
performs a Ôtechno-narrativeÕ where an assemblage of associating narratives and technology witnessed
of technology and strategic propositions are cou- by the number of cells filled out across the table.
pled to recast the past, present and future. As the cases will show some actors create stron-
In contrast to those who argue that ÔIS develop- ger networks and are able better than others to
ment is concerned with transforming a Ôlash-upÕ of align a series of narratives and technology. ERP
heterogeneous, disorderly, and unreliable allies is there as an important actor, but there are also
into an automation, which once it resembles an supplements in the form of additional technolo-
organizational whole, can be considered a black- gies, hand carried data sets, attempts to circum-
boxÕ (e.g. McMaster, Vidgen, & Wastell, 1998: p. vent the agency expressed by ERP. With these
253), our cases illustrate that networked activities differences in mind we focus on ERP systems, the
never settle—even when they appear to form a fi- work of making integration work and the resulting
nite action net at certain points in time. IS devel- forms of management control. Our aim is not pri-
opment is not about creating a black box—or if marily to illustrate the strength of different actors,
it is, our cases illustrate why this pursuit can never but rather to discuss how actor-networks and the
be successful, since it is always possible to reopen idea of corporate integration develop given the dif-
and re-debate the black box. However, as our ferent strengths of these actors and how this is

Table 2
Actors and boundary objects in TimeCorp and SpaceCorp
700 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

woven into a long series of relations that talk inventories and provided increased clarity and
about the affairs of the firms, their strategies, their assurance about data than previously had been
accountabilities, products and processes, and their possible. Throughout our interviews this was men-
futures. tioned several times as an example of how ERP
Our cases not only show a similarity in the way served organizational integration, and in turn of
strong and weak roles are constituted across the how the possibility of achieving continuous
two organizations. They also show that the net- improvement and world-class manufacturing was
works are connected differently, and how their present.
spokespersons put themselves at centre in different TimeCorpÕs goal was to become a Ôworld class
ways by different means and this is how we believe manufacturerÕ. This was introduced by a then re-
to go beyond Huberman and Miles (1998) notion cent consulting report that concluded that Time-
of a Ôcross-case analysisÕ. Corp practices were far from being Ôworld classÕ.
Our objective, in the following section, present- TimeCorp did not have a definition of what it
ing our case, is to restore the narratives of conflict meant to be a world class manufacturer, and it
and compromise inherent to the configurations of was not clear from the consulting report which
Ôglobal technologyÕ, boundary objectsÕ and other was a benchmark study against certain Ôbest in
un-naturalized classifications—in the context of classÕ but anonymousÕ companies. The consulting
which TimeCorp and SpaceCorp had become report suggested launching as many as 10 different
known for their success with ERP. improvement projects.
The corporate ambition at SpaceCorp was Ôcon-
tinuous improvementÕ (CI) just as Ôworld class
ERP practice in TimeCorp and SpaceCorp manufacturingÕ (WCM) was the ambition at Time-
Corp. Any initiative or project introduced to us
Our starting point is the strategies of World would be justified as steps towards of the proposi-
class management and continuous improvement tion of CI, or WCM. In this situation, integration
(1st subsection), which are Ôvisionary objectsÕ in was not just about being more interconnected.
each of the two firms. Then we analyse how early Integration was also bent around these visions that
choices in setting up the ERP systems have influ- were strong enough to set an agenda and yet ade-
enced integration and management control work quately plastic with regards to its possible prac-
(2nd subsection) before we analyse socio-technical tices. Integration was important; WCM and CI
trade-offs (3rd subsection); we show how the were ubiquitous.
means of interaction (4th subsection) and how
ÔauthorityÕ and ÔcompetenceÕ are shaped (5th sub-
Path dependency: management of technology
section); and finally we conclude by examining
the means through which the strategic goals are
However, both firms suggested—although in
made real (6th subsection).
different ways—that their implementation of SAP
some five years previously might not have been
Visionary objects: world class manufacturing and
as effective as it could have been. It appeared that
continuous improvement
elements of system ÔstructureÕ had been compro-
mised for cross-functional ÔprocessesÕ since it was
Both firms had experienced favourable effects
surprisingly difficult to make management
from their ERP systems. Very importantly, they
accounting work:
were said to have helped the firms towards im-
proved financial accounting (complete balance
Configuration work
and profit & loss statements) and towards more
precision and timeliness, because of automatic ‘‘Well, we must admit, it does not really
and ongoing data reconciliation. They also were reflect SAP. . . we omitted Ôbusiness areasÕ
said to help (particularly SpaceCorp) control because we did not want to enter the extra
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 701

two business area digits on every transaction. information systems. Management was aware that
Probably we made the world a little more corporate reporting at all levels of the firm was
complicated for ourselves. Now, it is a little supported by numerous unofficial spreadsheets
difficult to get the system to reflect our inter- and word documents. A SpaceCorp informant
est in divisional performances.’’ (TimeCorp recalls:
Informant 5)
‘‘Management decided that they wanted to
‘‘Probably we were thinking far too much in
get rid of all those spreadsheet, word docu-
terms of our old systems—and we were
ments etc. that individuals used and to inte-
afraid that you would have to manually code
grate everything in one, central information
every transaction with an additional number
system. So we literally went around in the
for the plant location. Of course, this is abso-
organization asking each person how many
lutely wrong. We didnÕt know this back
Excel sheets and Word documents they used
then’’ (SpaceCorp informant 6)
on a daily basis. We did this exercise on all
Both firms converted their old accounting sys- levels—even our CEO had to let go of his
tems into the new ERP systems directly and then personal spreadsheets. In total we found 31
hoped that a piecemeal and systematic, albeit slow, local subsystems in the organization that
change process would be a non-risk operation. were only known and maintained by one per-
This was wrong because in SAP technology the son for his or her personal control pur-
accounting ledger played a central role in how it poses.’’ (SpaceCorp Informant 9)
was possible to mobilize accounting and logistics
For TimeCorp, the SAP journey started with a
based management reporting and control. 7 At
perceived need for a new purchasing system, while
TimeCorp the implementation started by looking
in SpaceCorp management wanted to integrate
at the part of business operations most in demand
(standardize) all the local systems into one data-
for new management control systems, which at the
base. The configuration work resulted in systems
time was the purchasing process. A TimeCorp
that Ôdo not really reflect SAPÕ because the firms
informant recalls:
Ôwere thinking far too much in terms of their old
‘‘IÕm not even sure any longer, but as I recall systemsÕ, as stated by informants who pointed to
it we began with purchasing, then we imple- the residues of the past which turned out to set
mented ÔdistributionÕ, then the factories and serious constraints to the future development of
then we finally converted our old financials e.g. management control.
system over a couple of years ago. . . Of In order to understand precisely why and how
course it meant that accounting has had to technology intervened here and directed the firms
adjust to all the choices that had been made in their production of management control proce-
over the years—you know one company and dures, it is necessary to return to the suggestion
15 plants. ItÕs probably not the structure we made in the firms that the systems Ôdo not resemble
would choose today, if one were to begin SAPÕ. This was a possible statement because in
all over.’’ (TimeCorp Informant 8) spite of claims to the contrary, ERP systems—in
casu SAP—did have a set of properties that al-
There were similar traces of serendipity in
lowed them to be actors in the production of man-
SpaceCorp and at the outset the primary reason
agement control.
to take in SAP was to replace most, if not all, other
In the following subsection we discuss how
certain configuration choices made at the time
No. 31—April 27, 1999.
No. 53—April 15, 1999.
‘‘Organizational units control functions and processes in
R/3 through master data and control data’’ (SAP AG, 1997b,
CA ConsultantÕs handbook—p. 2–22/23).
8 9
No. 32—April 26, 1999. No. 37—October 19, 1999.
702 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

when the systems were implemented at Time- of hardware. As shown in Table 3, firms could
Corp and SpaceCorp created certain premises choose between three scenarios running from a
for their use. In order to establish the basis for basic through an advanced system to a multiple
this discussion we take point of departure in systems system. The technological complexity
the relational database technology of an SAP increased through the scenarios, and management
R/3 system. control processes differed with each technology
set-up. The basic set-up offered maximum data
The SAP model of management control integration and the multiple systems set-up offered
In principle, SAP could operate any functional- maximum system and management control flexi-
ity, but it was not easy to evade its Ôtechno-logicÕ bility. Also, however, the multiple systems system
through which configuration work combined three required more IT power which required invest-
elements, ÔtechnologyÕ, Ôaccounting structureÕ and ments in technology that could be avoided when
Ôlogistical processesÕ established via five concentric selecting the basic set-up.
levels of configuration work. Obviously, under- In the basic set-up all master data and transac-
standing the techno-logic is of crucial importance tion data resided in the same database. Redundant
to a corporation implementing ERP. But equally data were few, which produced fast system perfor-
it is important to understand the configuration of mance, but which also developed certain system
the technology, when researching ERP because inflexibilities as a company then was required to
the IT configuration defines the global standards do everything in uniform ways. With increasing
for the production of textual and/or material Ôob- complexity, performance decreased if no addi-
jectsÕ—including Ôboundary objectsÕ. In the termi- tional investments in hardware were committed,
nology of Bowker and Star (1999), the IT and it was costly to keep performance in place.
configuration—once it is set—becomes the classifi- The flexibility gradually built into the higher-level
cation through which organizational actors meet systems came at the cost of reduced performance
ÔinfrastructureÕ. or increased cost.
In this configuration, the first level concerned Both TimeCorp and SpaceCorp had chosen a
the choice of technology, i.e. the scope and scale basic set-up, which required them to standardize

Table 3
Three technology scenarioÕs at the first of 5 ERP configuration levels
Control Basic Advanced Multiple systems
Configuration One system set up One system set up One enterprise with
for 1 company for X companies X company systems
Logistical integration Within location of Focussed at the level of Only to a limited extent across
business strategic business units business units and enterprise
across legal entities
Technologies and process Standardized to a Different in various Different in various business units
large degree business units
Markets Uniform by structure Varying by structure Varying by structure
Internal transactions No internal pricing Transfer pricing Market based customer/vendor
Operational control Standard reporting Independent operative and Full BU autonomy: system master data,
stratigic control open items management and controlling
General reporting Reports for external Autonomous SBUÕs in Independent
rendering relation to profitability
of accounts and cost reporting
New business Legal entities should not be Legal entities planable Legal entities planable ÔanytimeÕ
planned in short term in medium term
Source: Based on SAP AG, 1997b—CA—Consultants handbook.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 703

their process data uniformly. The managerial In principle firms were not limited to a choice
implication was twofold. Firstly, the choices in between accounting and logistics structure, be-
relation to building an accounting structure were cause everything was possible if only the system
limited. Secondly, it was almost impossible to set-up was sufficiently advanced. However, often
make any changes to the coded system structure the situation was different. Technology was
once transaction data had been entered. These costly, and many corporations therefore did not
implications had not been clear to either of the implement systems that were sufficiently ad-
companies at the time when they implemented vanced to host and represent both structures
the systems. They may not even have been aware on equal terms. This was also the case for both
that the relational database technology took a SpaceCorp and TimeCorp, who by the basic sys-
point of departure in the configuration of an tem set-up had given primacy to the logistics
accounting structure. based representation of the firm through non-
The technological property of the relational financial information over an accounting based
database system was that once the technological representation of the firm through financial
platform was chosen, four additional levels of information.
configuration work emerged (see Fig. 2). The first Yet, the accounting structure defined the start-
two levels concerned accounting structure and ing point of any configuration because it defined
the last two the logistics structure. In SAPÕs pre- the master data tables, which were truly basic
sentation materials these technological details and could only be changed at very high inconve-
were not readily available. Instead of present- nience and cost once the system was working.
ing the 5 levels as relational layers in a system Whereas accounting at levels 2–3 in Fig. 2 was lar-
infrastructure, SAP introduced the accounting gely stable, once configured with a Ôbasic technol-
and logistics structures and their interaction by ogy set-upÕ, logistics at level 4 and 5 largely
means of a graphical illustration of pyramids remained plastic and changeable once the firms
(see Fig. 1): used their systems. The tables underlying the logis-
tics structure were namely peripheral to the core of
‘‘We can represent Accounting and Logistics accounting tables:
in R/3 as pyramids with three different seg- Neither TimeCorp nor SpaceCorp focused in
ments. . . The segments of the pyramids, in any detail on such issues at the outset. When dis-
each case, vary in size. In Logistics, business cussing this diagram with informants in TimeCorp
processes are of decisive importance. In and SpaceCorp, they explained that they had—at
Accounting the focus is on information the time, which was in the early days of ERP—
acquisition for internal and external report- not realized the global implications of the system
ing requirements.’’ architecture:



Organisational Organisational
Units Units

Business Processes

Business Processes

Fig. 1. The accounting-logistics structure of an ERP-system architecture. Source: SAP AG, 1997a. BC The R/3 Process Model. Release
704 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

Level 1 Client or Clients

Level 2 Chart of Controlling Credit Control
Accounts Area Area

Company Business
Level 3
Code(s) Area(s)

Logistics HR
Level 4
Purchasing Sales
Organization(s) Plant(s)

Distribution Sales
Level 5 Channel Division


Purchasing Shipping Storage Sales

Group Point Location Office

Loading Sales Person

Point Group

Fig. 2. The 5-ring configuration architecture of an ERP system based on relational database technology. Source: Based on SAP
Academy Handout, 1998.

‘‘Basically one could say that ÔplantsÕ and vidual business units, because they all were defined
Ôbusiness areasÕ have been mixed up. What as ÔplantsÕ that in the world of SAPÕs relational
we did instead is to employ exclusive num- database technology were not expected to record
bering of departments. Department 23 only and report financial data, and therefore they could
exists once in the entire firm. . . But, we not be financial entities. SAP turned out to be an
should not neglect that we have imposed on actor that pointed out how certain kinds of man-
ourselves some restrictions because we imple- agement control were not possible. This was felt
mented the new system by converting the old both in TimeCorp and in SpaceCorp. The technol-
system 1:1 into SAP.’’ (TimeCorp ogy, which was configured partly for productivity
Informant 10) purposes, stroke back and hampered certain
‘‘. . . I think that someone told us that it dimensions of productivity.
would be so much easier if we only had one Presenting it here is simple compared to the
plant. However, when you have two physical complexities of technology in the process of imple-
locations—then you also need to have two menting it. In this process, vendors assumed that
units in the system. Therefore, we have to the graphical map of organizational work was to
do some counting from time to time. Really, be preferred to the work of technology in the firm.
it is much more complicated than it needs to In SAPÕs written materials, we found that they
be.’’ (SpaceCorp informant 11) black-boxed the technology in such ways that the
managerial control implications of different config-
TimeCorp and SpaceCorp mingled different ob-
uration models (Table 3) were unclear:
jects. As a consequence in TimeCorp it was not
really possible to produce individual profit and ‘‘Business people benefit from business
loss statements and balance sheets for its 15 indi- objects as a means of abstraction. They are
not at all interested in the details of a ‘‘pur-
No. 31—April 27, 1999. chase requisition’’ programming code, for
Nos. 53 and 27 in a first round interview on April 15, example. Much more important is the fact
1999. that business people can continue to use their
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 705

own language in order to efficiently commu- liveÕ and the logistics structure remained plastic
nicate their business needs. Therefore, busi- was not clear.
ness and IT people can both talk about Whereas the configuration (vertical) architec-
identical business objects from two com- ture stipulated a sequence of decisions to be made
pletely different points of view. Business in order to make the configuration workable, the
objects close the communication gap EPC diagram did not. The EPC diagram presented
between IT and business. . .’’ (SAP AG, configuration work as a liberal (horizontal) pro-
1996), R/3 System—SAP Business cess, where it would be possible to start from many
Objects—p. 3 and 4. angles as long as the management ÔfunctionsÕ (the
things that need to be done) were accorded pri-
These business objects—of which there were
macy. In contrast, the vertical-structure approach
only four—were hosted by a graphic called the
did not speak about the things that needed to be
EPC diagram, which visualized the corporation
done as did the horizontal EPC diagram. It spoke
in terms of its ÔprocessesÕ (see Fig. 3). The four Ôob-
about recording data. In brief, two visualizations
jectsÕ answered one question each. One was Ôfunc-
produced potentially incoherent representations
tionÕ (what should be done?), another ÔeventÕ (when
(although not irreconcilable), and each unveiled
should something be done?), a third Ôorganization
certain contingencies but each also hid certain is-
unitÕ (who should do something?) and the fourth
sues that could strike back as the purpose of the
was Ôentity typeÕ (what information is necessary to
system developed over time. TimeCorp and partic-
do something?).
ularly SpaceCorp had experienced how ERP sys-
ÔBusiness objectsÕ made questions of technology
tem representations had acted upon them:
peripheral. In the words of SAP AG, business ob-
jects accomplish Ôa high degree of abstraction with- ‘‘When it comes to sales organization, we
out detailed information, allowing project team should have had 2 units. We only have
members to communicate at a business level that one—but then we have overcome that obsta-
does not require special technical knowledge of the cle by splitting it up into a number of sales
system. (SAP AG, 1997a: BC The Process Model offices. . .yes—You can say that as we go
(Release 3.1G): p. 2–6). down the structure, it all becomes more oper-
In the EPC diagram there was little attention to ational and pragmatic—a mistake here is not
the location of the accounting and the logistics as fatal as it is in the top of the organization
structures in relation to each other (as illustrated system structure. Down there, it is not that
in Fig. 3). In the EPC diagram, the consequence serious, but up here the mistakes have awful
that accounting structure was solidified around consequences in the long run. . .’’ (SpaceCorp
the central data modules when the system Ôwent Informant 12)
The obstacles developed at a high level in the
configuration architecture were mended at a low-
Event er level of the system configuration. SpaceCorp
had come to understand the vertical configura-
Entity Org.unit tion structure over time, and in particular it
type had learnt how the logistics structure could be
modelled at lower levels to compensate for the
type missing sales organization, mentioned in the
quotation above.
Technological obstacles created in the ERP
Event Event system at the time of the implementation were

Fig. 3. The EPC diagram—an implementation technology.

Source: Keller (1999). Nos. 53 and 27—April 15, 1999.
706 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

ÔresolvableÕ in practice to the extent that the logis- The significance of the ERP configuration is not
tics structure at the lowest level could be expanded only the way in which it creates a global standard
e.g. by adding of sales offices, as mentioned above. in the two corporations. What is equally signifi-
Moving down the structural hierarchy however cant is how this standard—with its specific print
had a consequence as it also meant sliding from in technology and configuration—inevitably cre-
a financial to a non-financial plane of data. Both ates both Ôblind spotsÕ that delete certain organiza-
firms compensated for the structuring effects of tional representations and what Galison (1999)
the core systems (configuration levels 1–3) through refers to as Ôtrading zonesÕ through which organi-
continuously adding to the peripheral objects zational representations are shifted from
(configuration levels 4 and 5). Not only did this accounting based to logistics (non-financial) repre-
change how it became possible to mobilize data sentations of organization.
for various purposes, it also created a peculiar lo- In the first instance Ôblind spotsÕ are created
cal Ôtechno-logicÕ of somewhat Ôcoincidental when an organization settles between one of three
boundariesÕ created by a system structure whose technology models that makes the organization ac-
reach was not readily understood by all actors. cord more relevance to either the integration of
By adding layers low in their ERP systems both logistical flows or the separation of organizational
organizations accomplished a response to organi- units through accounting. Secondly, we learn how
zational concerns. However, typically, these Ôblind spotsÕ happen as a result of the path depen-
improvements did not benefit accounting because dency behind technology, as for example in Space-
the new layers were added in the logistics structure Corp where 2 physical plants in the ERP
and were connected in such a way that it would configuration were reduced to 1 informational unit
not be possible to extract and consolidate financial and where the informational representation of 1
data. SAP in this sense became an active partici- sales organization integrated 2 actual sales organi-
pant in its own construction. zations, which had nothing do with one another,
The strategy to convert the old accounting sys- except that they once had shared the same geo-
tems piecemeal into the SAP technology had kicked graphical address. Thirdly, we learn how Ôtrading
back and continued to do so since the change po- zonesÕ are created in both corporations because
tential was not present in the form it originally during the configuration process the possibility
had been anticipated. Our informants talked about of an Ôbusiness areaÕ focused accounting represen-
this by acknowledging that their systems did not tation of TimeCorps business is traded (unknow-
really resemble Ôa SAP model of managementÕ. The ingly) for a logistical representation that replaces
complexity was that SAP spoke two languages— a business focus with an operations focus.
one about the technology of relational databases Our cases begin to show why we should perhaps
and another about possible forms of management. not continue to lend unilateral support to the argu-
On this basis we now begin to understand the ment that ÔIS development is about the transfor-
visionary objects introduced in the firms. The ben- mation of unreliable allies into an automated
efits accruing to accounting were found (only) in black boxÕ (e.g. McMaster et al., 1998: p. 253).
consolidation of data, while it was difficult to Our illustration of Ôblind spotsÕ and Ôtrading zonesÕ
move further with financial management control. suggest that IS Development—even when it is
This was explained by the choices made in relation scoped on an enterprise wide level—is hardly able
to both the accounting and the logistics structure to produce a complete calculation machine. Its
of the SAP systems. The firms gradually moved model of the firm once it Ôgoes liveÕ continues to re-
attention more to the logistics, as the systems quire a substantial amount of Ôboundary workÕ in
logistics structures remained more plastic and flex- order to render operational those Ôunreliable alliesÕ
ible that the accounting structures that are only who because they now have been inscribed to a
flexible during the configuration of the systems, global classification standard suddenly experience
before transaction data are entered into the and struggle altogether with uneven access to
databases. information.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 707

Certain actors are better positioned than others plained that, with a little creativity, most data
in relation to the global system configuration. For could be separated at the level of the individual
example in TimeCorp and SpaceCorp, visionary operations planner—the MRP-controller:
objects are prominent with their focus on non-
‘‘. . . we are focusing a lot on the level of
financial punch lines that for example allowed
MRP-controllers—i.e. we can do our statis-
Ôcontinuous improvementÕ and Ôworld class man-
tics at this level—because at this level we
agementÕ because they support the relevance of
can ask the system to select all transactions
non-financial forms of information over other
starting with a ÔWÕ or with a ÔZÕ. On this basis
types of information.
we can then separate Ôwork in progressÕ and
In the following sections, we continue by illus-
so on. Really, we can do it all; itÕs just a little
trating how organizational integration results
more complicated. . . we can control a lot on
from the combination of actors, the boundary ob-
this level—we can separate functions, prod-
jects they mobilize, the way they approach these to
ucts—so the MRP controller is tremendously
make them part of representational practice.
important as a concept—much more than
anything else we control. You know prod-
Bridging coincidental system boundaries
ucts are assigned to the MRP controller, so
if we need an overview—normally this is
In both firms SAPÕs architectural configuration
the place to look at. . .’’ 15
had to be accepted. It may be that the systems in
principle are capable of many things, but not in When a colleague from the accounting depart-
TimeCorp and SpaceCorp. What they had to ac- ment in the same meeting suggested that the data
cept was that standards start acting globally in retrieved from this level were ‘‘. . . of course purely
practice once the architecture or infrastructure statistical or operational. . .’’, 16 the prompt re-
has been configured. Given the configuration ap- sponse from IT/Logistics was:
proach, where the organization had ended rather
‘‘. . . yes but I think itÕs managerial! If you
than started with the configuration of the account-
have a specific problem, you can run MRP
ing module, in TimeCorp there were significant
on an individual product you donÕt need to
restrictions as to the possible uses of the account-
do MRP on everything that is out there!’’ 17
ing-based representations:
The poverty of the accounting structure was
‘‘Well, I guess that we suffer a little from the
reinterpreted from a limitation to a condition to
fact that financials and controlling was
be managed, which can also be illustrated by the
implemented after operations had been
following exchange of views
finished.’’ 13
‘‘. . . We had to, because we initially had a ‘‘[Accounting-Informant]. . . yes—You can
different system for financials. . . But, we say that as we go down the structure, it all
should not neglect that we have imposed on becomes more operational and pragmatic—
ourselves some restrictions because we imple- a mistake is not as fatal as it is in the top
mented the new system by converting the old of the system structure. Down there, it is
system 1:1 into SAP.’’ 14 not that serious, but up here the mistakes
have awful consequences in the long run. . .
SpaceCorp found itself in a slightly different sit-
[IT/Logistics Informant] well, it means
uation because IT/Logistics had found ways of
that you have to live with. . . [Accounting
using certain information properties of the ERP
informant]. . . .Yyeaaaaaaaaaaaaahh, well it
configuration as a Ôboundary objectÕ to compen-
sate for the lack of accounting structure. It was ex-
No. 53—April 15, 1999.
13 16
No. 21—April 28, 1999. No. 27—April 15, 1999.
14 17
No. 31—April 28, 1999. No. 53—April 15, 1999.
708 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

means that you will have to resolve it down Standard forms: producing ERP-enabled
here, what you forgot to take into account management control
up there (in the top of the configuration
cross-section diagram) . . . [IT/Logistics Configuration of system technology matters
Informant] if you donÕt get it right at the when the boundaries of different functions are
beginning you have to think creatively about crossed. In both firms the functions of Ôoperations
ways of employing the functionality at the managementÕ and Ôfinancial accountingÕ had bene-
MRP-controller level in order to get your fited. For financial accounting the main achieve-
work done.’’ 18 ment was the elimination of manual data
reconciliation, and for operations management,
IT/Logistics had found ways to bring compe-
the main achievement was Ôupstream visibilityÕ in
tence into SAP, so that it could be used for specific
the supply chain, which allowed them to Ômake in-
managerial issues. But in bringing in this compe-
formed decisionsÕ about operations planning.
tence, also IT/Logistics contributed to what Bow-
In line with previous research findings (e.g.
ker and Star (1999) define as an Ôact of organized
Granlund & Malmi, 2002), neither TimeCorp nor
unlearningÕ—erasure through the ongoing destruc-
SpaceCorp had adopted any sophisticated means
tion of selective traces in the present. Allying with
of management accounting. As we have already
the ERP-configuration served an accounting pur-
started to address, this does not mean though that
pose—the separation of Plants A and B. But the
management control was not being pursued rigor-
alliance also implied a technology of politics—
ously. In the following two sub-sections we follow
marginalizing the accounting function silently—
how various informants throughout the two firms
and as such only surfacing because of our presence
articulated their local control measures and how
when interviewing both parties concurrently,
they sought to mobilize these measures as and
which for example resulted in the above exchange
through boundary objects. In order to learn about
of views.
these matters our conversations focused on the Ôre-
IT/Logistics had an advantage over accounting.
pairÕ work that had to be undertaken on the ERP
This was not because they knew the system better.
configurations to facilitate such controls.
Their advantage was that they could manoeuvre
on the lower levels of the ERP system. This dia-
logue illustrates that a global IT configuration also
can become an object that can be acted on.
‘‘ÔOperationsÕ has benefited the most from
Through the reinterpretation of the functionality
our ERP. . . Then I would argue that
of IT classifications, the Ôtrading zoneÕ imposed
ÔaccountingÕ has been second in line to bene-
by the ERP system can be utilised as a Ôboundary
fit from the ERP systems. . . well, what it
objectÕ. In doing so, certain actors not only run
means is that we are now able to produce a
away with technology as they reconnect global
reliable profit and loss statement—each
standards with the idea of integration. They also
month. Before SAP, we simply were not in
materialize a marginalization—an act of organized
a position to do this with our old systems—
unlearning showing that accounting is not neces-
really. There were two problems at the time.
sarily the accountantÕs job—in particular in this
We ‘‘lost moneyÕ—and we didnÕt know why,
case were the accountants do not (any longer)
and therefore we had to write it off on high
command the global standards of communication.
scrap rates. At the time (before SAP) our
Visibility may not be maximized and certainly is
bottom-line result would come up with posi-
not automated, but it is made ÔperformativeÕ and
tive/negative variances of up to 2 million
Euros per month. This has been eliminated
now, where inventory values are adjusted
electronically—our financial reporting is
Nos. 53 and 27—April 15, 1999. based on a calculated unit value—you know
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 709

adjusted for Ôbeginning-and-end-of-monthÕ for doing much of what weÕre doing is really
data on inventory. The reason at the time simple—it has been done before.’’ 23
was that our people sometimes entered the
Integration through automated production
wrong numbers when counting invento-
schedules was rare. SAP was used to run MRP
ries—you know instead of entering actual
on individual components, but approximately
value, they accidentally entered their
70% of suggested production schedules were over-
employee ID. Our old systems didnÕt moni-
ruled manually. 24 From the perspective of opera-
tor it. Our SAP system does.’’ 19
tions, integration was to provide factories with
Global standards disciplined the organization access to sales subsidiariesÕ sales order systems
to do certain things systematically. Based on this, and to global distribution centresÕ inventory sys-
we first turned to operations and then to account- tems. By having access to remote systems, opera-
ing to analyse the use of SAP. In particular we tions planners in the factories were able to match
wanted to know more about the type of planning production orders from one region with available
activities—and related forms of integration—that inventories from distribution centres in another re-
SAP was able to mobilize. Was it e.g. a MRPII gion. Integration concerned a virtual centraliza-
system that integrated production volumes with tion of logistics and a refocusing of operations
capacity constraints? This turned out to be a very management from a factory level to a supply chain
optimistic expectation, because the rate of compli- level. Instead of optimizing operations at the fac-
ance (indicated by the number of machine-sug- tory level before focusing on the supply chain le-
gested production schedules authorized without vel, TimeCorp moved directly to the supply
manual adjustments) was below 50% and Ômore chain level trusting its ability to run the production
like 30%Õ. 20 Operations planning appeared unre- schedules in factories on the basis of experience.
lated to machine-capacity, 21 and operation plan- An informant explained:
ners did not plan operation schedules and
‘‘Well, you have got to understand this the
monitor factory performance according to time
way it was! Culture is actually quite impor-
standards. 22. As an informant told us:
tant also in this domain. In France they do
‘‘I suppose that our operation planning is things very differently than we do. They
mostly based on experience. There isnÕt the donÕt care about operations. The French
thing we will do tomorrow that we havenÕt Madame who Ôdown thereÕ is supposed to
done before. However, if suddenly I got a work with logistics—basically, all she does
new employee out here, who wants to run is to authorize every sales order as a produc-
2000 products in 4 weeks on assembly line tion order. You see what happens then is
2, then I have no quantitative data with that my production schedules become
which I can prove that itÕs impossible. I sim- directly dependent on weekly sales. Factories
ply know, what I know. But I canÕt prove do not work that way—I cannot know what
what I know because we donÕt schedule our is being sold on a weekly basis, and itÕs not
production on the basis of assembly-line relevant because we Ômanufacture-to-stockÕ.
capacity. So what we do now is that I access their sys-
We run the MRP only on the number of tems and then I authorize their sales orders—
units to be produced and then we schedule on their behalf—so that when they reach my
the production on the basis of our experi- desk, they fit the production schedule that I
ences with the machinery. You see—the basis have prepared.’’ 25

No. 2—April 28, 1999.
20 23
No. 14—January 11, 2000. No. 33—April 27, 1999.
21 24
No. 33—April 27, 1999. No. 14—January 11, 2000.
22 25
No. 33—April 27, 1999. No. 33—April 27, 1999.
710 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

Organizationally the local operations planners Ômake-to-customer orderÕ directly from one
had been given access to authorize production or- of our sales subsidiaries, because then it is
ders at distance by accessing national ERP the sales subsidiary that must be invoiced
configurations. From a technological point of view and not the distribution centre. However,
these systems could have been part of a truly our R/2 cannot do this—because it canÕt
(informational and geographical) integrated global see our subsidiaries and their customers
information architecture. They were not, however, because this information is stored in R/3,
as TimeCorp had not selected the advanced multi- which R/2 doesnÕt interface with. Our distri-
ple systems set-up, discussed in the previous sec- bution centres run both R/2 and R/3 so
tion on technology. Nevertheless, Ôremote accessÕ thatÕs why its less problematic with Ômake-
allowed operation planners placed in Denmark to-stock-orders, which they simply transfer
to access national ERP systems with sales compa- by retypingÕ them manually from one system
nies and distribution centres across the world. to the other.’’ 27
Integration remained ambiguous though because
Thus there are slippages in the integration work
TimeCorp operated different technological solu-
performed by TimeCorpÕs ERP systems, which
tions in the factories and in the sales companies
have to be repaired by visual access, person-to-per-
and distribution centres. Originally, when SAP
son communication and manual retyping. This re-
had been implemented in the factories, SAP had
pair work illustrates how integration becomes an
been implemented as a mainframe based R/2 sys-
objective that is aspired even when in fact the glo-
tem. Later when the sales companies and distribu-
bal standards of the ERP system are neither global
tion centres were upgraded to ERP, certain of the
from an informational nor from a geographical
functionalities needed were only available in the
point of view.
new client–server based R/3 system. This combina-
Also, repair work was conducted in the attempt
tion of two different ERP-technologies created
to make TimeCorp factories comparable by a stan-
communication problems and thus managerial
dardized monthly report format. From a business
point of view, the 15 factories at TimeCorp were
‘‘The problem is very simple. We got all our 15 individual Ôbusiness areasÕ with each their bal-
customers (sales subsidiaries) in SAP R/3, ance sheet and profit/loss statements. In SAP these
and it is not possible to connect that infor- 15 entities however had (accidentally) been re-
mation with our R/2 system (that hosts the duced to an informational representation as 15
factories). We are of course working to find ÔplantsÕ (within the same business). By means of
a solution—but even SAP AG canÕt help us the global system configuration, it was not possible
with this incompatibility.’’ 26 to consolidate and read this representation as 15
‘‘It works like. . . not very well—if you ask profit centres. To SAP, TimeCorp was one consol-
me. Orders are placed in R/3 and automati- idated business. To TimeCorp the SAP system in
cally go to our distribution centres. If they this particular regard was a technology that si-
are not able to match the order, people in lenced the organization given its coarse
the distribution centres retype the sales order configuration.
from R/3 into R/2 as a Ômake-to-customerÕ To repair this, a new Ôfactory report-formatÕ
production order, which is then dropped into was developed, by which factories (in their capac-
the factory that produces the item. Now if it ity as individual business units) on a monthly basis
had been a normal Ômake-to-stock orderÕ our had to report various financial and non-financial
R/2 can update inventories in R/3 through information (following the format using 31 differ-
EDI and invoice the distribution centres. ent performance measures illustrated in Fig. 4).
This however is not possible if it is a This had to be transferred manually from factories

26 27
No. 33—April 27, 1999. No. 18—October 20, 1999.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 711

to accounting. The procedure was simple in princi-

ple, but it was heavy because the system architec- Report Division:______
Division:______ Month:__
ture resisted. A memo distributed to introduce Category/1000 Euro Month Year -to-date
this report format stated: Customer
6 non -financial measures
‘‘The factory report—which from now on Labor Hours
6 non -financial measures
will be known as the divisional report—must No. Employees

be standardized. The standardization serves Value of 2 financial measures

the following four purposes: 1/visibility is
increased; 2/comparability is enabled; 3/ data Productivity 2 non -financial measures
redundancy is reduced by eliminating local
reports and 4/ standardization supports our Scrap 1 statistical / 1 fin.measure

world class manufacturing mission’’ 28. Value of

1 financial measure
Unplanned WIP
To establish some form of benchmarking, cen-
tral accounting asked factories to literally hand Inventory 1 statistical / 1 fin. measure
over data from their plant systems. Even if SAP
served financial accounting work (consolidation Investments 4 financial measures

on a monthly basis), its recognition of manage-

Calculations 5 statistical measures
ment accounting appeared to be fragile. Most of
the management accounting data was routed via Fixed Costs 3 financial measures
supplementary reports generated by the add-on
information queries found in SAS technology.
TimeCorp provided a graphic on management Fig. 4. TimeCorp divisional report.
control systems (Fig. 5), which was used to com-
municate the sources of management control
information. Only the white boxes in the figure reports from our ERP system. . . ItÕs good for
refer to TimeCorpÕs SAP. The shaded boxes refer transaction recording, but in order to pro-
to a variety of sub-systems, and HR related infor- duce reports we use SAS’’ 30
mation remained in CICS—the old financial ‘‘You know what, the ERP system works
accounting system. SAP recorded all financial reasonably well on the lowest level of plants.
transactions, but managerial statistics were in ThatÕs where you can go in and get specific
large measure maintained in specialized databases data on the maintenance cost for example
outside SAP and were drawn in for budgeting and on each machine individually. You can
management accounting purposes. Ad hoc analy- extract, what was spent on this and that.
sis was typically conducted by means of However on the aggregated level of Time-
spreadsheets. Corp, then it gets complicated—its not con-
At the time of the research, the only report dis- figured very well to high-level reporting.
tributed across all TimeCorp business units was Basically, you can say that it is good for sim-
the sales report, 29 which was produced and main- ple bookkeeping. But itÕs not good for man-
tained in SAS outside SAP. As commented by two agement reporting, which in consequence
informants: means that we export data to SAS and other
places in order to work with them. . . How-
‘‘Well, if you are asking me, I have to be
ever, where previously we would run the
honest with you. We cannot extract any
management reporting cycle on a monthly

Document made available by TimeCorp.
29 30
No. 32—April 26, 1999. No. 26—April 27, 1999.
712 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

M.C.Systems Management Control Components


Salaries Course-admin

Purchase Orders
SAP Asset
Transaction Financial
Sales Recording Registry
Inventory Equipment Accounting

Sales Equipment Equipment Accounts Assets

(Statistics) Asset Asset
TimeCorp TimeCorp TimeCorp
Mngmnt Costs

SAS Management SAP directly
(Reports) Budgets
Accounting ABAP from SAP

Excel Calculation
(Calculations) Extracts to

Fig. 5. Management control (systems) at TimeCorp.

basis, we actually donÕt do this any longer. challenge because of global configuration clotted
Today, we drill things as needed. I guess this with Ôblind spotsÕ and Ôtrading zonesÕ. Getting the
is new compared to the past—but at the time right information was also a challenge because
we didnÕt have the data power we have organizational Ôdata disciplineÕ was meagre. This
today’’ 31 for example meant that certain practices—such
as for example cost accounting—could hardly be
Somehow data and information were a patch-
work. Getting the right data in the right places
was difficult because technology had created a ‘‘You know one of the things that are for-
Ôblind spotÕ that did not allow subsidiaries to re- gotten when integrating everything with
port profit and loss when they were defined only everyone is that it supposes a perfect
as ÔplantsÕ. The patchwork of subsystems including world. . . the problem right now are certainly
the new reporting format shows that numerous not with our systems, but with our organi-
boundary objects were enrolled that allowed the zation. Users donÕt know the systems suffi-
organization in each their way to draw on the glo- ciently well, and therefore they constantly
bal configuration for purposes that the standards create disasters in the system. Given that
of this configuration did not foresee. everything is so incredibly integrated we
However, Ôboundary objectsÕ were not always (in corporate accounting) are usually the
the immediate cure. Even if they here are repre- first to know that something is wrong. False
sented as a means of creating what Suchman and data entry is revealed in accounting terms
Trigg (1993) refer to as the Ôartful integrationÕ of immediately.’’ 32
local constraints, sometimes the constraints were ‘‘Let me give you the example of full cost cal-
too strong for the actors to work around. Getting culations. The controllers are doing them in
the right data in the right places was not only a a spreadsheet. You know there are some

31 32
No. 31—April 27, 1999. No. 2—January 10, 2000.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 713

common costs. We assign these costs to the started to debate this assumption, and this debate
various production departments on the basis can be reduced to a question: what constitutes a
of their realized work plans. We have to do manageable relational language?
this in a spreadsheet, because we can see that However, before elaborating on this, in the
the way operations are entering direct hours, next-following section we first turn to the produc-
costs etc.—they simply donÕt have the disci- tion of ERP-enabled management control at
pline. . . You know instead of sitting up SpaceCorp.
here—educating and hunting down all the
people on the operations floor, we chose an SpaceCorp
easy short cut. Now weÕre doing it by our- At SpaceCorp, the exploration of technology
selves—in a spreadsheet. Staff functions— took a different path than at TimeCorp. At Space-
for example—they are allocated to the oper- Corp attention was directed firstly at finding ways
ative functions by letÕs say 10% to each—and to make new practices functional through SAP
this is what we enter in the SAP system, and and secondly at finding alternative ways through
so it is actually SAP that does the calcula- SAP to serve existing practices. The directly expe-
tion. But of course, weÕre cheating a little rienced successes at SpaceCorp with SAP were
to make it happen.’’ 33 similar to those that were mentioned to us at
It was a challenge to have faith in a highly
integrated system. Here, however, it was not so ‘‘. . . the Ôbig successÕ in terms of accounting is
much because of the technology configuration as that we can split the final result by means of
it was because of local classification schemes in the business areas. Last year when everything
action: was settled, we had a residual posting of only
2 Euros. Otherwise we had everything under
‘‘We donÕt employ the functionality in our
control—and that is of course because oper-
SAP for the calculation of material prices. . .
ations and sales were implemented the right
because in this house we lack Ôdata-disci-
way—we can do the profit and loss accounts
plineÕ, if I may say so. SAP is doing what
and two individual balances for our own
they call a floating average, on the basis
business and the other business—that still
of recorded invoices and orders placed.
resides in our system.’’ 35
But our purchasing department—if they
donÕt have the price, when ordering, they The Ôother businessÕ was a residue of an organi-
just fill in a fictional 1 Euro as the price. zational arrangement, which was terminated ex-
Therefore I do not dare to use that func- cept from in the SAP system. As we were to
tionality. You see our data-discipline is learn, the SAP configuration challenged the firm
lagging. . .Yeah—thatÕs the trouble with in several ways that were similar to TimeCorp
integration. Everything is so damnÕ inte- although informants tended to make less hereof:
grated, so if you donÕt exercise data-disci-
‘‘We have had great success—in some
pline, then the consequences flow directly
areas—and the basis of our success is SAP. . .
to the bottom line.’’ 34
With SAP, we are able to see what happens
The paradox espoused here is that integration almost while it happens. . . Before SAP, we
could actually end up disintegrating the firm! This controlled on a case-by-case basis. But,
is a paradox under the assumption that integration SAP incorporates the philosophy of continu-
with ERP systems should be total and technologi- ous improvements—lead-time is a standard
cal (as opposed to being social). TimeCorp had measure. . . Actually, the only real problem
is when we want to trace deviations in
No. 21—April 28, 1999.
34 35
Nos. 26 and 31—April 28, 1999. No. 53—April 15, 1999.
714 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

operations. There weÕre stuck with the two the present. This in turn ensured that a person is
plant managers who always argue that the left unable to act on scrap rates the way he/she
problem is not located at his site—and then did in the past. From a technological point of view
they argue back and forth—but again it isnÕt reality was that scrap rates and inventory prob-
serious—we can deal with that.’’ 36 lems by all available means were cumbersome to
trace. From a social point of view inventory prob-
Through our conversations with other infor-
lems were not much more than the concern of
mants we came to understand how it could be that
accountants; a minor default less significant Ôfor
this configuration ÔmistakeÕ was not considered
the businessÕ than other focus points and ways of
serious. An informant explained as follows:
managing that the SAP system—as configured—
‘‘Well, we have to do a few tricks to report allowed for.
on our two factories. Here, at SpaceCorp, We set out to trace these new ways of managing
we are very focused on inventories—Ôwork- and we found that sales and purchasing at Space-
in-progressÕ in particular. In order to sepa- Corp were less in focus than operations. However,
rate the two factories we divide data through they were affected by changes in operations since
queries by MRP-controllers (who mainly both purchasing and sales were dependent on the
work for either one or the other of the two possibilities of producing and shipping products.
factories, and therefore they make it possible Purchasing, for example, no longer pulled actual
to separate inventory data on the MRP con- supplies, which had become operationsÕ responsi-
trollers family name). But then we always bility, and sales had to share data with corporate
have some data that we cannot trace to the marketing through SAP. Both of these changes
factories. Say we know that the cash value were results of changes in operations. Our aim
of our inventories is 4.5 million Euros, and was therefore next to trace the changes in opera-
each of the factories claim that they should tions with a view to understand what kind of
account for only 1.5 million Euros. When information had become available? These changes
we show that the factories hereby only were related to the continuous improvement
account for 3.0 of 4.5 million Euros both fac- strategy. 38
tories point at each other.’’ 37 Operations at SpaceCorp were reluctant to use
formal forecasts except for annual budgeting and
Here the ERP system configurations had a
standard costing purposes. 39 Two people from
Ôblind spotÕ. Even the Ôtrading zoneÕ enabled
sales (called external logistics) and the two facto-
through the MRP-controller search did not allow
ries (called internal logistics) developed the pro-
anyone to separate these data. Given the system
duction plan. External logistics communicated
configuration, scrap values had been rendered a
with all sales subsidiaries on a weekly basis about
Ôblind spotÕ. There was no way to circumvent it,
sales orders and inventories. A production plan
unless of course each factory had been asked to
was drafted, and internal logistics would break it
count and account for scrap in separate systems.
down on the two factories and send it to the
In lieu of such a Ôboundary objectÕ the alternative
MRP-controllers. These would then allocate it to
was to reduce it to a local problem, something that
2-shift manufacturing teams and machines and
only concerned accountants and not the ÔbusinessÕ
communicate back to internal logistics that then
per se. In the terminology of Bowker and Star
would present a set of priorities to external logis-
(1999) the Ôblind spotÕ was used to create a Ôclear-
tics that finally established the manufacturing
anceÕ—in the sense that the Ôblind spotÕ handily
schedule. Once priorities were settled production
could be used to erect a barrier in the past that
did not allow any information to leak through to
No. 7—October 18; no. 3—October 19; no. 37—October
No. 53—April 15, 1999. 19; no. 19—October 20, 1999.
37 39
No. 27—April 16, 1999. No. 44—April 16, 1999.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 715

orders were entered into SAP. Most of the plan- ‘‘Our production batches do not represent
ning work was drafted by hand and broken down economic lot sizes. Our production schedule
via Excel spreadsheets. 40 As explained by an follows pretty much the flow of sales orders.
informant: You know, we do also not record set-up
times. In fact, it is only very recently that
‘‘Okay, well when everything is settled with
set-up times were entered. Well, I did that
[the external logistics person], everything is
in this area. I wanted to know the set-up
entered into SAP. I have created a spread-
times, because we have been working on
sheet for this purpose, which I then forward
the basis of Ôbest-guessesÕ. I wanted to know,
to the MRP-controllers. The spreadsheet
because it is something that you can take
specifies the products and quantities that I
into account. Look here [on a SAP screen],
want per month, and then they will do the
there is a blue and a red bar. The red one
capacity planning. ItÕs based on Ôcommon
is for set-up times. I wanted to know specif-
senseÕ. They know the order of things they
ically, so I asked people to fill it out for every
prefer, the batch sizes they prefer etc. It
order they do. It turns out that they vary
makes no sense if I specify this, because it
from 15 minutes to 4 hours [half a shift].’’
might end up giving them way to many
machinery set-ups. . . When they are done Along the same lines, another informant (a
with that part of their job, they return the manager) explained that they were about to intro-
spreadsheet to me and then I enter their data duce the concept of Ôcapacity levellingÕ in opera-
as Ôlocal sales forecastsÕ into SAP. . . Now tions, so that instead of producing all to stock,
when all this is done, we ask the machine several manufacturing strategies could be applied
to compare sales order to production orders. at the same time. SAP did provide a series of pos-
Whenever the ERP system turns out a code sibilities waiting to improve practice.
40, it means that [the external logistics per-
‘‘You know back in the old days the only
son] has more sales orders than I have pro-
manufacturing strategy was Ôto-stockÕ.
duction orders. . . and in this way we can
Today, we are working intensively to get
see well in advance whether there is a con-
our MRP-controllers to understand, that
straint. If there is, [the external logistics per-
SAP provides a lot of other options with
son] has to reorganize with our sales
regard to our choice of manufacturing strat-
companies, or ask us to change our plans
egies. You know with SAP we can basically
again’’ 41
run our products according to each their
Not only did the process appear tedious and in own manufacturing strategy. In order to
contrast to the original motive for implementing realize this we are also visiting our sales
SAP it did not eliminate local spreadsheets. 42 It subsidiaries just to make sure that we know
also changed the notion of Ôintegration workÕ from how they order. . . Once we do, we can intro-
one where SAP was an actor in front to one where duce lean manufacturing. However, it
it was an actor in the back, which is not very dif- requires that the individual MRP-controller
ferent from the practice at TimeCorp. In parallel realizes this—that they can do it. . .’’ 43
we also found that SAP in some instances was
SAP might speak loudly, but the MRP control-
emerging as a repository of functionality. This
ler had to be able to listen and understand. A pro-
was illustrated by one of the MRP-controllers
cess had been initiated where SpaceCorp gradually
talking about monitoring production set-up times:
refocused operations planning to its upstream

Nos. 3 and 4—October 19, 1999.
No. 3—October 19, 1999.
42 43
No. 37—October 19, 1999. No. 53—April 15, 1999.
716 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

sales subsidiaries based increasingly on a new form fill them out on paper as well—but thatÕs just
of relationship between SAP and the MRP con- a safety measure.’’ 46
trollers. Various informants gave examples: ‘‘Well, out there [in the factory] people must
for a particular unit produce anywhere
‘‘SAP is a very good tool to support decen-
between 5000 and 9000 units. It depends a
tralization, because what we are pursuing is
little on their shifts etc. If they donÕt, itÕs
to move information out to each individual
money right out of their pockets. But they
so that they are able to act and make
have to report it themselves. . . However, itÕs
decisions.’’ 44
my job to go in and close the order formally
The organization changed when SAP was in the system after the factory workers have
implemented—and people where not happy about entered their data. So I do check up on
it. . . things, but otherwise itÕs completely up to
them.’’ 47
‘‘Today, it is our MRP controllers who
authorize manufacturing supplies. It used As in TimeCorp, integration through ERP was
to be our purchasing department, who would not total, and it was not purely technological,
then call the suppliers and then get back to which we came to realize through our conversa-
the MRP controller. It means that our pur- tions about the organizational availability of man-
chasing function today can focus on the con- agement control information. At SpaceCorp pc-
tractual stuff exclusively. . . Another thing is terminals were placed throughout the whole firm
that we run operations in teams. . . We basi- from which it was possible to request a presenta-
cally ask them to solve things themselves in tion on selected performance measures. Even if
relation to the week-plans we provide them the presentation was run in presentation software
with. . . We for example also ask them to fill (MS-PowerPoint), people typically believed it to
out the purchasing requisitions directly for be Ôon-lineÕ as in constantly up-to-minute in real
manufacturing related products such as oil, time made possible by SAP:
glue etc.’’ 45
‘‘. . . let me introduce you to our slide
The increased centrality of MRP controllers show. . . this is what our operations con-
was related to SAP because the system would stantly monitor. The numbers are from
speak up if workers had not closed the orders they SAP. . . The presentation is not, but all the
were executing. In actor network terms, the system numbers come automatically from SAP. In
would show its policing actor-hood whenever the old days we had someone writing it
workers started getting sloppy. In our conversa- down—paper, pencil and spreadsheet. Today
tions we encountered informants that described we donÕt have to, because SAP records it
the consequences of SAP as a matter of decentral- automatically in relation to each transaction.
ization and empowerment while others focused on . . . This is management information—and
its ability to act at a distance: itÕs accessible by everyone, which I believe
is relevant. Even out in the factories they
‘‘Now, we have changed practice in terms of
can walk over to a terminal and get an over-
work sheets. Factory workers used to fill
view. . . It enables them to understand and
them out on paper, and hand them over to
appreciate when we are scaling up and
the MRP-controller who was authorized to
down.’’ 48
enter the data in our systems. Today they
enter them directly into SAP, well they still

No. 3—October 19, 1999.
44 47
No. 37—October 19, 1999. No. 4—October 19, 1999.
45 48
No. 3—October 19, 1999. No. 53—April 15, 1999.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 717

The presentation assembled what certain indi- to make integration happen as a mindset through
viduals had always been monitoring, but it had which to drive various other organizational
never been put together in one picture. 49 The initiatives.
new thing was not as much technology as it was
the idea of projecting SAP as the technology that
The relation between accounting, management
produces integration. What happened in actor-net-
control and SAP
work terms was that competences were swopped
Even if the various stories about SAPÕs conse-
between the SAP system and the actors extracting
quences sometimes had very little to do with tra-
data from the system and feeding them into a slide-
ditional notions of (cost) accounting, they all
show. Here the actors endowed the ERP system
have a very clear relation to the domain of man-
with representational capability it did not have in
agement control. The various stories illustrate an
return for organizational belief in the SAP system
emerging division of control and accounting.
as omnipotent (or at least hard to question) black-
Clearly, as reported, the ERP system in both cor-
box (given that everything was presented as inte-
porations had had positive effects mainly in rela-
grated and systematic).
tion to financial accounting in terms of accuracy
‘‘This is our on-line management report. and timeliness in consolidating the balance sheets
[The factory manager] maintains data on and profit and loss statements for business units
our lead-times. [The HR person] supplies both in SpaceCorp and TimeCorp. These
labour statistics. Our [quality manager] holds accounting effects had little to do with planning
data on scrap. Another one, who used to be and control, which—as the stories illustrate—
in our quality function, is responsible for were developed elsewhere, typically around pro-
data on machinery maintenance and service. duction issues. Accounting appears to be reduced
One of the controllers maintains data on our a ritual:
standard variable cost and variances. Our
‘‘WeÕre calculating standard costs here—and
safety manager enters statistical data on acci-
I think we will continue this, given that we
dents etc and I supply the data on our ability
see no reason to change this principle. . .
to deliver to sales subsidiaries on time. . . We
The production program is always based on
have always had these statistics. However,
quarterly forecasts. The manager of external
we never put them together before. This is
logistics and the two plant managers lay out
what we are doing now, and it is in that sense
these forecasts. They establish a best guess
mere coincidence that the people I mentioned
about the next quarter and from there, you
assume responsibility for this. ItÕs just
only have to look at the number of employ-
because their information was already in
ees available to calculate the rates. Certainly,
it. . . We each update our pictures on a
rates deviate as much as our forecasts. . .
weekly basis—approximately.’’ 50
weÕre a very traditional mass-producing
The Powerpoint slideshow was dramatic more firm. . . Our purpose is only to compare our
in its symbolic than in its material form. By re-rep- total production costs quarter by quarter.
resenting (manual) data extracts the slideshow was . . .From our work with the reports we have a
able to symbolize integration achievements. It was general knowledge. . . [But] often I donÕt need
a convenient technology bringing together people to look into the system to know if some-
from different functions. Rather than dismissing thingÕs wrong. For example in the canteen
the slideshow as Ôsmoke and mirrorsÕ the point is area you see the R&D people looking dis-
that there is no smoke without a burning interest tressed, then I already know that they will
come up with an explanation about current
projects. Management accounting isnÕt that
No. 3—October 19, 1999. difficult, when it comes to it. Usually you
No. 3—October 19, 1999. can tell from peopleÕs appearance—the way
718 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

they appear at lunch in the Cafeteria— was given reason with ÔprocedureÕ and given ÔhopeÕ
whether they are in trouble or not. Small of being on the way to integration. Some of these
signs are left everywhere—for that matter constellations focused the firms on the future,
also in the system. We are about 700 people whereas others presented TimeCorp and Space-
out here, but I can still overview a lot with- Corp as dragging residues and path dependencies
out going to the desk and drilling the from their pasts. They suggest that there in inte-
data.’’ 51 grated information system environments are per-
manent tensions on a range of scales that cannot
Obviously, this technique does not require
simply be inscribed by the juxtapositions of Ôglo-
boundary work that draws on objects—such as
balÕ and ÔlocalÕ. Other factors such as Ôpath depen-
the global system infrastructure. What our respon-
dencyÕ and culture of habits also play an important
dent willingly introduced was a local sorting mech-
role when it comes to explaining how practice with
anism that does not need sharing with others in
ERP is made workable, and how the representa-
the organization, but which therefore also indi-
tions of information across localities become
cates that accounting in relation to this organiza-
tion positions itself as a discretionary centre
Consecutive acts of thinking about respectively
(Latour, 1987, 1999a, 1999b) monitoring from
Ôthe businessÕ, Ôthe idea of integrationÕ and Ôthe
the periphery (of a canteen area) as opposed to:
ERP technologyÕ and the ways relationships be-
from the centre of operations. As we were about
tween these are constructed and/or deconstructed
to learn being a centre of discretion allowed the
are interesting as acts through which different par-
accounting function to play a role in the ongoing
ties in both organizations mobilized repositories.
integration, even though accounting in important
In this process, the ERP configuration was not
respects had been marginalized to the periphery.
simply given as the global standard for the repre-
The comment about monitoring faces in the can-
sentation of information across localities. Of
teen area is an example of the ways that account-
course, to some of the respondents SAP defined
ing even from the periphery of action was able to
the strategy for integration. But to others SAP
influence and effect integration—even without al-
was only a variable in making integration a strate-
ways having to use the information systems and/
gic priority.
or related boundary objects. To the experienced
These observations contribute to explaining Ôthe
accountant, a particular ÔairÕ among the R&D peo-
permanent tension between the formal and the
ple in the canteen area was usable as a means to
empirical, the local and the situatedÕ that Bowker
monitor the progress on R&D projects. Following
and Star (1996, p. 291) find to be both underex-
his/her experience from our respondent was able
plored and undertheorized as they question and
also to assess the coming of budget overrides even
comment: How can something be simultaneously
before the information systems would set of an
concrete and abstract? The same and yet different?
alarm. As the respondent mentions in the above
People are not (yet, we hope) used to thinking in this
quotation, small signs are left everywhere, and
fashion in science or technology. As information sys-
they too are being used for the production of
tems grow in scale and scope, however, the need for
such complex analyses grows as well.
We learned from our respondents that practitio-
Repositories: debating corporate integration
ners are used to thinking in fashions that in re-
search perhaps yet are considered avantgarde.
As the previous four sections have now illus-
Many of the respondents focusing on ways to pro-
trated Ôintegration workÕ was arranged and main-
duce ÔpracticeÕ through ERP, also were quite skilled
tained through boundary objects through which
in juxtaposing ERP with practice in order to de-
global standards of the configured ÔtechnologyÕ
bate respectively what was needed versus what
was possible. By keeping everything present as
No. 27—April 16, 1999. variables these respondents gave meaning to the
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 719

Ôconcrete/abstractÕ; Ôsame yet differentÕ dichotom- matter. Integration matters and the advan-
izations raised by Bowker and Star (1999) by tage is that it becomes possible to talk, with-
mobilizing another layer of reflection about the out all those interfaces. Basically the factory
broader ERP context in which they were situated. becomes able to see whether the sales com-
By reflecting on integration, business and technol- pany has an inventory or not.’’ 52
ogy in relation to each other the respondents nur-
IT, in other words, wanted to debate and ques-
tured the ability (at least hypothetically) of being
tion current Ôintegration workÕ arguing that the
able to reopening the black box that they and oth-
firm was not yet utilizing SAPÕs technological core
ers in order to keep reality workable at other times
to facilitate supply chain management. Integration
closed by means of various boundary objects.
work was dependent on human interfaces that
should be eliminated by means of technology.
TimeCorp The IT functionÕs perspective was drawn to us as
In TimeCorp the debate circled around the ques- illustrated in Fig. 6.
tion, how it were to make technology part of inte- The diagram is interesting not only because it
gration work in the future. IT people were busy explicitly promised that a future ERP system
creating a future SAP as a repository of opportuni- would integrate and automate the entire organiza-
ties to be explored and seized, and operations was tion in ways that were not yet possible at the time
equally busy creating alternative repositories of of our interviews. The diagram is also interesting
accountability and control. In SpaceCorp the de- because it drew up a past that contextualized the
bate was subtler and focused more on inter-profes- present as a step towards the future. The figure im-
sional relations and conflicts. The proficiency with plied that a future with only one global informa-
which the IT/logistics function had been able to tion system would be superior to the then current
make the current ERP-architecture resourceful state of integration. Through the figure the future
led the accounting into a corner where it started was made a resourceful repository that rendered
defending its domain. In both situations SAP invisible the risks, the complexities and the costs
brought resources to the idea of integration. that would follow with one global system.
TimeCorp was working with a bricolage of sys- Simultaneously, operations were busy building
tems that integrated the firm in some ways as much a very different repository where SAP was not
as they compelled it to do manual work in other the solution to integration work on a local produc-
ways. In the view of the IT function, TimeCorp tion level. From the perspective of operations it
was Ôstuck-in-the-middleÕ of its work with plural only worked on global supply chain level, as ex-
systems. TimeCorp appeared far from an ideal of plained by the following informant:
IT-based containment and visibility:
‘‘I would really ÔenjoyÕ if it would be possible
‘‘In the long run, when R/3 has been imple- to run a factory like a black box, if you know
mented, then you will have complete visibil- what I mean—only with my MRP as a calcu-
ity throughout the chain. The sales person lating black box. However since it is not pos-
will be able to follow the order back to oper- sible, instead I would like to free myself from
ations and have it confirmed directly. . . You continuous MRP batch runs. . . I would like
see our intention is to have full visibility— to be less dependent on SAP within opera-
once operations get onto R/3 in 2001-2-3? tions although I wouldnÕt mind even more
ItÕs basically about making sure that we integration across the supply chain to be
reconnect the chain—with full visibility—so based on SAP—between factory and distri-
that sales can look through the distribution bution centres. . . In operations I prefer
centres and into the various factories. . . visual planning by means of kan-ban
You see this is about integration—to have
total integration. Whether the system is
SAP or something else, that really doesnÕt 52
No. 18—April 28, 1999.
720 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

Time: System:
ERP Make to order Make to stock

2002 >

Production DC Sales

R/2 R/3

1993 >

manual Distribution Sales

Production transfer Centers

R/2 R/2 Other

1991 >

Production Logistics Sales

Fig. 6. Building the repository of opportunity.

systems. . . you see I donÕt want SAP to tell I think that this product configurator could
my factory workers how many units they become a tool that will qualify the skills of
must produce. I want them to see this our sales people. What of course happens is
themselves.’’ 53 that some start to question who should be
responsible for this task? So right now the
Although technology already was able to per-
project is moving a little back and forth until
form kan-ban control and was easy to use, opera-
we know who is in charge. Some of us believe
tions did not find that the kan-ban screen on a PC-
that this tool equals more flexibility. But, I
terminal had the same effect as physical cards in
donÕt think that this is about flexibility. The
three rows of colours. With this manual technol-
product configurator can enable a common
ogy, factory workers could assume more responsi-
language between sales, R&D and opera-
bility over their work compared with a situation
tions. Basically, it will allow us to define
where data were illustrated on a PC screen, it
the level at which we start customizing our
was suggested. Also, operations introduced a tech-
products to specific orders. And also this
nology—a Ôproduct configuratorÕ—that could be a
thing will give us the opportunity to define
repository of control over sales entities.
some rules of calculation.’’ 54
‘‘You see, my interest is to inform the sales
Whereas the IT-function focused on building a
subsidiaries about the component types I
future ERP implementation as a Ôrepository of
can produce into a variety of products.
opportunityÕ, operations promoted alternative
You see if they require something that forces
technologies as Ôrepositories of accountability and
me to build products with components I
controlÕ. Here, added perspectives were produced
donÕt have on the shelves, then it hurts—it
to illustrate what integration could be about.
really hurts. That is why we here at Time-
Corp have defined a project called the Prod-

53 54
No. 14—January 11, 2000. No. 14—January 11, 2000.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 721

SpaceCorp each of them with some help in formulating

In SpaceCorp the IT/logistics function was able their viewpoints. They will now discuss how
to drive accounting into a corner where it instead things look, and without any doubt, IÕm
of participating in the work of building reposito- quite sure that we decide to look on our
ries started to defend its eroding domain: way of managing projects!’’ 57
‘‘I will make sure that everything that relates The IT/logistics function appreciated the
to accounting also stays with accounting. dynamics of the situation. The proficiency with
Things become invisible, when other parties which they in the past had been able to make the
interfere—or you can say that the expertise current SAP architecture resourceful for example
is removed from the function that should by means of management concepts, first TQM
know and a systems department then appro- and second TBM made them realize that:
priates this knowledge. YouÕre basically
‘‘we constantly need to get better in promot-
moving knowledge about the company out
ing our ideas. We are not supposed to do
of the functions and into the systems depart-
anything of course—well of course we
ment. But a systems department should not
are—but we want the people out there to
be in charge of the accounts. However,
take ownership of the things we do. Our pri-
this is what often happens. IÕve seen it
mary task is to make them realize the ideas
before.’’ 55
themselves. . . I had to teach some of our peo-
At SpaceCorp, the accounting function was ple in operations. IÕm not sure that they
aware of its marginalization. Integration work know what TBM abbreviates, but they are
was about influence and dominance, and because very motivated. Just hearing them speak-
accounting was limited by technology it began ing—they are promoting the ideas them-
to safeguard its domain from the rest of the selves. Their arguments are TBM. It shows
firm: that they have adopted these issues—without
knowing that it is TBM. In that sense, I think
‘‘We have created this menu for divisional
that someone has been doing very good
management. Here we have gathered the
fieldwork.’’ 58
data, we believe are relevant. Currently, the
number of reports is limited. We are working IT/Logistics had already launched the third
on it. However, the problem is that manage- management concept, total productivity mainte-
ment information has become secondary to nance (TPM), and this conceptual repository
the operational. was linked to functionalities in SAP. IT/Logistics
In theory our relevance is whenever the var- attempted to bring management concepts and
ious departments are coming up with expla- technology in line. As an informant commented:
nations that donÕt fit. Then management
‘‘. . . you could probably call us the sys-
will employ us as internal auditors—and of
tems department—not systems because of
course, you have to have a neutral party to
IT. ItÕs more like systematic—weÕre con-
do this job. . ..’’ 56
cerned with the business processes. Given
Yet, the repository to nurture for accounting that SAP is our backbone it is only natural
was mobilized only occasionally when there was that weÕre concerned with how these things
a need for an explanation: relate.’’ 59
‘‘. . . Yesterday, I had all the managers in
here—well most of them, and IÕve provided
No. 27—April 16, 1999.
55 58
No. 27—April 16, 1999. No. 7—October 18, 2000.
56 59
No. 27—April 16, 1999. No. 7—October 18, 2000.
722 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

In this strategy, there was room more for strate- ‘‘I donÕt think that we can point out one pro-
gic visions such as TPM than for issues such as ject and say this is the most important. They
variance reporting. But again this had a lot to do each have their focus. You really canÕt say
both with the original configuration of the system that one is more or less important than the
and the subsequent additions to it: other. . . I would say, one area is concerned
with operations—issues like inventory reduc-
‘‘We have had great success in some areas—
tion, time-to-market, continuous flow manu-
and the basis of our success is SAP. . . With
facturing, kan-ban, reduction of lead times,
SAP, we are able to see what happens—
visibility, supply chain management. . . Then
almost while it happens. . . Before we had
there is something about total productivity
SAP, we controlled on a case-by-case basis.
maintenance—which is basically that if we
But, SAP incorporates the philosophy of
plan our maintenance we can improve our
continuous improvements—lead-time is a
capacity utilization—in some areas in fact
standard measure. . . Actually, the only real
by 15–20%. . . And then a third area is con-
problem is when we want to trace deviations
cerned with new performance management.
in operations.’’
This is about aligning factories by using the
same measures—making visible that these
Ideal types: a sense of direction through units are comparable. . . So I donÕt think that
management concepts you can point out one project and say this is
more important. They idea is to become
In neither TimeCorp nor SpaceCorp was world-class by realizing all of them—so
anyone effectively able to define for us what it really we canÕt do without any of them!’’
meant to be a Ôworld class manufacturerÕ or what (TimeCorp informant 62)
it took to Ôimprove continuouslyÕ. They functioned
All projects were deemed important because
as visionary objects with a global flair and with
they helped to do something in relation to world-
room for being given local content and meaning:
class manufacturing. Each provided a hint about
‘‘I guess we skew all the concepts a little bit. how the future could work in a different, more
Our employees have absolutely no problems coherent and more integrated way. Together the
with this’’ (SpaceCorp Informant 60). projects created not only a set of solutions—they
‘‘You know the other day we got this list also poured content into the very idea of World
defined for us—fancy words; things we need Class Manufacturing and thus defined its charac-
to do. New operations planning; supply ter as Ôideal typeÕ.
chains; total productivity maintenance; flexi-
bility—more flexibility in operations and so SpaceCorp
on. But you know what? When we reviewed At SpaceCorp a similar attempt to create the
this list, it actually occurred to us that we Ôideal typeÕ future was accomplished in a different
are doing most of these things already!’’ way. Here, an exploration of different manage-
(TimeCorp Informant 61) ment concepts and models was performed in se-
quence—first total quality management (TQM);
TimeCorp then time based management (TBM), which soon
TimeCorp had recently initiated ten projects in was to be succeeded by total productivity mainte-
order to make world class manufacturing clear in nance (TPM). SpaceCorp was busy visualizing
the firm. how Ôideal typesÕ defined the future—as a time-line;
as a house of activities and as a set of ideas of per-
formance as drawn up in Fig. 7.
No. 52—October 23, 2000.
61 62
No. 18—January 10, 2000. No. 18—January 10, 2000.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 723

These graphic visualizations (Fig. 7) showed the The Ôboundary objectsÕ introduced, represent
future and they communicated a road towards it. working arrangements that contribute to resolving
They were means of mobilizing the future and a anomalies in the operating ERP configurations. As
means for control, and they illustrated where ÔweÕ Bowker and Star (1999) reminds us these objects
had been and where ÔweÕ were going by firstly only become material in the context of their ac-
introducing the approximate content of the latest tion. It is in their use that we learn how and to
concept (middle image); secondly explaining what what extent they supplement the ERP configura-
the organization expected from the individual tions in place. As Bowker and Star (1999) also re-
(lower right image); and thirdly presenting a grand mind even when these boundary objects illustrate
picture of where SpaceCorp was coming from and durability, this does not predetermine whether
going to (upper left image). They placed the firm in they will ever become naturalized as part of the
a temporal grid. overall infrastructure (ERP + supplements). Our
cases also illustrate that the naturalization of infra-
Summing up: the politics and materiality of structure depends not only on ÔworkabilityÕ of a
actor-networks technology (whether global or based on different
boundary objects) but as much on: A/the extent
TimeCorp and SpaceCorp help to illustrate how to which it can mobilize Ôpolitical appealÕ—i.e.
integration can be constructed. On the one hand convince to be in line with corporate visions—
we saw how ÔERP in actionÕ engages corporations and B/the power with which it supports the
with a set of classifications that standardize across unlearning of alternative/competing modes of
the organization how informational representa- representation.
tions are possible. On the other, however, we also In SpaceCorp, IT/logistics creatively found
saw how ÔERP in actionÕ challenges people in these ways to utilize the ERP system in unconventional
organizations with informational Ôblind spotsÕ and ways for conventional purposes. It translated stan-
Ôtrading zonesÕ forcing them to create supplements dard cost deviations across the two factories
that can help to make ERP performative by build- (which the ERP system could not monitor very
ing representations that Ôdo the workÕ. well, because two physical factories had been

Time to Delivery Team Team



6. Tools - quality assurance
1. Continuous Improvement

5. Tools - repair & replacement

3. Maintenance Scheduling

8. Safety & Environment
7. TPM in administrations

9000 9000 9000 9000

2. Quality Circles

1970 1992 1996 1998 1999

4. Training

How do we measure the TPM outcome ?

It will be evaluated by those measures we already
employ on a daily basis. Most of these measures
Team work are known already.

Organization Involvement
Increasing Productivity

Top Management Commitment Decreasing costs of scrapped materials

Improving time to delivery performance

Decreasing Costs

Decreasing number of accidents

Increasing Overall Equipment Efficiency

Fig. 7. Ideal types in SpaceCorp.

724 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

replicated into one plant in the ERP system) into a cases also illustrate, the point about ERP is not
quality problem, which could be monitored by accounting but how to make sure that common-
scrap rates that could be traced by MRP-control- sense controls are put to work while at the same
lers (operation planners). The scrap rates told time making sure not to become an appendix to
about the production efficiency in some ways par- other parties. In this particular sense the Time-
allel to standard costs, and this made an alterna- Corp case is very similar to the case of SpaceCorp,
tive to Ôcost controlÕ possible via unconventional although Ôpractical politicsÕ is here much more
translations. Not only was this solution com- about becoming an autonomous Ôcentre of discre-
mon-sensical—given the ERP configuration in tionÕ than becoming a Ôcentre of calculationÕ con-
place. It also appealed to the policy of continuous trolling others.
improvement, and last but not least, it served to At TimeCorp, no one resisted integration as an
move attention away from traditional accounting idea, but informants had only vague ideas about
based controls towards ÔnewÕ non-financial mea- what to do with it and attempted to use the ERP
sures that in contrast to the traditional accounting system as a tool to guarantee their own autonomy.
based reports even had a graphical representation. Operations for example wanted to introduce a
An illustrative paradox is here was that the Ôproduct-configuratorÕ in order to give them access
accounting reports in fact were available from a to the entire supply chain and in particular to align
menu directly in the SpaceCorp ERP. They were sales with production. Yet they preferred to main-
on-line and always up-to-date in real time in con- tain physical kan-ban card systems in order to se-
trast to the PowerPoint show that was created cure themselves a place of informational darkness.
manually and only updated once a week. Even if In similar ways accountants chose not to do over-
the accounting reports technically were much more head calculations by the ERP system because data-
workable than the PowerPoint figures and graphs, discipline was too low but in effect also because
everyone (including management—except for doing these calculations in spreadsheets also kept
accounting of course) preferred the PowerPoint their activities in the dark. A new Ôfactory reportÕ
slides that illustrated graphically the extent of was about to be introduced with the purpose to
Ôcontinuous improvementÕ. standardize the reporting from business units to
Furthermore, in contrast to other actors IT/ headquarters. It was a traditional paper-based re-
Logistics provided answers to the question, what port. When comparing TimeCorp to SpaceCorp
is next? It provided the translations between what is similar is the use of Ôboundary objectsÕ.
TQM, TBM and TPM and showed how the ERP In this particular dimension the cases demon-
system was an integral part of the future. TQM, strates that the need for adaptations to informa-
TBM and TPM were given Ômaterial textureÕ tion systems is global. What is local is only their
through functionalities identified in the ERP sys- adaptation. At TimeCorp local adaptations were
tem. It told the storyline of the past to the future. debated in relation to the question: how to make
In several ways IT/Logistics thus constituted itself technology part of future integration work. In con-
as a Ôcalculation centreÕ that was able to connect it- trast at SpaceCorp, repositories were debated in
self, the ERP system and other users with certain relation to the properties of accounting and logis-
Ôboundary objectsÕ that even allowed the organiza- tics respectively.
tion to unlearn certain other practices (such as for
example reading accounting reports in order to
know the state of things). ERP systems and management control work
What our cases show is how both the prospect
and practice of integration through ERP with sup- The analysis of TimeCorp and SpaceCorp
plements changes Ôpractical politicsÕ about who shows that ERP systems are highly involved in
does what, when and where. This may not neces- transforming and establishing management con-
sarily lead to more sophisticated controls—at least trol agendas through concerns for integration.
not from an accounting perspective. But as our Integration is not a solution but rather a means
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 725

by which to problematize the process through Our cases illustrate that many of the material
which actors look for ways of transporting infor- consequences of ERP only arise once the technol-
mation across localities in order to establish infor- ogy is made part of something else. Aside from
mational representations that suit the needs and imposing a global standard—a technologic that
requirements of many different parties in different clearly states what it can and canÕt do given certain
places. configuration choices—significant consequences,
In neither of the firms was it possible to engage such as for example the creation of new and/or
ERP in similar forms of management control. His- the destabilization of old centres of control arise
tory matters because the configuration later be- not because of technology and not simply because
comes a significant part of the techno-logic that of its use but because of the (potential) conse-
the ERP system in place exercises on the quences of the whole situation. For example, the
organization. situation that one cannot separate factory A from
Our TimeCorp and SpaceCorp cases illustrate factory B can be mitigated by shifting from a
how the ERP configuration conditions the various financial to a non-financial representation. Our
actors in different ways. Some parties have to go cases illustrate how ERP in both organizations is
far out of their way to perform management con- acted upon and challenged. When and where they
trol (for example by using scrap rates as a surro- fail to provide Ôadequate answersÕ other actors in-
gate for variance reporting in SpaceCorp, or by vent supplements that change how and the extent
retrieving hand held and paper based management to which ERP remains a gateway—an obligatory
control reports in TimeCorp). For others, the ERP passage point. These supplements do not only
system provides a significant lever allowing them serve to make integration workable. They also
to align other functions (as for example to the tend to make ERP less immutable both in terms
agenda Ôtotal productivity maintenance programÕ of what it from an integration point of view means
in SpaceCorp IT/Logistics functions). And to yet to get things connected and in terms of what from
others, ERP systems are part of a bricolage of a management control point of view means to get
technologies (as e.g. a supplement such as the things right (Dechow & Mouritsen, 2004). In other
Ôproduct-configuratorÕ in TimeCorp). words, in spite of a global syntax that we can trace
These examples illustrate that ERP systems are to the (then new) technology of relational dat-
always situated; they exist in settings where they abases, ERP systems do not define what integra-
act and are acted upon. The technologic that tion is and how it is to be developed. Our cases
ERP configurations exercise as a global standard illustrate that integration is a process, not a final
of communication makes the ERP system a sys- idea of what integration is. This is produced con-
tematic actor and yet Ôblind spotsÕ and Ôtrading currently and episodically.
zonesÕ cause periodic disruptions to global stan- The message in relation to accounting and
dards. When certain actors go out of their way information systems research is that in an ERP-
to make integration work with ERP, it is because setting the Ôconstruction of factsÕ is complex not
of its configuration. In its instantiations it can ap- only in terms of the number of modalities operat-
pear as if ERP systems are at ÔfaultÕ because in the ing at the same time but also in terms of the way
situation where integration was promised, it was in that integration is produced and forwarded
fact destabilized. However, such inference would through these different modalities. In both Time-
be erroneous. Even when the ERP technology Corp and SpaceCorp the creation of facts (Latour
was described as unpredictable, it was most of & Woolgar, 1979) involves the deletion of modal-
the time because the configuration of this technol- ities as much as anywhere else. What makes the
ogy at an earlier point had produced unanticipated setting much more complex is that the production
outcomes. As our cases illustrate, this is because it of modalities becomes more flexible and therefore
was not anticipated during the implementation can take place in many places at the same time.
what technology cannot do once the configuration While this may have changed corporate reality in
is made. certain ways, probably it is precisely this feature
726 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

that will challenge and change what accounting of a state to be reached; it is a constant problema-
researchers thought they knew about management tization of the firmsÕ affairs. It is not possible to
control. Our ERP cases illustrate how people in know when integration is reached as there can al-
different places of the same organization find com- ways be more integration or different integration.
mon languages through which to conduct joint (al- Our TimeCorp and SpaceCorp informants mostly
beit politicizing) work. They illustrate that speak about integration in relation to visionary
accounting inscriptions may in fact be much less objects like world class manufacturing or continu-
important than accountants, researchers and sys- ous improvement, and less in relation to manage-
tem vendors prefer to believe. What matters for ment control practices that favour local
the process of integration is not the sophistication interventions often with a view to clean up mess.
of the system, as does clarity about the ways in Integration has thus no finite goal, nor is it
which the global standards can be and/or needs merely technological. It is a process through which
to be ÔrepairedÕ with supplements of various kinds ERP is associated with (organizational) hope, pro-
and various outreaches. cedure (that organize various parties in relation to
Our informants had realized how the impor- each other), and technology (that inscribes a set of
tance of traditional control criteria such as visibil- local practices). In this process, integration is more
ity and comparability is often less important than than talk because it has numerous varieties of
some of us thought—believing in a great divide be- instantiations and punctuations all demonstrating
tween management and control. Control could be that ERP has certain things to offer. Even if never
based easily in Ôcommon-senseÕ information related settled, integration is thus performed, but it always
to their own practices, where controlling actions remains open-ended for anyone to debate who
are instantiated. Because of ERP the divide be- wants to and have the arguments to get beyond
tween management and control in that sense dis- Ôgoing concernÕ. In our cases, an example illustrat-
appears. Local connections between the two are ing this point in TimeCorp was operationsÕ devel-
made in the actions through which people work. opment of the Ôproduct configuratorÕ. Not only
However, this is not to say that ERP creates global was it designed to improve integration. It was also
management control by itself. Where our infor- used to question who controls what is to be sold
mants expected to receive data that were accurate, and thus to redraw the boundaries between the
accessible and shareable, they instead encountered concerns of operations vis-à-vis the concerns of
Ôblind spotsÕ and Ôtrading zonesÕ that challenged sales. Another example from SpaceCorp is how
them to find ways of working with and around IT/Logistics mobilized the MRP-controller not
the global representation of things. The integra- only as a search engine by which to separate data
tion work produced in this process was about the from two factories but also a means to question
ways that local control issues in different parts of whether management reporting and control had
the organization created the notion of global man- to be based on financial data. In both situations,
agement concern. In the following three sections it was the pursuit of integration that justified ad-
we elaborate on ways that ERP is involved in the denda and supplements to ERP.
quest for integration, the search for management The examples help illustrate that integration
control and organizational visibility. can be mobilized towards the transformation of
a wide series of elements. Even to the point where,
The quest for integration in TimeCorp, integration is to exclude an object—
such as the production space—from the gaze of
Informants from TimeCorp and SpaceCorp ERP because a situation of Ôself controlÕ would
saw integration neither as total nor based solely be more conducive for workers to integrate with
on technology. It is an effect of ERP systems tell- each other flexibly without intervention from man-
ing people what to do and what people tell ERP agement, as it was explained. Integration is in itself
systems to do. Integration exists in the sense of a visionary object. Its relation to ERP is not linear.
an explicitly stated objective but less in the form Even though the prospect of a totalizing panoptic
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 727

kind of management control may be in place via turned out to be ÔdifficultÕ, because (in both Time-
ERP, in its instantiations integration has to be Corp and SpaceCorp) the debate on consequences
seized and can be taken in many directions from of such techno-logic were pushed away by con-
as many different positions. Often this is in the cerns to make the financial database work. The ef-
form of adding to the procedures that ERP has fect experienced by both companies is that
put in place in the first instance by means of sup- accounting based management control becomes
plements outside the ERP system itself. They (more) difficult. Therefore, the observation that
may not be beautiful supplements, but they allow ERP systems have Ômoderate impactÕ on manage-
integration to happen in certain episodes. ment accounting practices is understandable
Integration may thus both mean more and less (Booth et al., 2000; Granlund & Malmi, 2002; Spa-
than the possibility to make data accurate, share- this & Constatinides, 2002). But, this observation
able and available (Koch, 1997). It may mean less may not be related to systems of management
since accuracy, shareability and availability of control.
data does not translate easily into the panoptic Although ERP is a control technology it is not a
view of centralization. They have to be supple- complete calculation machine (Cooper & Kaplan,
mented with numerous engagements and interven- 1998), which is able to integrate and monitor all in-
tions between hierarchical levels and functions puts and outputs everywhere at the same time. Be-
across the firm about the incorporation of ERP. cause of its techno-logic where a distinction
Integration may also mean more than accurate, between an accounting structure and a logistics
shareable and available data. ERP helps to re- structure is in place, an ERP system such as SAP
shape the organizational landscape where new cen- R/3 is often under exploration to develop alterna-
tres and peripheries are created and recreated as tives to financial controls based more on the logis-
the social uses of information are intertwined with tics structureÕs non-financial data which is more
the ERP technologies in place including supple- plastic than the accounting structure. As an effect,
ments that either add to them or circumvent them. a form of trading between traditional management
These observations are suggestive of a changing set control based on accounting numbers and a sys-
of premises for management control. tem of management control based on operational
non-financial data can take place. Therefore, visi-
Systems of management control bility is never total. A bricolage of controls devel-
ops in many places of the firm each of which has
The mobilization of systems of management access to different kinds of information. TimeCorp
control is related to the language games performed and SpaceCorp help us go further than the emerg-
by ERP both before and after its implementation. ing theory Ôof moderate effectsÕ (Booth et al., 2000;
The ERP systems in place explicate technologies Granlund & Malmi, 2002; Spathis & Constati-
for management control, but not necessarily a nides, 2002). Instead, they illustrate that manage-
technology for management accounting. ERP sys- ment control—in an ERP setting—is not
tems from SAP AG (at the time of our research) necessarily limited to the accounting function
built on principles of relational database technol- and that some ERP configurations even may work
ogy (Chen, 1976) and accounting was fundamental to separate management accounting practices
in developing this structure. Once systems Ôgo liveÕ from financial reporting. As illustrated by our
the structure is very difficult—and often even informants, management control information is
impossible given performance and financial con- not only collected by other functions, it is also
straints—to change. Therefore, management used here to generate new ideas about different
accounting can risk finding itself outside the circles practices. Even if not intended by anyone, ERP
of managements control once ERP Ôgoes liveÕ as systems make it possible to think about manage-
reporting dimensions are carved in computer code. ment control as an activity that is integrated with
What seemed an ÔeasyÕ solution (namely to con- commercial management rather than being func-
vert the old system directly into the new one) tionally separated from it.
728 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

From global visibility to competing visibilities our, there is a friction between the possibilities to
enable all kinds of management control proce-
Our cases also help us to say that ERP systems dures with equal efficiency. Some possibilities are
change the meaning of organizational visibility. Of- omitted, and therefore there are voices of discon-
ten, the notion of visibility is the panoptic, centra- tent when it is realised that ERP prioritises be-
lised action form, something executed as Ôaction at tween various kinds of management control. As
a distanceÕ. But, with ERP another notion of visibil- an effect it becomes clear that there are
ity becomes possible. TimeCorp and SpaceCorp multiple possibilities of what management control
illustrate the struggle to become the voice of global is about. ERP is part of a process of informa-
interests mobilized from a local situation. While vis- tional devolution, but not primary by transform-
ibility is often related to accounting information ing organizationally displaced non-accountants
being stable, mobile and combinable (Robson, into hybrid-accountants (Caglio, 2003; Scapens &
1992), it is possible to say that with ERP and its Jazayeri, 2003). The point is perhaps more clearly
supplements many expressions can speak about that propositions of management control can be
the performance of the firm. Some expressions heard across the firm because ERP requires actors
may circumvent the ERP system such as the paper to speak up about their control problems in order
based divisional report from the factories or the to make them part of the systemÕs capability.
kan-ban cards for the factory workers, but other TimeCorp and SpaceCorp illustrate that ERP
expressions involve ERP in games for control and recasts relations between existing procedures and
visibility for example through the product configu- principles of control. It becomes clear that certain
rator which pulls sales closer to factories by show- kinds of informational requirements become more
ing how complexity of products impacts on interesting because they are inscribed more clearly
manufacturing performance, and the MRP control- by the ERP systems than others. Some need only
ler that stabilises the production rhythms of the fac- few supplements while others need a lot, and in
tories. All of them speak about performance of the our examples, accounting was hardly put at centre
firm and they mobilize their perspective as not only stage as lever of control. In other words, the issue
a locally interesting procedure but also a mecha- is less in the form of adding ÔnewÕ practices than to
nism that, if generalised, would direct the firm into draw up the boundaries between these procedures
a more interesting form of integration. Each prop- and equip them with organizational significances
osition speaks differently about the firmÕs perfor- in relation to each other.
mance and each pays different attention to ERP; Whereas it is often assumed that ERP systems
they are not uniform and may sometimes clash. provide integration through technology, our
When ERP systems traverse local purposes with examples suggest that ERP systems can also be
a global model of business, the result is that differ- integrated with other technologies. ERP acts on
ent voices are aired to claim a space for local con- the organization because it is not able to maintain
trol. Possibly, part of this is Ôintegration striking all the procedures and data points that it abolishes
backÕ (Hanseth et al., 2001), but another part is when it replaces other systems. It helps develop the
that lots of actors add to the ERP system to equip relative importance of the accounting structure
it with competencies and shortcuts that create sys- and logistics structure, and management control
tems of management control as a network between moves between hierarchical and the lateral kinds
different actors on the lookout for relevance across of interventions, and in established ERP systems
the firm. some versions of management control are simpler
Management control may thus not develop pri- to sustain than others. In our examples, operations
marily by ÔinventingÕ more dimensions such as new and logistics are favoured. This does not necessar-
costing and new planning systems, but by assem- ily transform their management control practices,
bling existing practices distributed across the firm. but primarily makes them organizationally more
Since there is a limit to the number of situationally visible and acceptable than other approaches. In
specific conditions that the ERP systems can hon- SpaceCorp, for example, the move to favour
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 729

non-financial information was largely extracted to change. This however does not suggest that
from the logistics structure of the system. ERP systems drift (Ciborra, 2000). Rather it shows
Although accounting was not left out, it was left that integration rarely follows a distinct learning
alone except for those special instances where it curve. Functionalities of ERP systems are mobi-
was specifically requested due e.g. to lacking finan- lized only in situations that vary across organiza-
cial performance. The situation at TimeCorp was tional problems and solutions. They are, in turn,
similar although perhaps more extreme because present as episodes that may only be loosely cou-
the accounting function could not extract the data pled to each other. E.g. when integration is per-
it needed from the ERP system, but instead had to formed by adding to the repertoire of ERP by a
create paper-based divisional reports in order to product configurator or when hand carried infor-
retrieve certain data from the factories. mation is necessary to make an accounting report.
This shows that there is limited sequential connec-
tivity in the solutions found to make integration
Conclusions and future research implications perform around ERP and that there is hardly an
overall process of implementation going on.
We have identified a series of effects of ERP sys- Our research also shows how ERP systems in-
tems that extend both the theory of Ômoderate ef- cur a techno-logic that conditions how manage-
fectsÕ (Granlund & Malmi, 2002) and the theory ment control can be performed which in the
of ÔdriftÕ (Ciborra, 2000). Many things and moves case of SAP is a distinction between financial
happen around the ERP system, and it is difficult and non-financial data, or between an accounting
to separate it from developments in management structure and a logistics structure. These are parts
control and corporate integration work generally. of the structuring of the technology, but while the
This makes its effect both significant and ambigu- logistics structure mostly remains plastic and can
ous at the same time. ERP systems force actors to be changed at any point in practice mode, the
go out of their way to solve problems and create accounting structure is only plastic in the pre-
solutions, and through their attempts to either practice project mode. Therefore, corrections to
work with the ERP system or to circumvent it, the accounting structure are laborious and some-
they show awareness of multiple ways that ERP times this imposes a burden on actors to find
systems can act. They hardly merely make firms ways to mitigate these restrictions with new and
drift because they rarely just leave things open creative ways of using non-financial data for
ended. Some things work e.g. within the areas of management control. These observations explain
Ôfinancial accountingÕ, and some things are being to us why traditional management control in an
explored such as for example management control ERP setting may have but Ômoderate impactÕ
even if it works in mysterious ways prima facie. but also why this is no moderate conclusion.
Our cases illustrates that integration can be ex- Even if the new quest for integration is new, man-
plored and managed in various ways in relation to agement control is not reinvented when firms
ERP. Integration is less a goal than a problematiz- implement ERP. Data become accurate, share-
ing activity because the ambition to achieve Ôfull able and available to many different parties but
integrationÕ cannot be fulfilled. ERP systems re- does hardly create the panoptic dream of visibil-
quire a lot of different supplements. Many of these ity and action at a distance. ERP systems are
supplements are created to respond to crises of configured with certain problems in mind and
integration, and often they are created outside even in these complex database technologies there
the ERP system and function so as to alleviate lim- is no place for all details of all management con-
itations in the ERP system. These limitations are trol problems. Something is always left on the
not merely Ôbad implementationÕ because what outside, even if no one is left out. But they are
was a solution at a certain point in time can be a prioritised, and some are not given a primary role
problem at a later point in time. The ERP system on the centre stage and that they therefore may
brings a residue of the past that makes it resistant have to create supplements that either performs
730 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

the needed repair work to the ERP system or cir- and centres of discretion that distributed across
cumvents it altogether. many organizational places call upon information
ERP systems, rather than principally turning from the ERP system and its supplements to pro-
non-accountants into hybrid-accounts, point out duce control and accountability as inherent repre-
that management control is a collective affair be- sentations of management. In spite of the massive
tween separate commercial entities that are mutu- investments in ERP, the control system therefore
ally dependent—or at least have become so, when becomes an Ôimmediate transitoryÕ—something
their local systems were replaced by one ERP sys- that is concrete and yet fragile in the quest for more
tem. Such systems help change what it means to integration and better management control and
create organizational visibility, and rather than which therefore constantly poses the organization
bringing panoptic visibility, there is an interplay in a Ôtrading zoneÕ where getting things connected
between ERP systems telling people what to do sometimes mediate what it means to get things
and people telling the ERP system what to do. right et vice versa. It is therefore not surprising
They are both constraints and enablers and neither why ERP systems measured by traditional account-
offers an open set of possible actions. Even if ERP ing criteria do not produce radical improvements
systems may make financial accounting stronger, over previous practices. What is surprising, how-
they do not automatically make management ever, is the extent to which that we expected to find
accounting stronger. This perhaps may be a para- this radical leap step forward; how profoundly it
dox since ERP systems often are presented as tech- was anticipated that ERP systems would take us
nologies for management control. Instead, ERP from a basic language of accounting to a language
systems make separated versions of control mutu- we anticipated to much more sophisticated. In ab-
ally dependent and management control becomes sence of finding evidence of radical disruptions of
a collective affair. When the ERP system is in management accounting practices, some have con-
place, the corporate management control agenda cluded that this must be because ERP is a jugger-
becomes a varied one where most of the control naut that is difficult to control, while others have
mechanisms previously sustained in the firm in concluded that ERP is really nothing but a technol-
separate places suddenly become visible because ogy of a society increasingly defying its categories
they somehow have to be inscribed in or related to become a hybrid Ôof many things consideredÕ.
to the collective database. What is perhaps most significant about such con-
Whereas accounting functions in both firms clusions, is that these studies in their search for
could easily point to the tangible effects of ERP something to account for, do not reflect on their
in relation to new and automated consolidation own metaphorising of ÔinfrastructureÕ. Our cases
procedures, others had to be included in order to illustrate why and how ÔinfrastructureÕ matters.
understand procedures of management control. Technology acts by forcing other actors to take
And, these othersÕ ideas about control took depar- its categories seriously. Given its configuration
ture in their local situations and attempted to make there are things ERP systems in place can and can-
them of global interest so that—to an extent—they not do, which is why there probably always will be
used ERP systems to make their strategies into cor- good organizational reasons for creating supple-
porate strategies. Neither at TimeCorp nor at ments that work around the formal systems. Albeit
SpaceCorp was top management as close to the some may argue that these effects of technology are
materiality and politics of ERP systems and the modest, we find that these implications—whether
direction that corporate integration was taking in modest or not—(should) have significant conse-
these two companies, as was middle management. quences for future approaches to technology med-
The consequence hereof is that management iated accounting research.
control in an ERP world is not a property of the A primary lesson (that can also lead into future
accounting function (if it ever was). It does not de- research implications) from our cases of ERP, man-
pend on one informational hierarchy, because it is agement control and the quest for integration is that
suspended via the interaction of calculation centres control cannot be studied apart from technology
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 731

and context, because one will never get to under- 2004) for comments and input. Thirdly, we would
stand the underlying ÔinfrastructureÕ—the meeting like to thank Anthony Hopwood, Chris Chapman
point of many technologies and many types of con- and Markus Granlund for comments and discus-
trols that at the same time are pragmatic and yet sion. Last but not least, also we would like to
sophisticated. These cannot be accounted for with thank our two anonymous AOS reviewers for
the current tools of research where things are re- valuable feedback.
duced to simply being present either at the centre
or at the periphery of a calculation.
The single most significant effect of ERP identi-
fied in this study is how management, control and
technology perform as continua of the same busi- Benbassat, I., Dexter, A. S., Drury, D. H., & Goldstein, R. C.
ness-oriented discourse: what are we doing and (1984). A critique of the stage hypothesis: theory and
where are going? What makes this discourse inter- empirical evidence. Communications of the Association for
esting is that it at the same time is epistemic (in the Computing Machinery, 27(5), S476–S485.
sense that it is about the representation of data), Booth, P., Matovsky, Z., & Wieder, B. (2000). Integrated
information systems (ERP-systems) and accounting prac-
sociologic (in that it at the same time about Ôinte- tice. The Australian Experience. Paper presented at the 3rd
grationÕ and ÔclearanceÕ) and technologic (in that European Conference on Accounting and Information Sys-
it involves choices about respectively technology tems. Munich, Germany: 27–28 March.
and its configuration), which is why it ought to Bowker, G. C., & Star, S. L. (1996). How things (actor-
be taken utterly serious if we want to maintain net)work: classification, magic and the ubiquity of stan-
dards. Philosophia.
the relevance of our research in something close Bowker, G. C., & Star, S. L. (1999). Sorting things out:
to real-time. classification and its consequences. Cambridge, MA: MIT
ERP is not about visibility as such. ERP is Press.
about visibilities involving a lot of trading zones. Briers, M., & Chua, W.-F. (2001). The role of actor-networks
ERP is not interesting because it will revolutionize and boundary objects in management accounting change:
a field study of an implementation of activity-based
management, nor because it will fulfil the dream of costing. Accounting, Organizations and Society (26),
accountants and the nightmare of critical account- 237–269.
ing researchers. Most likely ERP will never be- Burns, J., & Baldvinsdottir, G. (1999). Hybrid accountants:
come a complete calculation machine. What Where do they belong and what (are they expected) to do?
makes ERP systems most interesting is probably Paper presented at the Conference on management account-
ing change—a European perspective. Manchester: 14–17
not primarily all the things they make possible, April 1999.
but that they can make those things impossible Caglio, A. (2003). Enterprise resource planning systems and
that we have grown accustomed to taking for accountants: towards hybridization? European Accounting
granted as managers, accountants and researchers. Review, 12(1), 123–153.
In that sense, what we learned was the importance Chen, P. (1976). The entity-relationship model: toward a
unified view of data. ACM Transactions on Database
of understanding local Ôblind spotsÕ and Ôtrading Systems, 1(1).
zonesÕ in order to address the global question: Chua, W.-F. (1995). Experts, networks and inscriptions in the
What SAP did and does. fabrication of accounting images: a story of the represen-
tation of three public hospitals. Accounting, Organizations
and Society(20), 111–145.
Ciborra, C. (2000). From control to drift—the dynamics of
Acknowledgments corporate information structures. Oxford.
Cooper, R., & Kaplan, R. S. (1998). The promise —and peril—
First of all we would like to thank the organiza- of integrated cost systems. Harvard Business Review,
tions and managers that made this piece of re- 109–119.
search possible. Secondly, we would like to thank Davenport, T. (1995). SAP: big change comes in big packages.
In CIO Magazine. October 15, Available from: www.cio.
the participants respectively at the ÔERP technolo- com/archive/101595_davenpor_content.html.
gies and management controlÕ meeting (Brussels, Davenport, T. (1998). Putting the enterprise into the enterprise
2002) and at the GMARS meeting (Michigan, system—if youÕre not careful the dream of information
732 N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733

integration can turn into a nightmare. Harvard Business Lodh, S. C., & Gaffikin, M. J. R. (2003). Implementation of an
Review, 121–131. integrated accounting and cost management system using
Davenport, T. (2000). Mission critical—realizing the promise of the SAP system: a field study. European Accounting Review,
enterprise systems. Harvard Business School Press. 12(1), 85–121.
Dechow, N., & Mouritsen, J. (2004). ERP manuscripts of Latour, B., & Woolgar, S. (1979). Laboratory life: the
accounting and information systems Chapter 6. In K. V. construction of scientific facts. Thousand Oaks, CA: SAGE.
Andersen & M. T. Vendeloe (Eds.), The past and Latour, B. (1987). Science in action. Harvard University Press.
future of information systems. Elsevier Information Systems Latour, B. (1991). Technology is society made durable. In J.
Series. Law (Ed.), A sociology of monsters. essays on power,
Deloitte Consulting (1998). ERPÕs second wawe—maximizing technology and domination (pp. 103–131). London:
the value of ERP-enabled processes. New York: Deloitte Routledge.
Consulting. Latour, B. (1999a). On recalling ANT Chapter 2. In J. Law & J.
Galison, P., 1999. Image and logic: meaning and material Hassard (Eds.), Actor network theory and after. Blackwell
culture in big science. In H. U. Obrist, & B. Vanderlinden Publishers.
(Eds.), Laboratorium (pp. 219–224). Latour, B. (1999b). PandoraÕs hope—essays on the reality of
Granlund, M., & Malmi, T. (2002). Moderate impact of ERPS science studies. Harvard University Press.
on management accounting: a lag or permanent outcome? McMaster, T., Vidgen, R., Wastell, D. (1998). Networks of
Management Accounting Research, 13, 299–321. association and due process in IS development. In
Hansen, A., & Mouritsen, J. (1999). Managerial technology and De Gross, et al., (Eds.), Proceedings of the joint IFIP 8.2/
netted networks ÔcompetitivenessÕ in action: the work of 8.6 working conference. Helsinki, Finland: December
translating performance in a high-tech firm. Organization, 1998.
6(3), 391–404. Mouritsen, J., & Dechow, N. (2001). Technologies of managing
Hanseth, O., Ciborra, C. U., & Braa, K. (2001). The control and the mobilization of paths Chapter 13. In P. Karnøe &
devolution: ERP and the side effects of globalization. R. Garudh (Eds.), Path creation and path dependence.
Advances in Information Systems Fall, 32(4), 34–46. Lawrence Eagleburn Associates.
Hayes, D. C., Hunton, J., & Reck, J. (2001). Market reaction to Munro, R. (1993). Just when you thought it safe to enter the
ERP implementation announcements. Journal of Informa- water: accountability, language games and multiple control
tion Systems. technologies. Accounting, Management and Information
Hirschheim, R., Earl, M., Feeny, D., & Lockett, M. (1988). Technologies(3), 249–271.
Information technology, strategy and leadership. In M. Earl Newell, S., Huang, J. C., Galliers, R. D., & Pan, S. L. (2003).
(Ed.), Information management: the strategic dimension Implementing enterprise resource planning and knowledge
(pp. 242–253). Oxford: Oxford University Press. management systems in tandem: fostering efficiency and
Holland, C. P., & Light, B. (2001). A stage maturity model innovation complementarity. Information and Organiza-
for enterprise resource planning systems use. The DATA tion(13), 25–52.
BASE for Advances in Information Systems., 32(2), Nolan, L. R. (1979). Managing the crisis in data processing.
S34–S45. Harvard Business Review, 57(2), 115–126.
Huberman, A. M., & Miles, M. B. (1998). Data management OÕLeary, D. E. (2002). Knowledge management across the
and analysis methods. In N. K. Denzin & Y. S. Lincoln enterprise resource planning life cycle. International Journal
(Eds.), Strategies of qualitative inquiry (pp. 179–210). Sage of Accounting Information Systems(3), 99–110.
Publications Inc. PA Consulting Group (1999). ValueMining—What happened
Hunton, J. E., McEwen, R. A., & Wier, B. (2002). The to the business value of the ERP-system? PA Consulting
reaction of financial analysts to enterprise resource planning Group—Research Report.
(ERP) implementation plans. Journal of Information Poston, R., & Grabski, S. (2001). Financial impacts
systems. of enterprise resource planning implementations. Interna-
Keller, G. (1999). SAP R/3 prozeßorientiert anwenden. Iteratives tional Journal of Accounting Information Systems(2),
Prozeß-Prototyping mit Prozeßketten. München: Addison- 271–294.
Wesley. Quattrone, P., & Hopper, T. (2001). What does organizational
King, J. L., & Kraemer, K. L. (1984). Evolution and organi- change mean? Speculation on a taken for granted category.
zational information systems: an assessment of NolanÕs Management Accounting Research, 12, 403–435.
stage model. Communications of the Association for Com- Robson, K. (1992). Accounting numbers as inscription—action
puting Machinery, 27(5), S466–S475. at a distance and the development of accounting. Account-
Koch, C. (1997). The big uneasy. In CIO Magazine. October 15, ing, Organizations and Society, 17(7), P685–P708.
Available from: Ross, J., & Vitale, M. R. (2000). The ERP revolution:
content.html. surviving vs thriving. Information Systems Frontiers, 2(2),
KPMG Consulting (1997). Profit-focused software package 233–240.
implementation. Research Report. KPMG Management SAP AG (1996). R/3 system. SAP business objects.
Consulting, London. SAP AG (1997). BC the R/3 process model. Release 3.1G.
N. Dechow, J. Mouritsen / Accounting, Organizations and Society 30 (2005) 691–733 733

SAP AG (1997). CA consultants handbook. sionals in BerkeleyÕs museum of vertebrate zoology. Social
Scapens, R. W., & Jazayeri, M. (2003). ERP systems and man- Studies of Science, 387–420.
agement accounting change: opportunities or impacts?—a Star, S. L. (Ed.). (1995). Ecologies of knowledge—work and
research note. European Accounting Review, 122(1), politics in science and technology. SUNY Press.
201–233. Strauss, A. S., Fagerhaugh, B., Suczek, B., & Wiener, C. (1985).
Spathis, C., Constatinides, S. (2002). ERP systems and manage- Social organization of medical work. Chicago: University of
ment accounting practice. Paper presented at 3rd conference Chicago Press.
on new directions in management accounting: innovations in Suchman, L., & Trigg, R. (1993). Artificial intelligence as
practice and research. Brussels: December 12–14. craftwork. In S. Chaiklin & J. Lave (Eds.), Understanding
Star, S. L., & Griesemer, J. R. (1989). Institutional ecology, practice: perspectives on activity and context (pp. 144–172).
‘translationsÕ and boundary objects: amateurs and profes- New York: Cambridge University Press.

View publication stats