Description
The literature has demonstrated the complex relationship between information system integration approaches, such as
Enterprise Resource Planning systems, and management control. In this paper, we begin our analysis by focussing on just
one aspect of information system integration, namely in terms of data architecture, commonly referred to as the single
database concept. We argue that whilst this particular aspect of integration should be related to perceived system success,
the variety of ways in which information might be drawn on in practice means it provides no strong basis for predicting a
link to business unit performance. Instead, building on Adler and Borys [Adler, P., & Borys, B. (1996). Two types of
bureaucracy: Enabling and coercive.
Information system integration, enabling control
and performance
Christopher S. Chapman
a,
*
, Lili-Anne Kihn
b,1
a
Imperial College Business School, Imperial College London, South Kensington Campus, London SW7 2AZ, UK
b
University of Tampere, Department of Economics and Accounting, FI-33014 Tampere, Finland
Abstract
The literature has demonstrated the complex relationship between information system integration approaches, such as
Enterprise Resource Planning systems, and management control. In this paper, we begin our analysis by focussing on just
one aspect of information system integration, namely in terms of data architecture, commonly referred to as the single
database concept. We argue that whilst this particular aspect of integration should be related to perceived system success,
the variety of ways in which information might be drawn on in practice means it provides no strong basis for predicting a
link to business unit performance. Instead, building on Adler and Borys [Adler, P., & Borys, B. (1996). Two types of
bureaucracy: Enabling and coercive. Administrative Science Quarterly, 41(1), 61–90] we argue that the level of information
system integration fosters the four design characteristics that make up an enabling approach to management control. Each
of these in turn is related to both perceived system success and business unit performance. We present PLS analysis of
survey data collected from 169 managers that broadly supports these expectations.
Ó 2008 Elsevier Ltd. All rights reserved.
Introduction
Approaches to information system integration
(ISI), such as Enterprise Resource Planning (ERP)
systems, which integrate and draw data from a com-
mon database, are fundamentally bound up with
organisational processes of accounting. They seek
to systematise and co-ordinate record keeping, the
design and implementation of structures of categor-
isation and aggregation of transactions, ultimately
allowing for the generation and manipulation of
comprehensive virtual perspectives on the nature
and ?ow of operations and resources (Chapman,
2005). Such ISI has been argued to potentially
tighten traditional forms of control, since it con-
strains opportunities for individual choice, simulta-
neously enhancing the hierarchical visibility of
those choices that remain (Orlikowski, 1991; Sotto,
1997). However, ISI has also been argued (by ven-
dors not least) to support ?exibility and innovation.
ISI potentially allows for the discovery and testing of
assumptions and relationships, allowing managers
to drill down from abstract summary data through
progressively more detailed records of organisa-
tional activity (Davenport, 1998). Thus, ISI has been
seen to present apparently contradictory potentials
of ?exibility and control (Orlikowski, 1991).
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.07.003
*
Corresponding author. Tel.: +44 20 7594 9335.
E-mail addresses: [email protected] (C.S.
Chapman), lili.kihn@uta.? (L.-A. Kihn).
1
Tel.: +358 3 3551 6838.
Available online at www.sciencedirect.com
Accounting, Organizations and Society 34 (2009) 151–169
www.elsevier.com/locate/aos
Studies of the e?ects of such systems in practice
show that neither ?exibility nor control, organisa-
tion nor disorganisation, ?ow directly from devel-
opments in technology (e.g. Bashein, Markus, &
Finley, 1997; Knights & Vurdubakis, 2005). ISI
does help to spread standardised information archi-
tectures, and these do act to shape organisational
priorities and procedures, providing what Dechow
and Mouritsen (2005) refer to as a ‘Techno-logic’.
ISI is not a direct cause of behaviour, however,
but might be better considered as a formative con-
text (Ciborra & Lanzara, 1994). Field studies of
technology and control amply demonstrate the
complex nature of the relationship between technol-
ogy and management practice (e.g. Caglio, 2003;
Dechow & Mouritsen, 2005; Granlund & Malmi,
2002; Orlikowski, 1991; Quattrone & Hopper,
2001, 2005).
The limited number of studies that have
attempted to test the relationship between ISI, such
as ERP, and performance have also demonstrated
the complex nature of this relationship. Poston
and Grabski (2001), for example, found no signi?-
cant improvement following ERP adoption in terms
of residual income, an e?ect that has become known
in subsequent studies as the ‘‘productivity para-
dox”. They do ?nd an e?ect whereby the ratio of
cost/sales is improved after a three year period. In
discussing this e?ect they note the fact that ERP
represent not just an integrated information archi-
tecture, but may also entail business process re-engi-
neering which may have caused the e?ect that they
found. Hunton, Lippincott, and Reck (2003)
explore these ideas further, observing a di?erence
in performance e?ect between adopters and non-
adopters. In their study, however, these statistical
e?ects were caused not by increased performance
of adopters, but by decreased performance of non-
adopters. In their discussion, Hunton et al. (2003)
partially explain this e?ect through the bundling
of the technical aspects of ERP with the idea that
ERP might also represent an innovative business
strategy.
In considering the possible performance implica-
tions of ISI then, the literature suggests that it o?ers
the potential to enhance performance, but that real-
ising this potential is dependent on the ways in
which people work. This intuition is reinforced by
the literature on the Resource Based View of the
?rm that holds that Information and Communica-
tion Technologies do not represent a source of com-
petitive advantage since they are too easy to
replicate (Barney, Wright, & Ketchen, 2001). What
would be required to support competitive advantage
would be the development of managerial
competence.
Newell, Huang, Galliers, and Pan (2003) o?er
empirical support for this line of reasoning. They
report on the mutual dependence of the information
processing e?ciency emphasis (arising from an ERP
implementation) together with the development of
organisational knowledge exploration and exploita-
tion capability (arising from a simultaneous exercise
in knowledge management) in their case company.
Adler and Borys (1996) o?er a more speci?c frame-
work for thinking through the notion of ISI-related
managerial competence. Drawing on analysis in the
?eld of human machine interface design Adler and
Borys (1996) discuss an enabling approach to con-
trol. Their framework is particularly relevant since
it is developed from detailed analysis of technical
design characteristics together with an overarching
philosophy of control through which these technical
characteristics may be expected to shape action.
The overarching concept of enabling control is
articulated through four design principles. Each of
these are premised on a notion of control that works
with rather than replaces user’s intelligence and
experience: repair, internal transparency, global
transparency, and ?exibility. Repair refers to the
breakdown of control processes, providing capabil-
ities for ?xing them. Internal transparency is about
understanding of the working of local processes.
Global transparency refers to the understanding of
where and how the local processes ?t into the orga-
nisation as a whole. Flexibility attends to the organ-
isational members’ discretion over the use of control
processes (i.e., to the extent that they can turn them
o?).
In the restaurant chain that they study (Ahrens &
Chapman, 2004) ?nd that an enabling approach to
control helped committed employees to do their
jobs more e?ectively o?ering managers the chance
to contribute to objectives of both ?exibility and
e?ciency. Whilst developing the implications of
enabling control in the context of management con-
trol systems through their analysis of a longitudinal
?eld study, they suggest the potential of the frame-
work to inform future questionnaire-based research
looking at the role of management control systems
(MCS).
A particular challenge in approaching such
research lies in the measurement of performance.
In this study we build on the observation by Ittner,
152 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
Larcker, and Randall (2003) that whilst past
research has often examined perceived system suc-
cess as an outcome variable, its relationship to
actual performance (as demonstrated in their own
analysis) is unclear. They call for further research
that might begin to unpick the relationship between
the causal factors behind the technical and organisa-
tional factors that contribute to perceived system
success and those related to actual performance.
In this study we analyze associations between ISI,
enabling budgeting, system success, and business
unit performance.
We report below on the ?ndings of a question-
naire-based study that sought to test these ideas
through a cross-sectional questionnaire. The empir-
ical results presented are based on partial least
squares analysis of the 169 (56.3%) usable question-
naires returned by a sample of Finnish managers.
Consistent with the extant literature, discussed fur-
ther below, the level of ISI is found to have a posi-
tive direct e?ect on perceived system success,
however, we ?nd that it does not have direct links
to business unit performance. Instead we ?nd it
has direct positive associations with three of the
individual design characteristics of an enabling
approach to control, which in turn have links to
perceived system success, and also to business unit
performance.
Hypotheses development
A common claim made for ISI is that it enables
more sophisticated and ?exible forms of analysis
leading to enhanced performance. The guiding intu-
ition behind our hypotheses (summarised in Fig. 1
below) is that it is the integrated data architecture
characteristics underlying ISI that support these
capabilities, and that these in turn will positively
a?ect perceived system success. That is to say the
system will be perceived to be doing a better job
of ful?lling its role. Given the complex nature of
relationship between managerial action and ISI we
do not expect direct e?ects on performance how-
ever. Rather, we argue that the nature of ISI is such
that it fosters an enabling approach to management
control in terms of each of the four design charac-
teristics identi?ed by Adler and Borys (1996): repair,
internal transparency, global transparency, and ?ex-
ibility. We hypothesise that it is these characteristics
of the management control system that lead to
enhanced outcomes in terms of perceived system
success but also in terms of business unit
performance.
The literature has produced a well-agreed de?ni-
tion of what information system integration approa-
ches, such as ERP, are in principle. Quattrone and
Hopper (2005), for example, cite Newell et al.
(2003), in turn citing Lee and Lee’s (2000) de?nition
of ERP as:
‘‘enterprise wide packages that tightly integrate
business functions into a single system with a
shared database.”
Studies examining the introduction of ERP sys-
tems have demonstrated that they might be imple-
mented for a wide variety of reasons, some quite
modest in scope such as the standardisation of dis-
parate general ledger schemes (Bashein et al.,
1997), some far more ambitious in their implica-
tions for the re?guring of organisational processes
and truncation and automation of work?ows
(Chapman & Chua, 2003). The practical de?ni-
tional problems this raises in terms of what does
ISI mean can quickly be appreciated by examining
Appendix 1 of Granlund and Malmi’s (2002)
study. This appendix neatly demonstrates the con-
siderable diversity of motivation, objectives, and
scale (in terms of which ERP modules had been
implemented to date, for example) in the 10 case
companies they studied.
We argue that whilst, the scale and scope of ISI
may vary, a central concept from the de?nition
above that can inform our understanding of its nat-
ure is the integrated information architecture,
summed up as the single database concept. It is this
structured approach to the information architecture
that supports the information processing e?ciency
emphasis of ERP (e.g. Newell et al., 2003). This sin-
gle database approach is frequently referenced
through the counterpointing of ERP against previ-
ously fragmented legacy information systems
(Davenport, 1998).
The single database concept traces directly back
to the earlier technology of relational databases
(Kallinikos, 2004). Standard texts on databases,
such as Elmasri and Navathe (1989), explain that
rather than storing corporate information in a series
of independent systems, the relational database
approach integrates organisational data. This inte-
gration is achieved through the application of vari-
ous algorithms whose primary purposes are to
reduce data redundancy, thus reducing information
storage requirements. The main bene?t of such an
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 153
approach comes from the fact that it is no longer
necessary to make multiple updates of the same
piece of information in the various independent
systems.
To give a simple example, under an un-integrated
information system approach, a customer’s tele-
phone number might be stored in many independent
systems. When that telephone number changes it
must be updated correctly in each of these systems,
this is both time consuming and signi?cantly
increases the chances of inaccurate information
being retained and used. ISI by contrast would
involve the de?nition of the customer as a system-
wide entity, whose various attributes (including tele-
phone number) would be stored in only one place,
facilitating data-retrieval, updating and most
important, accuracy.
Nelson, Todd, and Wixom (2005) carried out a
comprehensive study of the antecedents of perceived
information system success, identifying a series of
drivers of both information, and system quality.
Their analysis identi?es data accuracy as an intrinsic
property of an information system that signi?cantly
drives perceived system success. Their study also
identi?es completeness of information coverage as
also strongly linked to success. These various argu-
ments lead us to our ?rst hypothesis:
H1: Information system integration is positively
related to perceived system success.
In the accounting literature, Ittner et al. (2003)
found very di?erent relationships between their
independent variables relating to performance mea-
surement practices and their two dependent vari-
ables, system success and ?rm performance. Their
footnote 18 reports on some supplementary analysis
in which they found a negative relationship between
their two dependent variables. In following up on
this they note that past research in accounting has
shown that perceived system success is a?ected by
various organisational and technical characteristics,
making it unsuitable as a proxy for actual ?rm per-
formance. An information system may be perceived
to be technically sophisticated in its own right, and
even to facilitate communication, or reduce the need
to carry out manual reconciliations of data drawn
from di?erent sources. None of these characteristics
automatically translate into enhancements in a
?rm’s abilities to identify and achieve appropriate
goals (such as ?nancial returns, market share, or
whatever other goals are deemed strategically
important) however.
Indeed, whilst a frequent argument for ISI is the
replacement of fragmented and expensive legacy
systems, ISI itself is frequently reported as expensive
and di?cult to implement (Chapman & Chua,
2003). Empirical studies of its e?ects in practice fre-
quently emphasise the unintended nature of out-
comes (Dechow & Mouritsen, 2005; Quattrone &
Hopper, 2005). As discussed above, both Poston
Fig. 1. Structural model: IS integration, enabling use of budgeting and outcomes.
154 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
and Grabski (2001) and Hunton et al. (2003) dem-
onstrate the complex nature of links between ERP
and performance. Whilst the former study ?nds a
direct link to cost/sales ratio, they do not ?nd such
a link to residual income. As such, they show a
direct link to one aspect of performance, but not
to their more comprehensive ?nancial measure,
making the general interpretation of their results
with regards to performance complex. In their dis-
cussion of their ?ndings they emphasise the need
to consider not simply questions of information
architecture, but also more varied implications of
ISI relating to strategic and business process
reengineering.
In considering the potential for ISI to have e?ects
on performance (as opposed to perceived system
success) we feel that the most important argument
comes from the literature on the Resource Based
View of the ?rm (e.g. Barney et al., 2001). In this lit-
erature it is generally accepted that Information
Communication Technologies do not constitute a
source of long term competitive advantage since
they are too easily replicable. This kind of argument
has recently been aired more widely (e.g. Carr, 2003;
Davenport, 2006).
Whilst we have argued the case for a direct link
between ISI and perceived system success then, we
would argue that in and of itself, ISI is insu?cient
to predict the resulting e?ects on performance. An
integrated data architecture represents perhaps the
minimum form of integration necessary to qualify
a system as being integrated. Studies have amply
demonstrated that going beyond this there are a
whole raft of further decisions to be made, not least
around rearrangements to the structure of business
process. Studies examining the link to performance
?nd that these decisions are a vital component in
determining possible performance e?ects. As such
we do not expect that ISI per se is directly associ-
ated with the level of business unit performance.
2
We argue that Adler and Borys (1996) four
design characteristics of an enabling approach to
control allow for the development of a more theo-
retically sound explanation of the relationship
between ISI and performance. We expect that the
integrated information architecture underlying ISI
will foster the four design characteristics underlying
an enabling approach to control, and that these will
then positively a?ect performance. The ?rst of the
four design principles required for enabling control
is that of repair. An enabling system would be
designed such that it might be re?gured by users,
acting as a valuable resource informing their
actions. The intuition is that not everything can be
foreseen in advance, and some intellectual work
(and consequent freedom) must be left to users to
determine the appropriate course of action in such
unforeseen circumstances. The information system
might be designed such that user-driven changes
to the format and make up in terms of measure-
ments of reports are possible, facilitating what-if
type analysis rather than simply facilitating the pro-
duction of routine reports.
For the overall objective of enabling control to
be achieved, however, allowing repair is not su?-
cient (Adler & Borys, 1996; Ahrens & Chapman,
2004). E?orts at repair must be informed by a thor-
ough understanding of the nature of the system
being repaired. This leads to the second design char-
acteristic of enabling control systems, internal trans-
parency. In considering the advantages and appeal
of ISI, Davenport (1998) emphasises the attractions
that they o?er in replacing multiple, fragmented leg-
acy systems that made up the IS landscape of many
large organisations in the 1990s. ISI seeks to replace
these with a common data set, and clearly mapped
business processes. The scope and content of this
data set, and the nature of, and linkages between
business processes are determined to extensive anal-
ysis such as event process mapping (cf. Blaine,
1999).
Even where analysis of business processes in a
speci?c setting may be relatively limited, the nature
of ISI is such that they are frequently pre-?gured
with basic outlines of ‘‘best practice” processes.
The technical analysis and modular architecture of
ISI that leads to such functional simpli?cation
(Kallinikos, 2005), provides an excellent platform
for the development of a control system that can
inform its users in detail concerning the inner work-
ings of the processes it acts upon, contributing
directly to internal transparency.
A further aspect of enabling system design is glo-
bal transparency. If a system is to facilitate
responses to emerging contingencies, then it must
be used to provide a sense of where local actions
sit in relation to larger organisational strategies,
goals and agendas, for enabling control is not sim-
ply an exercise in decentralisation and delegation,
2
Whilst we do not formally hypothesise the absence of a direct
relationship between ISI and performance, in order to provide
more explicit evidence on the e?ects we also allow ISI to directly
a?ect performance in the empirical tests that follow.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 155
it is an attempt to harness local creativity and ?exi-
bility (Ahrens & Chapman, 2004). We see that ISI
might contribute to global transparency in two main
ways. Firstly the extensive process mapping e?orts,
and the exercises in standardising general ledger
accounting structures that are reported to be a very
frequent aspect of ISI (e.g. Bashein et al., 1997), are
activities that seek to position individual units
within a larger representation of organisational
activities.
More directly, however, ISI encourages interac-
tion between previously distant individuals. The
truncation of work?ows and automated integration
of activities make visible individuals’ actions as data
entered in one place ?ows through to others (in
sometimes unexpected ways). In Quattrone and
Hopper (2005), for example, previously local analyt-
ical exercises undertaken in what was thought to be
an unused accounting code became a source of dis-
turbance in previously separate analysis of that code
due to the real-time updating that was a feature of a
newly installed ERP system. ISI opens up possibili-
ties for enhanced communication. The shift away
from data-reporting work, towards more business
support work has been much commented on in rela-
tion to ERP (e.g. Scapens, Jazayeri, & Scapens,
1998). A dialogue between managers and accoun-
tants o?ers the possibility that assumptions and
implications might be fruitfully unpacked, interro-
gated and developed (e.g. Chapman, 1998; Chenhall
& Morris, 1995; Simons, 1990) contributing to glo-
bal transparency.
The ?nal design characteristic of enabling control
systems is ?exibility. Enabling systems seek to facil-
itate ?exible responses to emerging events to the
extent that the control systems can be turned o?,
when not needed. While unconstrained ?exibility is
unlikely to be bene?cial, ISI o?ers an e?ective
framework for the mapping out of individuals’ areas
of responsibility and control. They allow for the set-
ting of a variety of options ranging from blocking
speci?c actions through to automatic reporting
options such that hierarchical visibility of ?exibly
de?ned exceptions can be quickly and easily deter-
mined and implemented (Bashein et al., 1997). As
such, ISI o?ers users ?exible (but constrained)
options. The above discussion leads us to expect
that:
H2: Information system integration is positively
related to an enabling approach to management
control as seen in its four design characteristics
of (a) repair, (b) local transparency, (c) global
transparency, and (d) ?exibility.
In discussing the ambitions of ISI to control the
world through preplanning Kallinikos (2005) points
out that such systems rely on functional simpli?-
cation in which the ‘‘complexity of the world is
reconstructed as a simpli?ed set of causal and
instrumental relations” (p. 189). If functional
closure is achieved, then this may operate, but
Kallinikos highlights ‘‘the frustration caused by
technological devices that fail to deliver their prom-
ise is a reminder of the complex relationship stan-
dardisation entertains with human purpose” p.
189. ISI probably over-promises in terms of their
potential to deliver totally controlled situations in
which human skill and judgement is automated
away from daily activity (see Zubo? (1988) for a
detailed analysis of user resistance and work-
arounds in the face of such attempts, but also
Orlikowski, 1991). Adler and Borys (1996, p. 83)
likewise note that most types of automation might
bene?t from an enabling approach to control. The
expression ‘‘work to rule” underlines the require-
ment for positive participation in systems of con-
trol, a ?nding also demonstrated powerfully in
works such as Zubo? (1988). By positively seeking
to work with users, an enabling approach to control
is less likely to engender the kinds of patterns of
resistance to more coercive approaches to control,
thus enhancing perceived system success.
Given the objective of enabling control systems
of supporting users in attempts to respond ?exibly
but in a controlled fashion to emerging contingen-
cies this would suggest that an enabling approach
o?ers distinct advantages. The four enabling design
characteristics delineate an active management role
in which ?exibility and repair are expected, but cru-
cially an enabling approach o?ers of guidance as to
the ways in which these might be carried out
through the design aspects of internal and global
transparency. Global transparency in particular
aims to provide a sense of integration between local
action and wider concerns. Local transparency,
repair and ?exibility all aim to contribute positively
to supporting managers in their attempts to manage
and so might be expected to enhance perceived sys-
tem success.
H3: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
156 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
transparency, and (d) ?exibility, is positively
related to perceived system success.
Over and above the argument that an enabling
approach to control might signi?cantly reduce
forms of resistance and other disruptive activity
generated through a more coercive approach to con-
trol, learning to work together, to develop a joint
understanding of the bounds within which local dis-
cretion is meant to be exercised, and working to
meet and develop objectives ?exibly represent a
complex social achievement. As such, whilst sup-
ported by easily replicable information system
architecture, an enabling approach to management
control would constitute a management compe-
tence. Such a managerial competence would be dif-
?cult to imitate, and as such might be expected to
lead to enhanced performance.
Competences might be directed towards di?erent
strategic ends however. Ittner et al. (2003) in their
study are able to use stock market data as a perfor-
mance variable. This has the advantage that it leaves
the determination of the ?exible yet speci?c nature of
performance to the judgement of the market. Long
run investments that might depress short run earn-
ings can still be valued highly in a stock price based
measure. However, such a measure of performance
limits analysis to listed companies. Where such data
is not available, then addressing the question of per-
formance becomes complex. Relying solely on short
run ?nancial data privileges certain strategic avenues
at the expense of others. It also undercuts the discre-
tion of ?rms to strike a context speci?c balance
between the needs of shareholders versus other
potential stakeholders. In terms of theoretical expec-
tations, however, the four design principles making
up an enabling approach to control might be
expected to contribute to various aspects of business
unit performance as locally de?ned.
Repair with its emphasis on the enhancing the
ability of users to directly address problems speaks
directly to the development of competence in man-
aging the performance of the business unit. Internal
transparency by o?ering a means of structuring and
understanding the nature of local operations might
also be expected to help focus management e?ort,
enhancing performance. Global transparency
addresses questions of the wider signi?cance of local
decisions. In the context of global transparency,
internal transparency (for example) might be seen
to be a valuable tool in supporting the mobilisation
of local e?ort, to whatever end. The role of the
design characteristics of an enabling approach to
control in the de?nition and practice of strategic
objectives, together with their possibility of consti-
tuting a managerial competence as a source of com-
petitive advantage leads us to our ?nal hypothesis.
H4: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to business unit performance.
Research design
The target sample of the mail survey consisted of
300 business unit level senior managers from 86
industrial companies. The selected companies were
identi?ed using the Research Institute of the Finnish
Economy’s (ETLA’s) database of the largest com-
panies in Finland. The targeted companies were
selected randomly from the following industries:
electronics, food products, chemistry and plastics,
metal, forest, multiple businesses, construction
materials, and textile. The initial targeted managers
were mostly ?nancial, production and R and D
directors/managers. The survey was extended to
these directors in order to increase the sample size.
These roles of directors were chosen to re?ect di?er-
ent tasks with likely di?erent levels of IS integration
associated with them following empirical work on
the fragmented nature of virtualisation within
organisations (Kraut, Steinfeld, Chan, Butler, &
Hoag, 1999). Telephone calls and e-mails con?rmed
which were the most suitable business groups run
from Finland, and who were the most suitable per-
sons to be contacted to participate in the survey.
The mail survey was administrated using a
slightly modi?ed version of Dillman’s (2000) mail
survey method. After translating the questionnaire
from English to Finnish and pre-testing it, it was dis-
tributed by ?rst class mail to the targeted partici-
pants in the late spring of 2004. The initial
questionnaire request and three follow-ups yielded
173 (57.7%) anonymously ?lled questionnaires by
the deadline. Since most respondents used pre-coded
return envelopes, it seems that there are responses
from at least 98 (83%) business groups and from at
least 75 (87%) industrial corporations. The average
age of respondents was about 46 years with an aver-
age length of employment in their current position of
5.7 years. One hundred and forty four of the respon-
dents were men, and 19 women. According to the
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 157
responses, the business units had 26–9700 employees
and, on average, about 820 employees. The reported
sales of the business units ranged from 0.5 million to
6000 million euros, and averaged 241 million euros
(n = 152). Further information about the functional
area, educational background and industry category
of the respondents is shown in Table 1.
Four responses with multiple missing values on
dependent variables were excluded from hypothesis
testing to avoid any arti?cial increase in relation-
ships with independent variables, as recommended
in Hair, Black, Babin, Anderson, and Tatham
(2006, p. 56). For the remaining sample of 169 cases,
the level of randomness was tested with Little’s
missing completely at random (MCAR) test and
the result was found acceptable (v
2
505,830, DF
577, and Sig. 0.985). Therefore, given that the level
of missing data was acceptably low and missing
data did not seem to occur in a speci?c non-random
fashion, in a small number of other cases, individual
missing values were replaced with mean values. This
resulted in 169 responses (56.3%) deemed suitable
for hypothesis testing. To test if the timing of
responding provided bias in results, tests of one-
way analysis of variance in construct scores across
the timing of received questionnaires (i.e., after the
?rst, second, third or fourth contact) were under-
taken. No statistically signi?cant di?erences were
found at the 5% level.
Measures
The survey questionnaire was designed to collect
information on the integration of information sys-
tems, the degree to which the budgeting processes
re?ected the four design principles of enabling con-
trol, perceived systems success, and performance. In
addressing management control systems we chose to
focus our questionnaire on the budgeting process in
particular since this has been identi?ed as one of the
most signi?cant, widespread and long standing
aspects of organisational management control sys-
tems (e.g. Covaleski, Evans, Luft, & Shields, 2003;
Hansen, Otley, & Stede, 2003). This general ?nding
is con?rmed in our speci?c research setting since
well over 90% of Finnish companies (with over
16.7 million euro turnover) have been found to have
an annual budget (Ekholm & Wallin, 2000). Whilst
recognising a limitation in our research of focussing
on only budgeting, our focus does overcome the
challenges of trying to understand and integrate
results from questions addressing a wider range of
likely di?erentially and far less frequently adopted
aspects of MCS.
New measures were developed for several of
these variables, in particular addressing the enabling
characteristics of the budgetary process. As a ?rst
step, two rounds of interviews were conducted in a
sample of nine managers of a highly successful busi-
ness unit in Helsinki to explore possible impacts of
ISI on enabling budgetary control. The researchers
then iteratively developed a large set of questions
that addressed the four design principles. These
were systematically re?ned and questions that
remained ambiguous or poorly related to the con-
struct were dropped. The validity of measures was
enhanced by textbook de?nitions, and extensive
pilot testing in a sample of 10 professionals (six
business managers and four academic scholars) to
gain their views on the clarity and focus of the ques-
tions.
3
Finally, after checking the factorability of
items,
4
the design of all new measurement instru-
ments was based on the results of principal compo-
Table 1
Respondents by functional area, education, and industry
(n = 169)
n %
Panel A: Functional area
Accounting and ?nance 64 38
Production 41 24
Research, development, etc., 38 22
Other areas 26 15
Panel B: Educational background
Academic degree 116
Community college degree 45
Advanced degree 16
Panel C: Industry category
Metal 37 21.9
Foodstu?s and beverages 33 19.5
Multiple industries 30 17.8
Electronics 23 13.6
Chemicals and plastics 22 13.0
Forest 17 10.1
Textile 2 1.2
Energy 2 1.2
Furniture 2 1.2
Construction materials 1 0.6
3
The questions were translated from English to Finnish by the
second author, who is ?uent in both languages. In addition, the
translation of the survey items was independently double-
checked.
4
The Bartlett test of sphericity showed that nonzero correla-
tions existed at the signi?cance level of 0.000 for all the variables.
The Kaiser–Meyer–Olkin measure of sampling adequacy was met
in all cases with an MSA of >0.542.
158 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
nents analysis with oblique Oblimin rotations,
5
and
Cronbach (1951) alpha statistics of internal reliabil-
ity (Nunnally, 1978). Table 2 contains an overview
of the wording of items in the ?nal analysis (the full
questionnaire is appendicised) together with the
results of the factor and reliability analysis. Descrip-
tive statistics, based on the average scores of multi-
item variables, are presented in Table 3.
Information system integration
To measure information system integration the
questionnaire contained four questions using a
seven-point Likert-type scale ranging from disagree
completely (1) to agree completely (7). A two com-
ponent solution was discovered. The ?rst compo-
nent is used here since it obtained a Cronbach
alpha of 0.68 that was judged acceptable (Nunnally,
1978).
6
It includes the following items: information
in reports produced by our information systems is
entirely based on common sources of data (e.g. a
common database), and we have fully-integrated
information systems that contain both ?nancial
and non-?nancial information. This component
explains 75.3% of the variation. It is called informa-
tion system integration (ISI).
The budgeting processes and management
Twenty items were initially designed to measure
budgeting processes, four questions for each of the
design characteristics of enabling control. Since
Ahrens and Chapman (2004) discuss the potential
for the enabling design characteristics to extend
our understanding beyond studies that have
explored the role of interactive control systems we
also included four questions addressing the possibil-
ity of interactive budgetary control as developed by
Abernethy and Brownell (1999) to explore this issue.
The de?nition of budgeting processes was provided
(i.e., the whole range of activities – including meet-
ings – relating to the preparation, monitoring and
updating of formal business unit quantitative plans
within the business unit). Likert-type scale ranged
from disagree completely (1) to agree completely
(7). As Table 2 shows, the results of the principal
components analysis reveal four factors whose
eigenvalues are greater than one. In all, they explain
about 64.8% of the variance. The items for interac-
tive control did not load on a separate factor, rather
one of them loaded on global transparency and
another on ?exibility as an item suggesting that
interactive control is related, but may not be a
broad enough concept when set in the context of a
wider analysis of budgetary process.
Success of budgetary control and information sys-
tems was assessed by four questions about whether
the overall bene?ts outweigh the costs and whether
they are right tools for managing the business unit,
and not simply whether the systems might be felt to
be a success on other terms (technical sophistica-
tion, etc.). The principal component factor analysis
yielded one factor, with an eigenvalue greater than
one. It explained 57.0% of the variance, and
obtained a reliability estimate of 0.74.
Performance was measured applying an instru-
ment developed by Govindarajan (1984) and
Govindarajan and Fisher (1990). Following Chen-
hall and Lang?eld-Smith (1998), respondents were
asked to rate the performance of their business unit
on ten dimensions relative to competitors during the
past year. A seven-point scale ranging from ‘‘unsat-
isfactory” to ‘‘excellent” was used. However, rather
than making an attempt to measure a weighted
overall organisational performance index, we con-
ducted factor analysis to analyze various dimen-
sions of performance in greater detail.
Principal components factor analysis results
identi?ed three components with eigenvalues over
one for the items measuring perceived performance
relative to competitors. The ?rst component can be
interpreted as market performance. It consists of
development of new products, sales volume, market
share and market development, and explains about
44.42% of total variance. The second component is
called ?nancial performance. It consists of return on
investment, pro?t, and cash ?ow from operations,
explaining 19.12% of variance. The third compo-
nent is labelled as social responsibility. It includes
personnel developments and political-public a?airs.
It explains 11.14% of variance. Financial perfor-
mance, market performance, and social responsibil-
ity were used in the subsequent data analysis to aid
in a more detailed analysis of non-?nancial and
5
According to Hair et al. (2006, p. 125), the oblique rotation
method is more ?exible and more realistic, because the theoret-
ically important underlying dimensions are not assumed to be
uncorrelated with each other. However, Varimax rotations were
also checked and found to provide identical components. Given
the sample size, an objective of obtaining a power level of 80%
and the use of a 0.05 signi?cance level, factor loadings were
expected to be around 0.45 at a minimum in the sample of 169 to
be considered signi?cant (Hair et al., 2006). Raw data rather than
factor scores were used in the PLS analysis however.
6
The second component was dropped from the analysis due to
too low reliability coe?cient.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 159
Table 2
Principal components analysis (Oblimin rotated) and Cronbach alphas (n = 169)
Components
I II III IV
A. Information system integration (a 0.68, Eigenvalue = 1.51, 75.27% of variance)
5. Information in reports produced by our information systems is 0.868
entirely based on common sources of data (e.g. a common database)
7. We have a fully-integrated information systems that contain both ?nancial and non-?nancial
information
0.868
B. Budgeting and management
Repair (a 0.67, Eigenvalue = 1.22, 8.74% of variance)
2. It is easy for me to modify budget information and reports (or have them modi?ed for me) 0.086 0.045 0.129 À0.793
5. I easily get access to very detailed information in order to investigate budget deviations 0.124 À0.001 À0.035 À0.828
Internal transparency (a 0.87, Eigenvalue = 4.901, 35.05% of variance)
6. The budgeting process helps to clarify the activities that make up my business unit 0.802 À0.066 À0.006 À0.054
7. The budgeting process increases my knowledge of the operations of my business unit 0.902 0.082 À0.080 À0.030
8. The budget process increases my understanding of what drives our revenue/cost levels 0.760 À0.061 0.049 À0.163
9. The budget process increases my knowledge of how my business unit works as a whole 0.767 0.038 0.187 0.007
Global transparency (a 0.81) (Eigenvalue = 1.325, 9.47% of variance)
10. The budgeting process helps to communicate business unit strategy 0.256 0.062 0.620 0.098
11. The budgeting process signals areas in which we may need to change business unit strategy 0.052 0.161 0.793 0.083
12. The budgeting process helps personnel in my business unit to understand the overall context in
which they are working
0.304 0.084 0.565 0.027
13. I analyse budget information in order to come up with ideas for improving operations under
my control
À0.008 À0.176 0.701 À0.065
16. I often think of new ways of doing things during the budgeting process À0.220 0.023 0.744 À0.205
19. Our budgeting process aims to generate regular and frequent ?ows of strategic information
between operational and senior management
0.141 À0.183 0.600 0.025
Flexibility (a 0.51, Eigenvalue = 1.623, 11.60% of variance)
14. I can only make expenditures that have been built into the budget
a
0.070 0.841 0.099 0.141
20. Discussion during the budgeting process focuses on ensuring strict À0.084 0.766 À0.135 À0.232
adherence to original assumptions and action plans
a
C. Success of budgeting process and information systems (a 0.75, Eigenvalue = 2.28, 57.02% of variance)
1. Overall the bene?ts of our budgeting process outweigh the costs 0.731
2. Overall the bene?ts of the information systems that we use in our budgeting process outweigh the
costs
0.766
3. I am convinced that our budgeting process is the right tool for managing this business unit 0.821
4. I am convinced that the information systems that we use in our budgeting process are the right
tools for managing this business unit
0.697
D. Business unit performance relative to competitors
Financial performance (a 0.92, Eigenvalue 1.72, 19.12% of variance)
1. Return on investment 0.057 À0.923 À0.024
2. Pro?t 0.096 À0.911 À0.034
3. Cash ?ow from operations À0.078 À0.905 0.032
Market performance (a 0.80, Eigenvalue 4.00, 44.32% of variance)
5. Development of new products 0.764 0.196 À0.028
6. Sales volume 0.769 À0.139 0.013
7. Market share 0.772 À0.178 0.029
8. Market developments 0.702 À0.178 0.147
Social responsibility (a 0.67, Eigenvalue 1.00, 11.14% of variance)
9. Personnel developments 0.122 À0.039 0.793
10. Political-public a?airs À0.077 0.061 0.920
a
Reverse coded question.
160 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
?nancial performance e?ects relative to peers, when
there is lack of archival performance data as sug-
gested in Kihn (2005).
The descriptive statistics in Table 3 show that on
average, the extent of information system integra-
tion is average (with a mean of 4.02), but the actual
range is large. Of the four components of enabling
budgeting, internal transparency receives the highest
score (with a mean of 5.48). The mean scores of glo-
bal transparency and repair are also fairly high (i.e.,
4.70 and 4.67, respectively). Flexibility is the one
with the lowest mean score (of 4.56). The perceived
success of budgetary control and information sys-
tems is somewhat above average against the scale
(with a mean of 4.50). Of the three performance
components, ?nancial performance is perceived to
be highest (with a mean of 4.65), followed by market
performance (with a mean of 4.53), and by social
responsibility (with a mean of 4.47).
Pearson correlation coe?cients presented in
Table 4 indicate several associations between the
design characteristics of enabling control. Repair
is positively associated with all the other design
characteristics, and it has statistically signi?cant
relationships with both global transparency (0.280,
p < 0.000) and internal transparency (0.318,
p < 0.000). As could be expected, there is also a
strong positive correlation between internal and glo-
bal transparency (0.552, p < 0.000). Flexibility,
however, has close to zero and statistically insignif-
icant correlations with the other design
characteristics.
Survey data analysis
Greater use of structural equation modelling
(SEM) in management accounting research has
been called to overcome some of the limitations of
more traditional statistical techniques (Shields,
1997; Shields & Shields, 1998; Smith & Lang?eld-
Smith, 2004). It has been considered particularly
useful when one dependent variable becomes an
independent variable in subsequent dependence
relationships (Hair et al., 2006). In this study, the
hypotheses were tested with SEM analysis using
partial least squares (PLS) technique. There are a
Table 3
Descriptive statistics for variables (n = 169)
Variable Mean SD Theoretical range Actual range
1. IS integration 4.02 1.56 1.00–7.00 1.00–7.00
2. Repair 4.67 1.41 1.00–7.00 1.00–7.00
3. Internal transparency 5.48 1.01 1.00–7.00 2.00–7.00
4. Global transparency 4.70 1.00 1.00–7.00 1.33–7.00
5. Flexibility 4.56 1.25 1.00–7.00 1.00–7.00
6. Success 4.50 1.01 1.00–7.00 1.75–6.75
7. Financial performance 4.65 1.40 1.00–7.00 1.00–7.00
8. Market performance 4.53 1.04 1.00–7.00 2.00–7.00
9. Social responsibility 4.47 0.96 1.00–7.00 1.50–7.00
Table 4
Pearson correlation coe?cients (n = 169)
1 2 3 4 5 6 7 8 9
1. IS integration 1.000
2. Repair 0.197
**
1.000
3. Internal transparency 0.222
***
0.318
**
1.000
4. Global transparency 0.277
***
0.280
****
0.552
****
1.000
5. Budgeting: ?exibility À0.152
**
0.100 0.029 À0.063 1.000
6. Success 0.341
****
0.323
****
0.421
***
0.382
****
0.029 1.000
7. Financial performance 0.045 0.201
***
À0.065 À0.013 0.169 0.114 1.000
8. Market performance 0.216
**
0.238
***
0.126 0.193
**
0.016 0.279
****
0.446
****
1.000
9. Social responsibility 0.121 0.059 0.182
**
0.219
***
0.039 0.130 0.144
*
0.409
****
1.000
*
p < 0.10 (two-tailed).
**
p < 0.05.
***
p < 0.01.
****
p < 0.000.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 161
growing number of studies using PLS in the
accounting literature (see Hall, 2008 for a recent
example and list of other studies). PLS and covari-
ance-based SEM have di?erent strengths. PLS is
suitable for causal-predictive analysis of complex
relationships with multiple dependent variables,
when there is scarcity of prior theoretical knowl-
edge, and/or when the sample size is relatively small
(Chin & Newsted, 1999; Wold, 1980). It o?ers the
ability to perform path-analytic modelling with
latent variables. In our study, PLS was used to test
re?ective links between constructs and measures
(indicators) meaning that indicators are believed
to re?ect the unobserved, underlying construct, with
the construct giving rise to the observed measures
(Bisbe, Batista-Foguet, & Chenhall, 2007; Hulland,
1999).
The PLS approach comprises a measurement
model that speci?es relations between manifest items
(observed values for speci?c survey questions) and
the latent constructs that they represent (i.e., unob-
served values de?ned as the sum of their respective
indicators). It also comprises a structural model that
identi?es relations among constructs. To maximize
the interpretability of both models, the PLS model
was analyzed and interpreted sequentially in two
stages: the reliability and validity of the measure-
ment model was assessed ?rst, followed by the
assessment of the structural model (Hulland, 1999).
The preliminary analyses of uni-dimensionality
and reliability of multi-item constructs (i.e., the fac-
tor analysis and Cronbach alphas) were reported
and presented above (see Table 2). As Table 5
shows, the PLS analysis con?rmed the preliminary
factor analysis of the variables by showing high
composite reliability of latent variables (i.e., from
0.78 to 0.95) and high (over 0.70) loadings of man-
ifest variables on latent variables. A few items
loaded marginally below 0.70 in some of the tests
(i.e., three global transparency items being 0.693,
0.686 and 0.556; and a ?exibility item being
0.634). According to Hulland (1999), these loadings
can be considered acceptable for explorative
research.
Convergent validity of constructs was demon-
strated by calculating average variance extracted
(AVE). As the fourth column in Table 5 shows, all
variables were well above 0.50, which according to
Chenhall (2005, p. 411) is the conventional guideline
for adequate convergent validity. The discriminant
validity of the measurement model was also tested
using AVE, by comparing it with the squared corre-
lations between constructs. This determines the
extent to which a construct shares more variance
with its measures than it shares with other con-
structs (Fornell & Larcker, 1981). The AVE results
of Table 5 were compared with the correlations of
latent variables derived from Table 6. All AVE mea-
sures appeared greater than the respective squared
correlations suggesting satisfactory discriminant
validity.
7
Analysis of results
In the second phase of our PLS analysis, the ade-
quacy of the structural model was assessed with
standardised b-statistics, used as path coe?cients
and generated by the PLS, from ordinary least
squares regression (OLS). Fig. 1 shows the full
PLS model tested. Bootstrapping using 500 samples
with replacement was used to assess the signi?cance
of the path coe?cients. As recommended in the
PLS-graph user’s manual (Chin, 2001, p. 14), each
sample consisted of the same number of cases as
our original sample set (i.e., n = 169). Given that
the use of overall goodness-of-?t measures would
not be appropriate, the ?t of the PLS model was
estimated by the overall incidence of signi?cant rela-
tionships between constructs and the explained var-
iance of the dependent variables (Chenhall, 2005).
Tables 6 and 7 and Fig. 2 summarise the signi?cant
results numerically and graphically.
Test of hypotheses
As shown in Fig. 1, the research model tests the
extent to which ISI is associated with perceived sys-
tem success (H1) and performance, the extent to
which it is associated with design characteristics of
enabling approach to management control (H2a–
d), and how these design characteristics of an
enabling approach to management control are asso-
ciated with perceived system success (H3a–d) and
with performance (H4a–d).
H1: Information system integration is positively
related to perceived system success.
The PLS results in Table 7 indicate that the asso-
ciation between ISI and perceived success of systems
7
Note also that the correlation coe?cients presented in Table 4
do not indicate too high multicollinearity.
162 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
is positive and very strongly statistically signi?cant
(0.237, p < 0.005). This result provides support for
the ?rst hypotheses.
As expected, the results also indicate a lack of
signi?cant links between ISI and various perfor-
mance items. The associations between information
system integration and ?nancial performance
(0.015) and social responsibility (0.046) are close
to zero and insigni?cant. The association between
information system integration and market perfor-
mance is positive (0.115), but also statistically
insigni?cant.
H2: Information system integration is positively
related to an enabling approach to management
control as seen in its four design characteristics
of (a) repair, (b) local transparency, (c) global
transparency, and (d) ?exibility.
Table 5
Loadings of manifest variables in PLS analysis (n = 169)
Items Loadings Composite reliability Average variance extracted
A. IS integration
Item 5 0.886 0.858 0.752
Item 7 0.848
B. Budgeting and management
Repair
Item 2 0.889 0.858 0.752
Item 5 0.845
Internal transparency
Item 6 0.811 0.909 0.713
Item 7 0.875
Item 8 0.822
Item 9 0.868
Global transparency
Item 10 0.763 0.860 0.510
Item 11 0.765
Item 12 0.795
Item 13 0.686
Item 16 0.556
Item 19 0.693
Flexibility
Item 14 0.634 0.780 0.649
Item 20 0.946
C. Success of budgeting process and information systems
Item 1 0.713 0.839 0.567
Item 2 0.712
Item 3 0.842
Item 4 0.738
D. Business unit performance relative to competitors
Financial performance
Item 1 0.954 0.947 0.857
Item 2 0.950
Item 3 0.870
Market performance
Item 5 0.716 0.872 0.631
Item 6 0.818
Item 7 0.804
Item 8 0.835
Social responsibility
Item 9 0.914 0.852 0.744
Item 10 0.807
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 163
Consistent with expectations Table 7 shows sig-
ni?cant positive relationships between information
system integration and repair (0.197, p < 0.01),
internal transparency (0.219, p < 0.005), and global
transparency (0.234, p < 0.005). These results pro-
vide full support for Hypotheses 3a–c. In addition,
the results suggest a weak negative association
between information system integration and ?exible
use of budgets (À0.176) at 0.10 level of statistical
signi?cance (for a one-tailed test).
H3: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to perceived system success.
Table 8 shows how the various aspects of
enabling budgeting are associated with perceived
system success. Table 8 shows signi?cant positive
relationships between all aspects of enabling use of
budgets but ?exibility and perceived system success.
The perceived level of repair is signi?cantly associ-
ated with success (1.40, p < 0.05) suggesting that
when the level of repair as a budgeting characteristic
increases, then the perceived success of budgetary
control and information systems also increases.
The perceived extent of internal transparency is also
signi?cantly associated with success (0.208,
p < 0.025). Likewise, the level of global transpar-
ency is signi?cantly associated with success (0.216,
p < 0.01). In contrast, the association between ?exi-
bility in the use of budgets and success is close to
zero and not signi?cant in the statistical sense
(0.039, n.s.). These results suggest that most of the
investigated aspects of enabling use of budgets are
positively related to perceived system success. The
explained variance (R
2
) of this dependent variable
is 0.310. These results provide partial support for
Hypothesis 3.
H4: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to business unit performance.
The results reported in Table 8 suggest several
associations between characteristics of enabling
use of budgets and various aspects of business unit
performance. The extent of repair has positive asso-
ciations between all aspects of business unit perfor-
mance, except social responsibility. In particular, a
high extent of repair seems to increase ?nancial per-
formance (0.159, p < 0.05) and to a smaller degree
market performance (0.172, p < 0.10).
Contrary to expectations, the results reported in
Table 8 do not suggest signi?cant positive associa-
tions between the extent of internal transparency
and business unit performance. The associations
with market performance and social responsibility
Table 6
Correlations of latent variables from PLS (n = 169)
1 2 3 4 5 6 7 8 9
1. IS integration 1.000
2. Repair 0.197 1.000
3. Internal transparency 0.219 0.320 1.000
4. Global transparency 0.234 0.290 0.574 1.000
5. Flexibility À0.176 0.126 À0.012 À0.121 1.000
6. Success 0.354 0.321 0.428 0.427 À0.014 1.000
7. Financial performance À0.001 0.156 À0.075 0.030 0.184 0.034 1.000
8. Market performance 0.169 0.222 0.101 0.201 À0.004 0.241 0.422 1.000
9. Social responsibility 0.086 0.051 0.124 0.220 À0.005 0.145 0.158 0.417 1.000
Table 7
Results from PLS analysis (path coe?cients, n = 169)
Paths to Paths from
IS integration Mult. R
2
Repair 0.197
***
0.039
Internal transparency 0.219
*****
0.048
Global transparency 0.234
*****
0.055
Flexibility À0.176
*
0.031
Success 0.237
*****
Financial performance 0.015
Market performance 0.115
Social responsibility 0.046
*
p < 0.10 (one-tailed test).
**
p < 0.05.
***
p < 0.01.
****
p < 0.01.
*****
< 0.005.
164 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
are close to zero and statistically insigni?cant. The
relationship between internal transparency and
?nancial performance is statistically signi?cant,
but negative (À0.191, p < 0.05).
As expected, the level of global transparency
appears to have signi?cant positive associations
with several aspects of business unit performance.
It has a strong relationship with social responsibility
(0.225, p < 0.025), and a weaker association with
market performance (0.170, p < 0.10), but not with
?nancial performance (0.112, n.s.) in a statistical
sense.
The extent of ?exibility in the use of budgets
seems to be associated with ?nancial performance
(0.177, p < 0.05), but not with other analyzed
aspects of business unit performance. Overall, the
explained variance is 0.075 for ?nancial perfor-
mance, 0.055 for market performance, and 0.031
for social responsibility. Overall, these results pro-
vide partial support for the fourth and ?nal
hypotheses.
8
Fig. 2. Partial least squares model: IS integration, enabling use of budgeting and outcomes (n = 169). (Only statistically signi?cant paths
are shown).
Table 8
Results from PLS analysis (path coe?cients, n = 169)
Paths from Paths to
Success Financial performance Market performance Social responsibility
Repair 0.140
**
0.159
**
0.172
*
À0.025
Internal transparency 0.208
***
À0.191
**
À0.076 À0.007
Global transparency 0.216
****
0.112 0.170
*
0.225
***
Flexibility 0.039 0.177
**
0.014 0.034
Mult. R
2
0.310 0.075 0.055 0.031
*
p < 0.10 (one-tailed test).
**
p < 0.05.
***
p < 0.025.
****
p < 0.01.
8
In gathering a suitably large sample size we sent our
questionnaire to di?erent functional respondents within the same
?rm (we have 169 responses from 75 ?rms). This does raise the
possibility that non-independence of responses poses a threat to
the validity of our analysis. However, an additional ?rm-level
analysis was also conducted in which di?erent functional
respondents’ answers were averaged, and more stringent loading
requirements used resulting from the signi?cantly reduced sample
size. Whilst not identical, we feel that the qualitative similarities
in the constructs and their relationships in this second analysis
represent quite a strong test of the intuition behind our individual
hypotheses and of the analysis presented in full here.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 165
Discussion and conclusions
This study aimed to improve understanding of
how information system integration assists manag-
ers in enhancing performance. We analyzed direct
associations between these variables as well as
whether ISI is associated with the four design char-
acteristics of enabling control systems, and whether
these, in turn, are associated with performance. In
developing our hypotheses and statistical analysis
we explicitly sought to unpack the di?erence
between system success and performance as sug-
gested by Ittner et al. (2003), seeking to understand
implications for success, ?nancial and non-?nancial
aspects of business unit performance.
9
We measured system success in terms of whether
or not the system (both budgetary, and its support-
ing IS) was felt to be a cost-e?cient and an appro-
priate tool for managing the business unit. With
regards to performance, we used a factor analysis
to breakdown the overall measure of self-reported
business unit performance relative to competitors
to allow detailed analysis of the speci?c ways in
which enabling aspects of budgetary processes were
felt to contribute to ?nancial performance, market
performance and social responsibility. Such a
perceptions-based approach seemed useful in
increasing understanding of various ?nancial and
non-?nancial performance e?ects relative to com-
petitors, when archival performance data was not
available (Kihn, 2005).
As expected, our data shows a direct association
between information system integration and per-
ceived system success, but not between ISI and the
other aspects of performance. However, as
expected, positive and statistically signi?cant paths
are found between ISI and all the design principles
of enabling budgeting except ?exibility. A more
marginal negative e?ect is found between ISI and
?exibility suggesting di?culties in ?exible use of
budgets in the context of ISI. As noted in the
hypothesis development, ?exibility is an important
part of the enabling approach to control, but it is
a part of a set of design characteristics. Consistent
with prior research it seems that ?exibility might
need some supporting structure in order to lead to
positive outcomes, and so ?exibility on its own rep-
resents a necessary but not su?cient condition for
performance.
Our empirical results further show that it is the
enabling approach to budgeting that has direct
e?ects on both system success and performance.
Consistent with expectations we ?nd direct links
between all aspects of enabling budgetary control,
except ?exibility, and perceived system success.
Our results also indicate several associations
between enabling budgetary control and various
aspects of performance. Some of the statistically sig-
ni?cant e?ects between budgeting processes and
performance vary by performance dimensions, but
nearly all of them are positive as expected. The only
exception is internal transparency, which appears to
be negatively associated with ?nancial performance.
This result may re?ect greater use of budgets to try
to understand internal processes when ?nancial per-
formance is poor. That is, perhaps poor perfor-
mance leads to budget-based investigations which
lead to greater transparency, rather than greater
transparency leading to better ?nancial perfor-
mance. Strong positive associations between repair
and ?nancial performance receive support. In addi-
tion, a statistically signi?cant association between
repair and market performance is found. Global
transparency appears to be signi?cantly related to
social responsibility and to a lesser extent to market
performance.
The R
2
values of links to perceived system success
appear stronger than those with the other dimen-
sions of business unit performance (cf. Ittner
et al., 2003). Notwithstanding our e?orts to direct
our questions away from simply assessing satisfac-
tion with the system in its own terms, but the system
as a means of managing business unit performance,
some element of this kind of thinking would seem to
underlie the very strong relationship between the
extent of ISI and success, but not necessarily busi-
ness unit performance. Hence, our results also sug-
gest that management can be highly satis?ed with
their systems even if the systems do not directly
9
Our theoretical model (Fig. 1) did not include a relationship
between perceived success and performance. It was not expected.
However, following a query from a reviewer we conducted
additional tests to analyze this relationship in our data. The four
additional tests were otherwise similar, but di?ered in whether
system success or the three performance items was/were inter-
vening variables, and whether enabling budgeting items were
directly linked to both set of outcome variables, or to just one of
them. The main result was that as past studies have suggested,
our data showed no statistically signi?cant relationships between
perceived system success and ?nancial performance. Some minor
results were that the link between system success and market
performance appeared statistically signi?cant in all four cases,
and the link between system success and social responsibility
appeared signi?cant in one case. It would be interesting to try and
understand this relationship in more detail in future studies.
166 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
enhance performance in ?nancial and other terms.
Taken together, we did not get all the results we
had expected, but ?nd partial support for our
hypothesis. We draw some reassurance for the basis
for our theoretical argument from the robustness
check we carried out through a supplementary
?rm-level analysis. Whilst the results were not iden-
tical, they con?rmed the main premise concerning
direct and indirect e?ects, and also demonstrated
positive links between the enabling design charac-
teristics and various aspects of business unit
performance.
In this study, we did not aim to analyze the over-
all concept of enabling budgeting. However, note
that the empirical results of our study suggest rela-
tively high levels of design characteristics of
enabling control, and as could be expected, there
were several associations between these variables.
Repair was positively associated with all the other
design principles, and internal and global transpar-
ency were positively and signi?cantly correlated
with each other. Flexibility, however, was not signif-
icantly correlated with the other variables. Hope-
fully future research will further develop our
understanding of the complex role of ?exibility in
driving performance.
This exploratory study has several limitations
that should be considered when drawing conclu-
sions from the results. First, while prior research
has mostly analyzed ERPs, it is noteworthy that
we have contributed to the literature by analyzing
the extent of information system integration directly
through an attempt to measure the degree to which
information is drawn and integrated from a com-
mon database. More speci?c impacts of various
types of ERPs could be analyzed in future research.
Second, our analysis has focused on the e?ects of
ISI on enabling use of budgets. Possible e?ects of
other antecedent variables on enabling use of bud-
gets could also be analyzed. Likewise, enabling use
of other aspects of MCS could be explored. Our
choice to focus on budgets as a commonly used
form of MCS limits our wider understanding.
Third, some of our research instruments, in particu-
lar ?exibility, could be still further tested and devel-
oped. Fourth, the survey data report managers’
perceptions and, therefore, may not accurately rep-
resent actual business practice. Finally, the observed
e?ects do not indicate causal relationships, rather
statistical associations consistent with the theory.
Notwithstanding these various limitations we feel
that this study contributes several valuable insights
into the nature and ?nancial and non-?nancial
e?ects of ISI and budgetary control. Galbraith
(1973) argues that one way to reduce organisational
uncertainty is to invest in information processing
capabilities. Today, corporations often attempt this
through extensive investments in both information
systems, and management control systems such as
budgeting. In this study we analyzed the extent to
which ISI and enabling approaches to budgeting
go hand in hand, and, the extent to which they
impact business unit performance and success.
Consistent with past ?eld studies of technology
and control our analysis suggests that in considering
investments in ISI and MCS, managers would be
wise to explicitly include a detailed analysis of con-
trol as both a technical and social phenomenon. In
terms of pursuing a process view of budgetary con-
trol, this study demonstrates that the four design
principles of enabling control o?er a fruitful way
of systematically unpacking this challenge.
Acknowledgements
The authors thank the Academy of Finland and
the second author also the Foundation of Economic
Education for ?nancial support. We would also like
to thank participants at the American Accounting
Association Management Accounting Section Mid-
year Meeting, January 2006, Clearwater Beach,
Florida. In particular we would like to thank Anne
Lillis, Habib Mahama and Sally Widener for their
valuable comments, and Pekka Taipale for research
assistance. Finally, we also gratefully acknowledge
the many helpful comments and questions of two
anonymous reviewers.
References
Abernethy, M. A., & Brownell, P. (1999). The role of budgets in
organizations facing strategic change: An exploratory study.
Accounting, Organizations and Society, 24(3), 198–204.
Adler, P., & Borys, B. (1996). Two types of bureaucracy:
Enabling and coercive. Administrative Science Quarterly,
41(1), 61–90.
Ahrens, T., & Chapman, C. S. (2004). Accounting for ?exibility
and e?ciency: A ?eld study of management control systems in
a restaurant chain. Contemporary Accounting Research, 21(2),
271–301.
Barney, J., Wright, M., & Ketchen, D. J. (2001). The resource-
based view of the ?rm: Ten years after 1991. Journal of
Management, 27, 625–641.
Bashein, B., Markus, L., & Finley, J. (1997). Safety nets: Secrets
of e?ective information technology controls. Financial Execu-
tives Research Foundation.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 167
Bisbe, J., Batista-Foguet, J.-M., & Chenhall, R. (2007). De?ning
management accounting constructs: A methodological note
on the risks of conceptual misspeci?cation. Accounting,
Organizations and Society, 32(7–8), 789–820.
Blaine, J. (1999). Special edition using SAP R/3 (3rd ed.).
Caglio, A. (2003). Enterprise resource planning systems and
accountants: Towards hybridization?. European Accounting
Review 12(1), 123–154.
Carr, N. G. (2003). IT doesn’t matter. Harvard Business Review,
81(5), 41–49.
Chapman, C. S. (1998). Accountants in organizational networks.
Accounting, Organizations and Society, 23(8), 737–766.
Chapman, C. S. (2005). Not because they are new: Developing
the contribution of enterprise resource planning systems to
management control research. Accounting, Organizations and
Society, 30(7–8), 685–689.
Chapman, C. S., & Chua, W. F. (2003). Technology-driven
integration, automation and standardisation of business
processes: Implications for accounting. In A. Bhimani (Ed.),
Management accounting in the digital economy (pp. 74–94).
Oxford: Oxford University Press.
Chenhall, R. H. (2005). Integrative strategic performance mea-
surement systems: Strategy, strategic alignment of manufac-
turing, learning and organizational performance. Accounting,
Organizations and Society, 30(5), 395–422.
Chenhall, R. H., & Lang?eld-Smith, K. (1998). The relationship
between strategic priorities, management techniques and
management accounting: An empirical investigation using a
systems approach. Accounting, Organizations and Society,
23(3), 243–264.
Chenhall, R. H., & Morris, D. (1995). Organic decision and
communication processes and management accounting sys-
tems in entrepreneurial and conservative business organiza-
tions. Omega, International Journal of Management Science,
23(5), 485–497.
Chin, W. W. (2001). PLS graph user’s guide (Version 3.0.
February).
Chin, W. W., & Newsted, P. R. (1999). Structural equation
modelling analysis with small samples using partial least
squares. In R. H. Hoyle (Ed.), Statistical strategies for small
sample research. Thousand Oaks, CA: SAGE Publications.
Ciborra, C. U., & Lanzara, G. F. (1994). Formative contexts and
information technology. Accounting, Management and Infor-
mation Technologies, 4, 611–626.
Covaleski, M. A., Evans, H., III, Luft, J., & Shields, M. (2003).
Budgeting research: Three theoretical perspectives and criteria
for selective integration. Journal of Management Accounting
Research, 15, 3–49.
Cronbach, L. J. (1951). Coe?cient alpha and the internal
structure of tests. Psychometrika, 16, 297–334.
Davenport, T. (1998). Putting the enterprise into the enterprise
system. Harvard Business Review (July–August), 121–131.
Davenport, T. (2006). Competing on analytics. Harvard Business
Review (January), 99–107.
Dechow, N., & Mouritsen, J. (2005). On enterprise resource
planning systems, the quest for integration and management
control. Accounting, Organizations and Society, 30(7–8),
691–733.
Dillman, D. A. (2000). Mail and internet surveys: The tailored
design method. New York: Wiley.
Ekholm, B. G., & Wallin, J. (2000). Is the annual budget really
dead?. The European Accounting Review 9(4), 519–539.
Elmasri, R., & Navathe, S. B. (1989). Fundamentals of database
systems. Redwood, CA: Benjamin.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural
equation models with unobservable variables and measure-
ment error. Journal of Marketing Research, 18, 39–50.
Galbraith, J. R. (1973). Designing complex organizations. New
York, NY: Addison-Wesley.
Govindarajan, V. (1984). Appropriateness of accounting data in
performance evaluation: An empirical examination of envi-
ronmental uncertainty as an intervening variable. Accounting,
Organizations and Society, 9(2), 125–135.
Govindarajan, V., & Fisher, J. (1990). Strategy, control systems
and resource sharing: E?ects on business unit performance.
Academy of Management Journal, 33, 259–285.
Granlund, M., & Malmi, T. (2002). Moderate impact of ERPS on
management accounting: A lag or permanent outcome?.
Management Accounting Research 13(3), 299–321.
Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., &
Tatham, R. L. (2006). Multivariate data analysis (2nd ed.).
New Jersey: Pearson Prentice Hall.
Hall, M. (2008). The e?ect of comprehensive performance
measurement systems on role clarity, psychological empow-
erment and managerial performance. Accounting, Organiza-
tions and Society, 33(2–3), 141–163.
Hansen, S., Otley, D., & Stede, W. A. V. D. (2003). Practice
developments in budgeting: An overview and research
perspective. Journal of Management Accounting Research,
15, 49–70.
Hulland, J. (1999). Use of partial least squares (PLS) in strategic
management research: A review of four recent studies.
Strategic Management Journal, 20, 195–204.
Hunton, J., Lippincott, B., & Reck, J. (2003). Enterprise resource
planning systems: Comparing performance of adopters and
non-adopters. International Journal of Accounting Information
Systems, 4, 165–184.
Ittner, C., Larcker, D., & Randall, T. (2003). Performance
implications of strategic performance measurement in ?nan-
cial services ?rms. Accounting, Organizations and Society,
28(7/8), 715–741.
Kallinikos, J. (2004). Farewell to constructivism: Technology and
context-embedded action. In C. Avergrou, C. Ciborra, & F.
Land (Eds.), Social studies of information systems and
communication technologies (pp. 140–161). Oxford: Oxford
University Press.
Kallinikos, J. (2005). The order of technology: Complexity and
control in a connected world. Information and Organization,
15(3), 185–202.
Kihn, L. A. (2005). Comparing performance measurement
approaches in accounting research. Finnish Journal of Busi-
ness Economics, 5(2), 143–184.
Knights, D., & Vurdubakis, T. (2005). Information technology as
organization/disorganization. Information and Organization,
15(3), 181–184.
Kraut, R., Steinfeld, C., Chan, A., Butler, B., & Hoag, A. (1999).
Coordination and virtualisation: The role of electronic
networks and personal relationships. Organization Science,
10(6), 722–740.
Lee, Z., & Lee, J. (2000). An ERP implementation case study
from a knowledge transfer perspective. Journal of Information
Technology, 15(281–288).
Nelson, R., Todd, P., & Wixom, B. (2005). Antecedents of
information and system quality: An empirical examination
168 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
within the context of data warehousing. Journal of Manage-
ment Information Systems, 21(4), 199–235.
Newell, S., Huang, J., Galliers, R., & Pan, S. (2003). Implement-
ing enterprise resource planning and knowledge management
systems in tandem: Fostering e?ciency and innovation in
complementarity. Information and Organization, 13(1), 25–52.
Nunnally, J. C. (1978). Pscychometric theory (2nd ed.). New
York: McGraw-Hill.
Orlikowski, W. (1991). Integrated information environment or
matrix of control? The contradictory implications of infor-
mation technology. Accounting, Management and Information
Technology, 1(1), 9–42.
Poston, R., & Grabski, S. (2001). Financial impacts of enterprise
resource planning implementations. International Journal of
Accounting Information Systems, 2, 271–294.
Quattrone, P., & Hopper, T. (2001). What does organizational
change mean? Speculations on a taken for granted category.
Management Accounting Research, 12, 403–435.
Quattrone, P., & Hopper, T. (2005). A time–space odyssey:
Management control systems in two multinational organisa-
tions. Accounting, Organizations and Society, 30(7–8), 735–764.
Scapens, R., Jazayeri, M., & Scapens, J. (1998). SAP integrated
information systems and the implications for management
accountants. Management Accounting (September).
Shields, M. (1997). Research in management accounting by
North Americans in the 1990s. Journal of Management
Accounting Research, 9, 1–61.
Shields, J., & Shields, M. (1998). Antecedents of participative
budgeting. Accounting, Organizations and Society, 23(1),
49–76.
Simons, R. (1990). The role of management control systems in
creating competitive advantage: New perspectives. Account-
ing, Organizations and Society, 15(1/2), 127–143.
Smith, D., & Lang?eld-Smith, K. (2004). Structural equation
modelling in management accounting research: Critical anal-
ysis and opportunities. Journal of Accounting Literature, 23,
49–86.
Sotto, R. (1997). The virtual organization. Accounting, Manage-
ment and Information Technology, 7(1), 37–51.
Wold, H. (1980). Model construction and evaluation when
theoretical knowledge is scarce. In J. Kmenta & J. B. Ramsey
(Eds.), Evaluation of econometric models. New York: Aca-
demic Press.
Zubo?, S. (1988). In the age of the smart machine: The future of
work and power. New York: Basic books.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 169
doc_885623804.pdf
The literature has demonstrated the complex relationship between information system integration approaches, such as
Enterprise Resource Planning systems, and management control. In this paper, we begin our analysis by focussing on just
one aspect of information system integration, namely in terms of data architecture, commonly referred to as the single
database concept. We argue that whilst this particular aspect of integration should be related to perceived system success,
the variety of ways in which information might be drawn on in practice means it provides no strong basis for predicting a
link to business unit performance. Instead, building on Adler and Borys [Adler, P., & Borys, B. (1996). Two types of
bureaucracy: Enabling and coercive.
Information system integration, enabling control
and performance
Christopher S. Chapman
a,
*
, Lili-Anne Kihn
b,1
a
Imperial College Business School, Imperial College London, South Kensington Campus, London SW7 2AZ, UK
b
University of Tampere, Department of Economics and Accounting, FI-33014 Tampere, Finland
Abstract
The literature has demonstrated the complex relationship between information system integration approaches, such as
Enterprise Resource Planning systems, and management control. In this paper, we begin our analysis by focussing on just
one aspect of information system integration, namely in terms of data architecture, commonly referred to as the single
database concept. We argue that whilst this particular aspect of integration should be related to perceived system success,
the variety of ways in which information might be drawn on in practice means it provides no strong basis for predicting a
link to business unit performance. Instead, building on Adler and Borys [Adler, P., & Borys, B. (1996). Two types of
bureaucracy: Enabling and coercive. Administrative Science Quarterly, 41(1), 61–90] we argue that the level of information
system integration fosters the four design characteristics that make up an enabling approach to management control. Each
of these in turn is related to both perceived system success and business unit performance. We present PLS analysis of
survey data collected from 169 managers that broadly supports these expectations.
Ó 2008 Elsevier Ltd. All rights reserved.
Introduction
Approaches to information system integration
(ISI), such as Enterprise Resource Planning (ERP)
systems, which integrate and draw data from a com-
mon database, are fundamentally bound up with
organisational processes of accounting. They seek
to systematise and co-ordinate record keeping, the
design and implementation of structures of categor-
isation and aggregation of transactions, ultimately
allowing for the generation and manipulation of
comprehensive virtual perspectives on the nature
and ?ow of operations and resources (Chapman,
2005). Such ISI has been argued to potentially
tighten traditional forms of control, since it con-
strains opportunities for individual choice, simulta-
neously enhancing the hierarchical visibility of
those choices that remain (Orlikowski, 1991; Sotto,
1997). However, ISI has also been argued (by ven-
dors not least) to support ?exibility and innovation.
ISI potentially allows for the discovery and testing of
assumptions and relationships, allowing managers
to drill down from abstract summary data through
progressively more detailed records of organisa-
tional activity (Davenport, 1998). Thus, ISI has been
seen to present apparently contradictory potentials
of ?exibility and control (Orlikowski, 1991).
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.07.003
*
Corresponding author. Tel.: +44 20 7594 9335.
E-mail addresses: [email protected] (C.S.
Chapman), lili.kihn@uta.? (L.-A. Kihn).
1
Tel.: +358 3 3551 6838.
Available online at www.sciencedirect.com
Accounting, Organizations and Society 34 (2009) 151–169
www.elsevier.com/locate/aos
Studies of the e?ects of such systems in practice
show that neither ?exibility nor control, organisa-
tion nor disorganisation, ?ow directly from devel-
opments in technology (e.g. Bashein, Markus, &
Finley, 1997; Knights & Vurdubakis, 2005). ISI
does help to spread standardised information archi-
tectures, and these do act to shape organisational
priorities and procedures, providing what Dechow
and Mouritsen (2005) refer to as a ‘Techno-logic’.
ISI is not a direct cause of behaviour, however,
but might be better considered as a formative con-
text (Ciborra & Lanzara, 1994). Field studies of
technology and control amply demonstrate the
complex nature of the relationship between technol-
ogy and management practice (e.g. Caglio, 2003;
Dechow & Mouritsen, 2005; Granlund & Malmi,
2002; Orlikowski, 1991; Quattrone & Hopper,
2001, 2005).
The limited number of studies that have
attempted to test the relationship between ISI, such
as ERP, and performance have also demonstrated
the complex nature of this relationship. Poston
and Grabski (2001), for example, found no signi?-
cant improvement following ERP adoption in terms
of residual income, an e?ect that has become known
in subsequent studies as the ‘‘productivity para-
dox”. They do ?nd an e?ect whereby the ratio of
cost/sales is improved after a three year period. In
discussing this e?ect they note the fact that ERP
represent not just an integrated information archi-
tecture, but may also entail business process re-engi-
neering which may have caused the e?ect that they
found. Hunton, Lippincott, and Reck (2003)
explore these ideas further, observing a di?erence
in performance e?ect between adopters and non-
adopters. In their study, however, these statistical
e?ects were caused not by increased performance
of adopters, but by decreased performance of non-
adopters. In their discussion, Hunton et al. (2003)
partially explain this e?ect through the bundling
of the technical aspects of ERP with the idea that
ERP might also represent an innovative business
strategy.
In considering the possible performance implica-
tions of ISI then, the literature suggests that it o?ers
the potential to enhance performance, but that real-
ising this potential is dependent on the ways in
which people work. This intuition is reinforced by
the literature on the Resource Based View of the
?rm that holds that Information and Communica-
tion Technologies do not represent a source of com-
petitive advantage since they are too easy to
replicate (Barney, Wright, & Ketchen, 2001). What
would be required to support competitive advantage
would be the development of managerial
competence.
Newell, Huang, Galliers, and Pan (2003) o?er
empirical support for this line of reasoning. They
report on the mutual dependence of the information
processing e?ciency emphasis (arising from an ERP
implementation) together with the development of
organisational knowledge exploration and exploita-
tion capability (arising from a simultaneous exercise
in knowledge management) in their case company.
Adler and Borys (1996) o?er a more speci?c frame-
work for thinking through the notion of ISI-related
managerial competence. Drawing on analysis in the
?eld of human machine interface design Adler and
Borys (1996) discuss an enabling approach to con-
trol. Their framework is particularly relevant since
it is developed from detailed analysis of technical
design characteristics together with an overarching
philosophy of control through which these technical
characteristics may be expected to shape action.
The overarching concept of enabling control is
articulated through four design principles. Each of
these are premised on a notion of control that works
with rather than replaces user’s intelligence and
experience: repair, internal transparency, global
transparency, and ?exibility. Repair refers to the
breakdown of control processes, providing capabil-
ities for ?xing them. Internal transparency is about
understanding of the working of local processes.
Global transparency refers to the understanding of
where and how the local processes ?t into the orga-
nisation as a whole. Flexibility attends to the organ-
isational members’ discretion over the use of control
processes (i.e., to the extent that they can turn them
o?).
In the restaurant chain that they study (Ahrens &
Chapman, 2004) ?nd that an enabling approach to
control helped committed employees to do their
jobs more e?ectively o?ering managers the chance
to contribute to objectives of both ?exibility and
e?ciency. Whilst developing the implications of
enabling control in the context of management con-
trol systems through their analysis of a longitudinal
?eld study, they suggest the potential of the frame-
work to inform future questionnaire-based research
looking at the role of management control systems
(MCS).
A particular challenge in approaching such
research lies in the measurement of performance.
In this study we build on the observation by Ittner,
152 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
Larcker, and Randall (2003) that whilst past
research has often examined perceived system suc-
cess as an outcome variable, its relationship to
actual performance (as demonstrated in their own
analysis) is unclear. They call for further research
that might begin to unpick the relationship between
the causal factors behind the technical and organisa-
tional factors that contribute to perceived system
success and those related to actual performance.
In this study we analyze associations between ISI,
enabling budgeting, system success, and business
unit performance.
We report below on the ?ndings of a question-
naire-based study that sought to test these ideas
through a cross-sectional questionnaire. The empir-
ical results presented are based on partial least
squares analysis of the 169 (56.3%) usable question-
naires returned by a sample of Finnish managers.
Consistent with the extant literature, discussed fur-
ther below, the level of ISI is found to have a posi-
tive direct e?ect on perceived system success,
however, we ?nd that it does not have direct links
to business unit performance. Instead we ?nd it
has direct positive associations with three of the
individual design characteristics of an enabling
approach to control, which in turn have links to
perceived system success, and also to business unit
performance.
Hypotheses development
A common claim made for ISI is that it enables
more sophisticated and ?exible forms of analysis
leading to enhanced performance. The guiding intu-
ition behind our hypotheses (summarised in Fig. 1
below) is that it is the integrated data architecture
characteristics underlying ISI that support these
capabilities, and that these in turn will positively
a?ect perceived system success. That is to say the
system will be perceived to be doing a better job
of ful?lling its role. Given the complex nature of
relationship between managerial action and ISI we
do not expect direct e?ects on performance how-
ever. Rather, we argue that the nature of ISI is such
that it fosters an enabling approach to management
control in terms of each of the four design charac-
teristics identi?ed by Adler and Borys (1996): repair,
internal transparency, global transparency, and ?ex-
ibility. We hypothesise that it is these characteristics
of the management control system that lead to
enhanced outcomes in terms of perceived system
success but also in terms of business unit
performance.
The literature has produced a well-agreed de?ni-
tion of what information system integration approa-
ches, such as ERP, are in principle. Quattrone and
Hopper (2005), for example, cite Newell et al.
(2003), in turn citing Lee and Lee’s (2000) de?nition
of ERP as:
‘‘enterprise wide packages that tightly integrate
business functions into a single system with a
shared database.”
Studies examining the introduction of ERP sys-
tems have demonstrated that they might be imple-
mented for a wide variety of reasons, some quite
modest in scope such as the standardisation of dis-
parate general ledger schemes (Bashein et al.,
1997), some far more ambitious in their implica-
tions for the re?guring of organisational processes
and truncation and automation of work?ows
(Chapman & Chua, 2003). The practical de?ni-
tional problems this raises in terms of what does
ISI mean can quickly be appreciated by examining
Appendix 1 of Granlund and Malmi’s (2002)
study. This appendix neatly demonstrates the con-
siderable diversity of motivation, objectives, and
scale (in terms of which ERP modules had been
implemented to date, for example) in the 10 case
companies they studied.
We argue that whilst, the scale and scope of ISI
may vary, a central concept from the de?nition
above that can inform our understanding of its nat-
ure is the integrated information architecture,
summed up as the single database concept. It is this
structured approach to the information architecture
that supports the information processing e?ciency
emphasis of ERP (e.g. Newell et al., 2003). This sin-
gle database approach is frequently referenced
through the counterpointing of ERP against previ-
ously fragmented legacy information systems
(Davenport, 1998).
The single database concept traces directly back
to the earlier technology of relational databases
(Kallinikos, 2004). Standard texts on databases,
such as Elmasri and Navathe (1989), explain that
rather than storing corporate information in a series
of independent systems, the relational database
approach integrates organisational data. This inte-
gration is achieved through the application of vari-
ous algorithms whose primary purposes are to
reduce data redundancy, thus reducing information
storage requirements. The main bene?t of such an
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 153
approach comes from the fact that it is no longer
necessary to make multiple updates of the same
piece of information in the various independent
systems.
To give a simple example, under an un-integrated
information system approach, a customer’s tele-
phone number might be stored in many independent
systems. When that telephone number changes it
must be updated correctly in each of these systems,
this is both time consuming and signi?cantly
increases the chances of inaccurate information
being retained and used. ISI by contrast would
involve the de?nition of the customer as a system-
wide entity, whose various attributes (including tele-
phone number) would be stored in only one place,
facilitating data-retrieval, updating and most
important, accuracy.
Nelson, Todd, and Wixom (2005) carried out a
comprehensive study of the antecedents of perceived
information system success, identifying a series of
drivers of both information, and system quality.
Their analysis identi?es data accuracy as an intrinsic
property of an information system that signi?cantly
drives perceived system success. Their study also
identi?es completeness of information coverage as
also strongly linked to success. These various argu-
ments lead us to our ?rst hypothesis:
H1: Information system integration is positively
related to perceived system success.
In the accounting literature, Ittner et al. (2003)
found very di?erent relationships between their
independent variables relating to performance mea-
surement practices and their two dependent vari-
ables, system success and ?rm performance. Their
footnote 18 reports on some supplementary analysis
in which they found a negative relationship between
their two dependent variables. In following up on
this they note that past research in accounting has
shown that perceived system success is a?ected by
various organisational and technical characteristics,
making it unsuitable as a proxy for actual ?rm per-
formance. An information system may be perceived
to be technically sophisticated in its own right, and
even to facilitate communication, or reduce the need
to carry out manual reconciliations of data drawn
from di?erent sources. None of these characteristics
automatically translate into enhancements in a
?rm’s abilities to identify and achieve appropriate
goals (such as ?nancial returns, market share, or
whatever other goals are deemed strategically
important) however.
Indeed, whilst a frequent argument for ISI is the
replacement of fragmented and expensive legacy
systems, ISI itself is frequently reported as expensive
and di?cult to implement (Chapman & Chua,
2003). Empirical studies of its e?ects in practice fre-
quently emphasise the unintended nature of out-
comes (Dechow & Mouritsen, 2005; Quattrone &
Hopper, 2005). As discussed above, both Poston
Fig. 1. Structural model: IS integration, enabling use of budgeting and outcomes.
154 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
and Grabski (2001) and Hunton et al. (2003) dem-
onstrate the complex nature of links between ERP
and performance. Whilst the former study ?nds a
direct link to cost/sales ratio, they do not ?nd such
a link to residual income. As such, they show a
direct link to one aspect of performance, but not
to their more comprehensive ?nancial measure,
making the general interpretation of their results
with regards to performance complex. In their dis-
cussion of their ?ndings they emphasise the need
to consider not simply questions of information
architecture, but also more varied implications of
ISI relating to strategic and business process
reengineering.
In considering the potential for ISI to have e?ects
on performance (as opposed to perceived system
success) we feel that the most important argument
comes from the literature on the Resource Based
View of the ?rm (e.g. Barney et al., 2001). In this lit-
erature it is generally accepted that Information
Communication Technologies do not constitute a
source of long term competitive advantage since
they are too easily replicable. This kind of argument
has recently been aired more widely (e.g. Carr, 2003;
Davenport, 2006).
Whilst we have argued the case for a direct link
between ISI and perceived system success then, we
would argue that in and of itself, ISI is insu?cient
to predict the resulting e?ects on performance. An
integrated data architecture represents perhaps the
minimum form of integration necessary to qualify
a system as being integrated. Studies have amply
demonstrated that going beyond this there are a
whole raft of further decisions to be made, not least
around rearrangements to the structure of business
process. Studies examining the link to performance
?nd that these decisions are a vital component in
determining possible performance e?ects. As such
we do not expect that ISI per se is directly associ-
ated with the level of business unit performance.
2
We argue that Adler and Borys (1996) four
design characteristics of an enabling approach to
control allow for the development of a more theo-
retically sound explanation of the relationship
between ISI and performance. We expect that the
integrated information architecture underlying ISI
will foster the four design characteristics underlying
an enabling approach to control, and that these will
then positively a?ect performance. The ?rst of the
four design principles required for enabling control
is that of repair. An enabling system would be
designed such that it might be re?gured by users,
acting as a valuable resource informing their
actions. The intuition is that not everything can be
foreseen in advance, and some intellectual work
(and consequent freedom) must be left to users to
determine the appropriate course of action in such
unforeseen circumstances. The information system
might be designed such that user-driven changes
to the format and make up in terms of measure-
ments of reports are possible, facilitating what-if
type analysis rather than simply facilitating the pro-
duction of routine reports.
For the overall objective of enabling control to
be achieved, however, allowing repair is not su?-
cient (Adler & Borys, 1996; Ahrens & Chapman,
2004). E?orts at repair must be informed by a thor-
ough understanding of the nature of the system
being repaired. This leads to the second design char-
acteristic of enabling control systems, internal trans-
parency. In considering the advantages and appeal
of ISI, Davenport (1998) emphasises the attractions
that they o?er in replacing multiple, fragmented leg-
acy systems that made up the IS landscape of many
large organisations in the 1990s. ISI seeks to replace
these with a common data set, and clearly mapped
business processes. The scope and content of this
data set, and the nature of, and linkages between
business processes are determined to extensive anal-
ysis such as event process mapping (cf. Blaine,
1999).
Even where analysis of business processes in a
speci?c setting may be relatively limited, the nature
of ISI is such that they are frequently pre-?gured
with basic outlines of ‘‘best practice” processes.
The technical analysis and modular architecture of
ISI that leads to such functional simpli?cation
(Kallinikos, 2005), provides an excellent platform
for the development of a control system that can
inform its users in detail concerning the inner work-
ings of the processes it acts upon, contributing
directly to internal transparency.
A further aspect of enabling system design is glo-
bal transparency. If a system is to facilitate
responses to emerging contingencies, then it must
be used to provide a sense of where local actions
sit in relation to larger organisational strategies,
goals and agendas, for enabling control is not sim-
ply an exercise in decentralisation and delegation,
2
Whilst we do not formally hypothesise the absence of a direct
relationship between ISI and performance, in order to provide
more explicit evidence on the e?ects we also allow ISI to directly
a?ect performance in the empirical tests that follow.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 155
it is an attempt to harness local creativity and ?exi-
bility (Ahrens & Chapman, 2004). We see that ISI
might contribute to global transparency in two main
ways. Firstly the extensive process mapping e?orts,
and the exercises in standardising general ledger
accounting structures that are reported to be a very
frequent aspect of ISI (e.g. Bashein et al., 1997), are
activities that seek to position individual units
within a larger representation of organisational
activities.
More directly, however, ISI encourages interac-
tion between previously distant individuals. The
truncation of work?ows and automated integration
of activities make visible individuals’ actions as data
entered in one place ?ows through to others (in
sometimes unexpected ways). In Quattrone and
Hopper (2005), for example, previously local analyt-
ical exercises undertaken in what was thought to be
an unused accounting code became a source of dis-
turbance in previously separate analysis of that code
due to the real-time updating that was a feature of a
newly installed ERP system. ISI opens up possibili-
ties for enhanced communication. The shift away
from data-reporting work, towards more business
support work has been much commented on in rela-
tion to ERP (e.g. Scapens, Jazayeri, & Scapens,
1998). A dialogue between managers and accoun-
tants o?ers the possibility that assumptions and
implications might be fruitfully unpacked, interro-
gated and developed (e.g. Chapman, 1998; Chenhall
& Morris, 1995; Simons, 1990) contributing to glo-
bal transparency.
The ?nal design characteristic of enabling control
systems is ?exibility. Enabling systems seek to facil-
itate ?exible responses to emerging events to the
extent that the control systems can be turned o?,
when not needed. While unconstrained ?exibility is
unlikely to be bene?cial, ISI o?ers an e?ective
framework for the mapping out of individuals’ areas
of responsibility and control. They allow for the set-
ting of a variety of options ranging from blocking
speci?c actions through to automatic reporting
options such that hierarchical visibility of ?exibly
de?ned exceptions can be quickly and easily deter-
mined and implemented (Bashein et al., 1997). As
such, ISI o?ers users ?exible (but constrained)
options. The above discussion leads us to expect
that:
H2: Information system integration is positively
related to an enabling approach to management
control as seen in its four design characteristics
of (a) repair, (b) local transparency, (c) global
transparency, and (d) ?exibility.
In discussing the ambitions of ISI to control the
world through preplanning Kallinikos (2005) points
out that such systems rely on functional simpli?-
cation in which the ‘‘complexity of the world is
reconstructed as a simpli?ed set of causal and
instrumental relations” (p. 189). If functional
closure is achieved, then this may operate, but
Kallinikos highlights ‘‘the frustration caused by
technological devices that fail to deliver their prom-
ise is a reminder of the complex relationship stan-
dardisation entertains with human purpose” p.
189. ISI probably over-promises in terms of their
potential to deliver totally controlled situations in
which human skill and judgement is automated
away from daily activity (see Zubo? (1988) for a
detailed analysis of user resistance and work-
arounds in the face of such attempts, but also
Orlikowski, 1991). Adler and Borys (1996, p. 83)
likewise note that most types of automation might
bene?t from an enabling approach to control. The
expression ‘‘work to rule” underlines the require-
ment for positive participation in systems of con-
trol, a ?nding also demonstrated powerfully in
works such as Zubo? (1988). By positively seeking
to work with users, an enabling approach to control
is less likely to engender the kinds of patterns of
resistance to more coercive approaches to control,
thus enhancing perceived system success.
Given the objective of enabling control systems
of supporting users in attempts to respond ?exibly
but in a controlled fashion to emerging contingen-
cies this would suggest that an enabling approach
o?ers distinct advantages. The four enabling design
characteristics delineate an active management role
in which ?exibility and repair are expected, but cru-
cially an enabling approach o?ers of guidance as to
the ways in which these might be carried out
through the design aspects of internal and global
transparency. Global transparency in particular
aims to provide a sense of integration between local
action and wider concerns. Local transparency,
repair and ?exibility all aim to contribute positively
to supporting managers in their attempts to manage
and so might be expected to enhance perceived sys-
tem success.
H3: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
156 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
transparency, and (d) ?exibility, is positively
related to perceived system success.
Over and above the argument that an enabling
approach to control might signi?cantly reduce
forms of resistance and other disruptive activity
generated through a more coercive approach to con-
trol, learning to work together, to develop a joint
understanding of the bounds within which local dis-
cretion is meant to be exercised, and working to
meet and develop objectives ?exibly represent a
complex social achievement. As such, whilst sup-
ported by easily replicable information system
architecture, an enabling approach to management
control would constitute a management compe-
tence. Such a managerial competence would be dif-
?cult to imitate, and as such might be expected to
lead to enhanced performance.
Competences might be directed towards di?erent
strategic ends however. Ittner et al. (2003) in their
study are able to use stock market data as a perfor-
mance variable. This has the advantage that it leaves
the determination of the ?exible yet speci?c nature of
performance to the judgement of the market. Long
run investments that might depress short run earn-
ings can still be valued highly in a stock price based
measure. However, such a measure of performance
limits analysis to listed companies. Where such data
is not available, then addressing the question of per-
formance becomes complex. Relying solely on short
run ?nancial data privileges certain strategic avenues
at the expense of others. It also undercuts the discre-
tion of ?rms to strike a context speci?c balance
between the needs of shareholders versus other
potential stakeholders. In terms of theoretical expec-
tations, however, the four design principles making
up an enabling approach to control might be
expected to contribute to various aspects of business
unit performance as locally de?ned.
Repair with its emphasis on the enhancing the
ability of users to directly address problems speaks
directly to the development of competence in man-
aging the performance of the business unit. Internal
transparency by o?ering a means of structuring and
understanding the nature of local operations might
also be expected to help focus management e?ort,
enhancing performance. Global transparency
addresses questions of the wider signi?cance of local
decisions. In the context of global transparency,
internal transparency (for example) might be seen
to be a valuable tool in supporting the mobilisation
of local e?ort, to whatever end. The role of the
design characteristics of an enabling approach to
control in the de?nition and practice of strategic
objectives, together with their possibility of consti-
tuting a managerial competence as a source of com-
petitive advantage leads us to our ?nal hypothesis.
H4: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to business unit performance.
Research design
The target sample of the mail survey consisted of
300 business unit level senior managers from 86
industrial companies. The selected companies were
identi?ed using the Research Institute of the Finnish
Economy’s (ETLA’s) database of the largest com-
panies in Finland. The targeted companies were
selected randomly from the following industries:
electronics, food products, chemistry and plastics,
metal, forest, multiple businesses, construction
materials, and textile. The initial targeted managers
were mostly ?nancial, production and R and D
directors/managers. The survey was extended to
these directors in order to increase the sample size.
These roles of directors were chosen to re?ect di?er-
ent tasks with likely di?erent levels of IS integration
associated with them following empirical work on
the fragmented nature of virtualisation within
organisations (Kraut, Steinfeld, Chan, Butler, &
Hoag, 1999). Telephone calls and e-mails con?rmed
which were the most suitable business groups run
from Finland, and who were the most suitable per-
sons to be contacted to participate in the survey.
The mail survey was administrated using a
slightly modi?ed version of Dillman’s (2000) mail
survey method. After translating the questionnaire
from English to Finnish and pre-testing it, it was dis-
tributed by ?rst class mail to the targeted partici-
pants in the late spring of 2004. The initial
questionnaire request and three follow-ups yielded
173 (57.7%) anonymously ?lled questionnaires by
the deadline. Since most respondents used pre-coded
return envelopes, it seems that there are responses
from at least 98 (83%) business groups and from at
least 75 (87%) industrial corporations. The average
age of respondents was about 46 years with an aver-
age length of employment in their current position of
5.7 years. One hundred and forty four of the respon-
dents were men, and 19 women. According to the
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 157
responses, the business units had 26–9700 employees
and, on average, about 820 employees. The reported
sales of the business units ranged from 0.5 million to
6000 million euros, and averaged 241 million euros
(n = 152). Further information about the functional
area, educational background and industry category
of the respondents is shown in Table 1.
Four responses with multiple missing values on
dependent variables were excluded from hypothesis
testing to avoid any arti?cial increase in relation-
ships with independent variables, as recommended
in Hair, Black, Babin, Anderson, and Tatham
(2006, p. 56). For the remaining sample of 169 cases,
the level of randomness was tested with Little’s
missing completely at random (MCAR) test and
the result was found acceptable (v
2
505,830, DF
577, and Sig. 0.985). Therefore, given that the level
of missing data was acceptably low and missing
data did not seem to occur in a speci?c non-random
fashion, in a small number of other cases, individual
missing values were replaced with mean values. This
resulted in 169 responses (56.3%) deemed suitable
for hypothesis testing. To test if the timing of
responding provided bias in results, tests of one-
way analysis of variance in construct scores across
the timing of received questionnaires (i.e., after the
?rst, second, third or fourth contact) were under-
taken. No statistically signi?cant di?erences were
found at the 5% level.
Measures
The survey questionnaire was designed to collect
information on the integration of information sys-
tems, the degree to which the budgeting processes
re?ected the four design principles of enabling con-
trol, perceived systems success, and performance. In
addressing management control systems we chose to
focus our questionnaire on the budgeting process in
particular since this has been identi?ed as one of the
most signi?cant, widespread and long standing
aspects of organisational management control sys-
tems (e.g. Covaleski, Evans, Luft, & Shields, 2003;
Hansen, Otley, & Stede, 2003). This general ?nding
is con?rmed in our speci?c research setting since
well over 90% of Finnish companies (with over
16.7 million euro turnover) have been found to have
an annual budget (Ekholm & Wallin, 2000). Whilst
recognising a limitation in our research of focussing
on only budgeting, our focus does overcome the
challenges of trying to understand and integrate
results from questions addressing a wider range of
likely di?erentially and far less frequently adopted
aspects of MCS.
New measures were developed for several of
these variables, in particular addressing the enabling
characteristics of the budgetary process. As a ?rst
step, two rounds of interviews were conducted in a
sample of nine managers of a highly successful busi-
ness unit in Helsinki to explore possible impacts of
ISI on enabling budgetary control. The researchers
then iteratively developed a large set of questions
that addressed the four design principles. These
were systematically re?ned and questions that
remained ambiguous or poorly related to the con-
struct were dropped. The validity of measures was
enhanced by textbook de?nitions, and extensive
pilot testing in a sample of 10 professionals (six
business managers and four academic scholars) to
gain their views on the clarity and focus of the ques-
tions.
3
Finally, after checking the factorability of
items,
4
the design of all new measurement instru-
ments was based on the results of principal compo-
Table 1
Respondents by functional area, education, and industry
(n = 169)
n %
Panel A: Functional area
Accounting and ?nance 64 38
Production 41 24
Research, development, etc., 38 22
Other areas 26 15
Panel B: Educational background
Academic degree 116
Community college degree 45
Advanced degree 16
Panel C: Industry category
Metal 37 21.9
Foodstu?s and beverages 33 19.5
Multiple industries 30 17.8
Electronics 23 13.6
Chemicals and plastics 22 13.0
Forest 17 10.1
Textile 2 1.2
Energy 2 1.2
Furniture 2 1.2
Construction materials 1 0.6
3
The questions were translated from English to Finnish by the
second author, who is ?uent in both languages. In addition, the
translation of the survey items was independently double-
checked.
4
The Bartlett test of sphericity showed that nonzero correla-
tions existed at the signi?cance level of 0.000 for all the variables.
The Kaiser–Meyer–Olkin measure of sampling adequacy was met
in all cases with an MSA of >0.542.
158 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
nents analysis with oblique Oblimin rotations,
5
and
Cronbach (1951) alpha statistics of internal reliabil-
ity (Nunnally, 1978). Table 2 contains an overview
of the wording of items in the ?nal analysis (the full
questionnaire is appendicised) together with the
results of the factor and reliability analysis. Descrip-
tive statistics, based on the average scores of multi-
item variables, are presented in Table 3.
Information system integration
To measure information system integration the
questionnaire contained four questions using a
seven-point Likert-type scale ranging from disagree
completely (1) to agree completely (7). A two com-
ponent solution was discovered. The ?rst compo-
nent is used here since it obtained a Cronbach
alpha of 0.68 that was judged acceptable (Nunnally,
1978).
6
It includes the following items: information
in reports produced by our information systems is
entirely based on common sources of data (e.g. a
common database), and we have fully-integrated
information systems that contain both ?nancial
and non-?nancial information. This component
explains 75.3% of the variation. It is called informa-
tion system integration (ISI).
The budgeting processes and management
Twenty items were initially designed to measure
budgeting processes, four questions for each of the
design characteristics of enabling control. Since
Ahrens and Chapman (2004) discuss the potential
for the enabling design characteristics to extend
our understanding beyond studies that have
explored the role of interactive control systems we
also included four questions addressing the possibil-
ity of interactive budgetary control as developed by
Abernethy and Brownell (1999) to explore this issue.
The de?nition of budgeting processes was provided
(i.e., the whole range of activities – including meet-
ings – relating to the preparation, monitoring and
updating of formal business unit quantitative plans
within the business unit). Likert-type scale ranged
from disagree completely (1) to agree completely
(7). As Table 2 shows, the results of the principal
components analysis reveal four factors whose
eigenvalues are greater than one. In all, they explain
about 64.8% of the variance. The items for interac-
tive control did not load on a separate factor, rather
one of them loaded on global transparency and
another on ?exibility as an item suggesting that
interactive control is related, but may not be a
broad enough concept when set in the context of a
wider analysis of budgetary process.
Success of budgetary control and information sys-
tems was assessed by four questions about whether
the overall bene?ts outweigh the costs and whether
they are right tools for managing the business unit,
and not simply whether the systems might be felt to
be a success on other terms (technical sophistica-
tion, etc.). The principal component factor analysis
yielded one factor, with an eigenvalue greater than
one. It explained 57.0% of the variance, and
obtained a reliability estimate of 0.74.
Performance was measured applying an instru-
ment developed by Govindarajan (1984) and
Govindarajan and Fisher (1990). Following Chen-
hall and Lang?eld-Smith (1998), respondents were
asked to rate the performance of their business unit
on ten dimensions relative to competitors during the
past year. A seven-point scale ranging from ‘‘unsat-
isfactory” to ‘‘excellent” was used. However, rather
than making an attempt to measure a weighted
overall organisational performance index, we con-
ducted factor analysis to analyze various dimen-
sions of performance in greater detail.
Principal components factor analysis results
identi?ed three components with eigenvalues over
one for the items measuring perceived performance
relative to competitors. The ?rst component can be
interpreted as market performance. It consists of
development of new products, sales volume, market
share and market development, and explains about
44.42% of total variance. The second component is
called ?nancial performance. It consists of return on
investment, pro?t, and cash ?ow from operations,
explaining 19.12% of variance. The third compo-
nent is labelled as social responsibility. It includes
personnel developments and political-public a?airs.
It explains 11.14% of variance. Financial perfor-
mance, market performance, and social responsibil-
ity were used in the subsequent data analysis to aid
in a more detailed analysis of non-?nancial and
5
According to Hair et al. (2006, p. 125), the oblique rotation
method is more ?exible and more realistic, because the theoret-
ically important underlying dimensions are not assumed to be
uncorrelated with each other. However, Varimax rotations were
also checked and found to provide identical components. Given
the sample size, an objective of obtaining a power level of 80%
and the use of a 0.05 signi?cance level, factor loadings were
expected to be around 0.45 at a minimum in the sample of 169 to
be considered signi?cant (Hair et al., 2006). Raw data rather than
factor scores were used in the PLS analysis however.
6
The second component was dropped from the analysis due to
too low reliability coe?cient.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 159
Table 2
Principal components analysis (Oblimin rotated) and Cronbach alphas (n = 169)
Components
I II III IV
A. Information system integration (a 0.68, Eigenvalue = 1.51, 75.27% of variance)
5. Information in reports produced by our information systems is 0.868
entirely based on common sources of data (e.g. a common database)
7. We have a fully-integrated information systems that contain both ?nancial and non-?nancial
information
0.868
B. Budgeting and management
Repair (a 0.67, Eigenvalue = 1.22, 8.74% of variance)
2. It is easy for me to modify budget information and reports (or have them modi?ed for me) 0.086 0.045 0.129 À0.793
5. I easily get access to very detailed information in order to investigate budget deviations 0.124 À0.001 À0.035 À0.828
Internal transparency (a 0.87, Eigenvalue = 4.901, 35.05% of variance)
6. The budgeting process helps to clarify the activities that make up my business unit 0.802 À0.066 À0.006 À0.054
7. The budgeting process increases my knowledge of the operations of my business unit 0.902 0.082 À0.080 À0.030
8. The budget process increases my understanding of what drives our revenue/cost levels 0.760 À0.061 0.049 À0.163
9. The budget process increases my knowledge of how my business unit works as a whole 0.767 0.038 0.187 0.007
Global transparency (a 0.81) (Eigenvalue = 1.325, 9.47% of variance)
10. The budgeting process helps to communicate business unit strategy 0.256 0.062 0.620 0.098
11. The budgeting process signals areas in which we may need to change business unit strategy 0.052 0.161 0.793 0.083
12. The budgeting process helps personnel in my business unit to understand the overall context in
which they are working
0.304 0.084 0.565 0.027
13. I analyse budget information in order to come up with ideas for improving operations under
my control
À0.008 À0.176 0.701 À0.065
16. I often think of new ways of doing things during the budgeting process À0.220 0.023 0.744 À0.205
19. Our budgeting process aims to generate regular and frequent ?ows of strategic information
between operational and senior management
0.141 À0.183 0.600 0.025
Flexibility (a 0.51, Eigenvalue = 1.623, 11.60% of variance)
14. I can only make expenditures that have been built into the budget
a
0.070 0.841 0.099 0.141
20. Discussion during the budgeting process focuses on ensuring strict À0.084 0.766 À0.135 À0.232
adherence to original assumptions and action plans
a
C. Success of budgeting process and information systems (a 0.75, Eigenvalue = 2.28, 57.02% of variance)
1. Overall the bene?ts of our budgeting process outweigh the costs 0.731
2. Overall the bene?ts of the information systems that we use in our budgeting process outweigh the
costs
0.766
3. I am convinced that our budgeting process is the right tool for managing this business unit 0.821
4. I am convinced that the information systems that we use in our budgeting process are the right
tools for managing this business unit
0.697
D. Business unit performance relative to competitors
Financial performance (a 0.92, Eigenvalue 1.72, 19.12% of variance)
1. Return on investment 0.057 À0.923 À0.024
2. Pro?t 0.096 À0.911 À0.034
3. Cash ?ow from operations À0.078 À0.905 0.032
Market performance (a 0.80, Eigenvalue 4.00, 44.32% of variance)
5. Development of new products 0.764 0.196 À0.028
6. Sales volume 0.769 À0.139 0.013
7. Market share 0.772 À0.178 0.029
8. Market developments 0.702 À0.178 0.147
Social responsibility (a 0.67, Eigenvalue 1.00, 11.14% of variance)
9. Personnel developments 0.122 À0.039 0.793
10. Political-public a?airs À0.077 0.061 0.920
a
Reverse coded question.
160 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
?nancial performance e?ects relative to peers, when
there is lack of archival performance data as sug-
gested in Kihn (2005).
The descriptive statistics in Table 3 show that on
average, the extent of information system integra-
tion is average (with a mean of 4.02), but the actual
range is large. Of the four components of enabling
budgeting, internal transparency receives the highest
score (with a mean of 5.48). The mean scores of glo-
bal transparency and repair are also fairly high (i.e.,
4.70 and 4.67, respectively). Flexibility is the one
with the lowest mean score (of 4.56). The perceived
success of budgetary control and information sys-
tems is somewhat above average against the scale
(with a mean of 4.50). Of the three performance
components, ?nancial performance is perceived to
be highest (with a mean of 4.65), followed by market
performance (with a mean of 4.53), and by social
responsibility (with a mean of 4.47).
Pearson correlation coe?cients presented in
Table 4 indicate several associations between the
design characteristics of enabling control. Repair
is positively associated with all the other design
characteristics, and it has statistically signi?cant
relationships with both global transparency (0.280,
p < 0.000) and internal transparency (0.318,
p < 0.000). As could be expected, there is also a
strong positive correlation between internal and glo-
bal transparency (0.552, p < 0.000). Flexibility,
however, has close to zero and statistically insignif-
icant correlations with the other design
characteristics.
Survey data analysis
Greater use of structural equation modelling
(SEM) in management accounting research has
been called to overcome some of the limitations of
more traditional statistical techniques (Shields,
1997; Shields & Shields, 1998; Smith & Lang?eld-
Smith, 2004). It has been considered particularly
useful when one dependent variable becomes an
independent variable in subsequent dependence
relationships (Hair et al., 2006). In this study, the
hypotheses were tested with SEM analysis using
partial least squares (PLS) technique. There are a
Table 3
Descriptive statistics for variables (n = 169)
Variable Mean SD Theoretical range Actual range
1. IS integration 4.02 1.56 1.00–7.00 1.00–7.00
2. Repair 4.67 1.41 1.00–7.00 1.00–7.00
3. Internal transparency 5.48 1.01 1.00–7.00 2.00–7.00
4. Global transparency 4.70 1.00 1.00–7.00 1.33–7.00
5. Flexibility 4.56 1.25 1.00–7.00 1.00–7.00
6. Success 4.50 1.01 1.00–7.00 1.75–6.75
7. Financial performance 4.65 1.40 1.00–7.00 1.00–7.00
8. Market performance 4.53 1.04 1.00–7.00 2.00–7.00
9. Social responsibility 4.47 0.96 1.00–7.00 1.50–7.00
Table 4
Pearson correlation coe?cients (n = 169)
1 2 3 4 5 6 7 8 9
1. IS integration 1.000
2. Repair 0.197
**
1.000
3. Internal transparency 0.222
***
0.318
**
1.000
4. Global transparency 0.277
***
0.280
****
0.552
****
1.000
5. Budgeting: ?exibility À0.152
**
0.100 0.029 À0.063 1.000
6. Success 0.341
****
0.323
****
0.421
***
0.382
****
0.029 1.000
7. Financial performance 0.045 0.201
***
À0.065 À0.013 0.169 0.114 1.000
8. Market performance 0.216
**
0.238
***
0.126 0.193
**
0.016 0.279
****
0.446
****
1.000
9. Social responsibility 0.121 0.059 0.182
**
0.219
***
0.039 0.130 0.144
*
0.409
****
1.000
*
p < 0.10 (two-tailed).
**
p < 0.05.
***
p < 0.01.
****
p < 0.000.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 161
growing number of studies using PLS in the
accounting literature (see Hall, 2008 for a recent
example and list of other studies). PLS and covari-
ance-based SEM have di?erent strengths. PLS is
suitable for causal-predictive analysis of complex
relationships with multiple dependent variables,
when there is scarcity of prior theoretical knowl-
edge, and/or when the sample size is relatively small
(Chin & Newsted, 1999; Wold, 1980). It o?ers the
ability to perform path-analytic modelling with
latent variables. In our study, PLS was used to test
re?ective links between constructs and measures
(indicators) meaning that indicators are believed
to re?ect the unobserved, underlying construct, with
the construct giving rise to the observed measures
(Bisbe, Batista-Foguet, & Chenhall, 2007; Hulland,
1999).
The PLS approach comprises a measurement
model that speci?es relations between manifest items
(observed values for speci?c survey questions) and
the latent constructs that they represent (i.e., unob-
served values de?ned as the sum of their respective
indicators). It also comprises a structural model that
identi?es relations among constructs. To maximize
the interpretability of both models, the PLS model
was analyzed and interpreted sequentially in two
stages: the reliability and validity of the measure-
ment model was assessed ?rst, followed by the
assessment of the structural model (Hulland, 1999).
The preliminary analyses of uni-dimensionality
and reliability of multi-item constructs (i.e., the fac-
tor analysis and Cronbach alphas) were reported
and presented above (see Table 2). As Table 5
shows, the PLS analysis con?rmed the preliminary
factor analysis of the variables by showing high
composite reliability of latent variables (i.e., from
0.78 to 0.95) and high (over 0.70) loadings of man-
ifest variables on latent variables. A few items
loaded marginally below 0.70 in some of the tests
(i.e., three global transparency items being 0.693,
0.686 and 0.556; and a ?exibility item being
0.634). According to Hulland (1999), these loadings
can be considered acceptable for explorative
research.
Convergent validity of constructs was demon-
strated by calculating average variance extracted
(AVE). As the fourth column in Table 5 shows, all
variables were well above 0.50, which according to
Chenhall (2005, p. 411) is the conventional guideline
for adequate convergent validity. The discriminant
validity of the measurement model was also tested
using AVE, by comparing it with the squared corre-
lations between constructs. This determines the
extent to which a construct shares more variance
with its measures than it shares with other con-
structs (Fornell & Larcker, 1981). The AVE results
of Table 5 were compared with the correlations of
latent variables derived from Table 6. All AVE mea-
sures appeared greater than the respective squared
correlations suggesting satisfactory discriminant
validity.
7
Analysis of results
In the second phase of our PLS analysis, the ade-
quacy of the structural model was assessed with
standardised b-statistics, used as path coe?cients
and generated by the PLS, from ordinary least
squares regression (OLS). Fig. 1 shows the full
PLS model tested. Bootstrapping using 500 samples
with replacement was used to assess the signi?cance
of the path coe?cients. As recommended in the
PLS-graph user’s manual (Chin, 2001, p. 14), each
sample consisted of the same number of cases as
our original sample set (i.e., n = 169). Given that
the use of overall goodness-of-?t measures would
not be appropriate, the ?t of the PLS model was
estimated by the overall incidence of signi?cant rela-
tionships between constructs and the explained var-
iance of the dependent variables (Chenhall, 2005).
Tables 6 and 7 and Fig. 2 summarise the signi?cant
results numerically and graphically.
Test of hypotheses
As shown in Fig. 1, the research model tests the
extent to which ISI is associated with perceived sys-
tem success (H1) and performance, the extent to
which it is associated with design characteristics of
enabling approach to management control (H2a–
d), and how these design characteristics of an
enabling approach to management control are asso-
ciated with perceived system success (H3a–d) and
with performance (H4a–d).
H1: Information system integration is positively
related to perceived system success.
The PLS results in Table 7 indicate that the asso-
ciation between ISI and perceived success of systems
7
Note also that the correlation coe?cients presented in Table 4
do not indicate too high multicollinearity.
162 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
is positive and very strongly statistically signi?cant
(0.237, p < 0.005). This result provides support for
the ?rst hypotheses.
As expected, the results also indicate a lack of
signi?cant links between ISI and various perfor-
mance items. The associations between information
system integration and ?nancial performance
(0.015) and social responsibility (0.046) are close
to zero and insigni?cant. The association between
information system integration and market perfor-
mance is positive (0.115), but also statistically
insigni?cant.
H2: Information system integration is positively
related to an enabling approach to management
control as seen in its four design characteristics
of (a) repair, (b) local transparency, (c) global
transparency, and (d) ?exibility.
Table 5
Loadings of manifest variables in PLS analysis (n = 169)
Items Loadings Composite reliability Average variance extracted
A. IS integration
Item 5 0.886 0.858 0.752
Item 7 0.848
B. Budgeting and management
Repair
Item 2 0.889 0.858 0.752
Item 5 0.845
Internal transparency
Item 6 0.811 0.909 0.713
Item 7 0.875
Item 8 0.822
Item 9 0.868
Global transparency
Item 10 0.763 0.860 0.510
Item 11 0.765
Item 12 0.795
Item 13 0.686
Item 16 0.556
Item 19 0.693
Flexibility
Item 14 0.634 0.780 0.649
Item 20 0.946
C. Success of budgeting process and information systems
Item 1 0.713 0.839 0.567
Item 2 0.712
Item 3 0.842
Item 4 0.738
D. Business unit performance relative to competitors
Financial performance
Item 1 0.954 0.947 0.857
Item 2 0.950
Item 3 0.870
Market performance
Item 5 0.716 0.872 0.631
Item 6 0.818
Item 7 0.804
Item 8 0.835
Social responsibility
Item 9 0.914 0.852 0.744
Item 10 0.807
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 163
Consistent with expectations Table 7 shows sig-
ni?cant positive relationships between information
system integration and repair (0.197, p < 0.01),
internal transparency (0.219, p < 0.005), and global
transparency (0.234, p < 0.005). These results pro-
vide full support for Hypotheses 3a–c. In addition,
the results suggest a weak negative association
between information system integration and ?exible
use of budgets (À0.176) at 0.10 level of statistical
signi?cance (for a one-tailed test).
H3: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to perceived system success.
Table 8 shows how the various aspects of
enabling budgeting are associated with perceived
system success. Table 8 shows signi?cant positive
relationships between all aspects of enabling use of
budgets but ?exibility and perceived system success.
The perceived level of repair is signi?cantly associ-
ated with success (1.40, p < 0.05) suggesting that
when the level of repair as a budgeting characteristic
increases, then the perceived success of budgetary
control and information systems also increases.
The perceived extent of internal transparency is also
signi?cantly associated with success (0.208,
p < 0.025). Likewise, the level of global transpar-
ency is signi?cantly associated with success (0.216,
p < 0.01). In contrast, the association between ?exi-
bility in the use of budgets and success is close to
zero and not signi?cant in the statistical sense
(0.039, n.s.). These results suggest that most of the
investigated aspects of enabling use of budgets are
positively related to perceived system success. The
explained variance (R
2
) of this dependent variable
is 0.310. These results provide partial support for
Hypothesis 3.
H4: An enabling approach to management con-
trol as seen in its four design characteristics of
(a) repair, (b) internal transparency, (c) global
transparency, and (d) ?exibility, is positively
related to business unit performance.
The results reported in Table 8 suggest several
associations between characteristics of enabling
use of budgets and various aspects of business unit
performance. The extent of repair has positive asso-
ciations between all aspects of business unit perfor-
mance, except social responsibility. In particular, a
high extent of repair seems to increase ?nancial per-
formance (0.159, p < 0.05) and to a smaller degree
market performance (0.172, p < 0.10).
Contrary to expectations, the results reported in
Table 8 do not suggest signi?cant positive associa-
tions between the extent of internal transparency
and business unit performance. The associations
with market performance and social responsibility
Table 6
Correlations of latent variables from PLS (n = 169)
1 2 3 4 5 6 7 8 9
1. IS integration 1.000
2. Repair 0.197 1.000
3. Internal transparency 0.219 0.320 1.000
4. Global transparency 0.234 0.290 0.574 1.000
5. Flexibility À0.176 0.126 À0.012 À0.121 1.000
6. Success 0.354 0.321 0.428 0.427 À0.014 1.000
7. Financial performance À0.001 0.156 À0.075 0.030 0.184 0.034 1.000
8. Market performance 0.169 0.222 0.101 0.201 À0.004 0.241 0.422 1.000
9. Social responsibility 0.086 0.051 0.124 0.220 À0.005 0.145 0.158 0.417 1.000
Table 7
Results from PLS analysis (path coe?cients, n = 169)
Paths to Paths from
IS integration Mult. R
2
Repair 0.197
***
0.039
Internal transparency 0.219
*****
0.048
Global transparency 0.234
*****
0.055
Flexibility À0.176
*
0.031
Success 0.237
*****
Financial performance 0.015
Market performance 0.115
Social responsibility 0.046
*
p < 0.10 (one-tailed test).
**
p < 0.05.
***
p < 0.01.
****
p < 0.01.
*****
< 0.005.
164 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
are close to zero and statistically insigni?cant. The
relationship between internal transparency and
?nancial performance is statistically signi?cant,
but negative (À0.191, p < 0.05).
As expected, the level of global transparency
appears to have signi?cant positive associations
with several aspects of business unit performance.
It has a strong relationship with social responsibility
(0.225, p < 0.025), and a weaker association with
market performance (0.170, p < 0.10), but not with
?nancial performance (0.112, n.s.) in a statistical
sense.
The extent of ?exibility in the use of budgets
seems to be associated with ?nancial performance
(0.177, p < 0.05), but not with other analyzed
aspects of business unit performance. Overall, the
explained variance is 0.075 for ?nancial perfor-
mance, 0.055 for market performance, and 0.031
for social responsibility. Overall, these results pro-
vide partial support for the fourth and ?nal
hypotheses.
8
Fig. 2. Partial least squares model: IS integration, enabling use of budgeting and outcomes (n = 169). (Only statistically signi?cant paths
are shown).
Table 8
Results from PLS analysis (path coe?cients, n = 169)
Paths from Paths to
Success Financial performance Market performance Social responsibility
Repair 0.140
**
0.159
**
0.172
*
À0.025
Internal transparency 0.208
***
À0.191
**
À0.076 À0.007
Global transparency 0.216
****
0.112 0.170
*
0.225
***
Flexibility 0.039 0.177
**
0.014 0.034
Mult. R
2
0.310 0.075 0.055 0.031
*
p < 0.10 (one-tailed test).
**
p < 0.05.
***
p < 0.025.
****
p < 0.01.
8
In gathering a suitably large sample size we sent our
questionnaire to di?erent functional respondents within the same
?rm (we have 169 responses from 75 ?rms). This does raise the
possibility that non-independence of responses poses a threat to
the validity of our analysis. However, an additional ?rm-level
analysis was also conducted in which di?erent functional
respondents’ answers were averaged, and more stringent loading
requirements used resulting from the signi?cantly reduced sample
size. Whilst not identical, we feel that the qualitative similarities
in the constructs and their relationships in this second analysis
represent quite a strong test of the intuition behind our individual
hypotheses and of the analysis presented in full here.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 165
Discussion and conclusions
This study aimed to improve understanding of
how information system integration assists manag-
ers in enhancing performance. We analyzed direct
associations between these variables as well as
whether ISI is associated with the four design char-
acteristics of enabling control systems, and whether
these, in turn, are associated with performance. In
developing our hypotheses and statistical analysis
we explicitly sought to unpack the di?erence
between system success and performance as sug-
gested by Ittner et al. (2003), seeking to understand
implications for success, ?nancial and non-?nancial
aspects of business unit performance.
9
We measured system success in terms of whether
or not the system (both budgetary, and its support-
ing IS) was felt to be a cost-e?cient and an appro-
priate tool for managing the business unit. With
regards to performance, we used a factor analysis
to breakdown the overall measure of self-reported
business unit performance relative to competitors
to allow detailed analysis of the speci?c ways in
which enabling aspects of budgetary processes were
felt to contribute to ?nancial performance, market
performance and social responsibility. Such a
perceptions-based approach seemed useful in
increasing understanding of various ?nancial and
non-?nancial performance e?ects relative to com-
petitors, when archival performance data was not
available (Kihn, 2005).
As expected, our data shows a direct association
between information system integration and per-
ceived system success, but not between ISI and the
other aspects of performance. However, as
expected, positive and statistically signi?cant paths
are found between ISI and all the design principles
of enabling budgeting except ?exibility. A more
marginal negative e?ect is found between ISI and
?exibility suggesting di?culties in ?exible use of
budgets in the context of ISI. As noted in the
hypothesis development, ?exibility is an important
part of the enabling approach to control, but it is
a part of a set of design characteristics. Consistent
with prior research it seems that ?exibility might
need some supporting structure in order to lead to
positive outcomes, and so ?exibility on its own rep-
resents a necessary but not su?cient condition for
performance.
Our empirical results further show that it is the
enabling approach to budgeting that has direct
e?ects on both system success and performance.
Consistent with expectations we ?nd direct links
between all aspects of enabling budgetary control,
except ?exibility, and perceived system success.
Our results also indicate several associations
between enabling budgetary control and various
aspects of performance. Some of the statistically sig-
ni?cant e?ects between budgeting processes and
performance vary by performance dimensions, but
nearly all of them are positive as expected. The only
exception is internal transparency, which appears to
be negatively associated with ?nancial performance.
This result may re?ect greater use of budgets to try
to understand internal processes when ?nancial per-
formance is poor. That is, perhaps poor perfor-
mance leads to budget-based investigations which
lead to greater transparency, rather than greater
transparency leading to better ?nancial perfor-
mance. Strong positive associations between repair
and ?nancial performance receive support. In addi-
tion, a statistically signi?cant association between
repair and market performance is found. Global
transparency appears to be signi?cantly related to
social responsibility and to a lesser extent to market
performance.
The R
2
values of links to perceived system success
appear stronger than those with the other dimen-
sions of business unit performance (cf. Ittner
et al., 2003). Notwithstanding our e?orts to direct
our questions away from simply assessing satisfac-
tion with the system in its own terms, but the system
as a means of managing business unit performance,
some element of this kind of thinking would seem to
underlie the very strong relationship between the
extent of ISI and success, but not necessarily busi-
ness unit performance. Hence, our results also sug-
gest that management can be highly satis?ed with
their systems even if the systems do not directly
9
Our theoretical model (Fig. 1) did not include a relationship
between perceived success and performance. It was not expected.
However, following a query from a reviewer we conducted
additional tests to analyze this relationship in our data. The four
additional tests were otherwise similar, but di?ered in whether
system success or the three performance items was/were inter-
vening variables, and whether enabling budgeting items were
directly linked to both set of outcome variables, or to just one of
them. The main result was that as past studies have suggested,
our data showed no statistically signi?cant relationships between
perceived system success and ?nancial performance. Some minor
results were that the link between system success and market
performance appeared statistically signi?cant in all four cases,
and the link between system success and social responsibility
appeared signi?cant in one case. It would be interesting to try and
understand this relationship in more detail in future studies.
166 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
enhance performance in ?nancial and other terms.
Taken together, we did not get all the results we
had expected, but ?nd partial support for our
hypothesis. We draw some reassurance for the basis
for our theoretical argument from the robustness
check we carried out through a supplementary
?rm-level analysis. Whilst the results were not iden-
tical, they con?rmed the main premise concerning
direct and indirect e?ects, and also demonstrated
positive links between the enabling design charac-
teristics and various aspects of business unit
performance.
In this study, we did not aim to analyze the over-
all concept of enabling budgeting. However, note
that the empirical results of our study suggest rela-
tively high levels of design characteristics of
enabling control, and as could be expected, there
were several associations between these variables.
Repair was positively associated with all the other
design principles, and internal and global transpar-
ency were positively and signi?cantly correlated
with each other. Flexibility, however, was not signif-
icantly correlated with the other variables. Hope-
fully future research will further develop our
understanding of the complex role of ?exibility in
driving performance.
This exploratory study has several limitations
that should be considered when drawing conclu-
sions from the results. First, while prior research
has mostly analyzed ERPs, it is noteworthy that
we have contributed to the literature by analyzing
the extent of information system integration directly
through an attempt to measure the degree to which
information is drawn and integrated from a com-
mon database. More speci?c impacts of various
types of ERPs could be analyzed in future research.
Second, our analysis has focused on the e?ects of
ISI on enabling use of budgets. Possible e?ects of
other antecedent variables on enabling use of bud-
gets could also be analyzed. Likewise, enabling use
of other aspects of MCS could be explored. Our
choice to focus on budgets as a commonly used
form of MCS limits our wider understanding.
Third, some of our research instruments, in particu-
lar ?exibility, could be still further tested and devel-
oped. Fourth, the survey data report managers’
perceptions and, therefore, may not accurately rep-
resent actual business practice. Finally, the observed
e?ects do not indicate causal relationships, rather
statistical associations consistent with the theory.
Notwithstanding these various limitations we feel
that this study contributes several valuable insights
into the nature and ?nancial and non-?nancial
e?ects of ISI and budgetary control. Galbraith
(1973) argues that one way to reduce organisational
uncertainty is to invest in information processing
capabilities. Today, corporations often attempt this
through extensive investments in both information
systems, and management control systems such as
budgeting. In this study we analyzed the extent to
which ISI and enabling approaches to budgeting
go hand in hand, and, the extent to which they
impact business unit performance and success.
Consistent with past ?eld studies of technology
and control our analysis suggests that in considering
investments in ISI and MCS, managers would be
wise to explicitly include a detailed analysis of con-
trol as both a technical and social phenomenon. In
terms of pursuing a process view of budgetary con-
trol, this study demonstrates that the four design
principles of enabling control o?er a fruitful way
of systematically unpacking this challenge.
Acknowledgements
The authors thank the Academy of Finland and
the second author also the Foundation of Economic
Education for ?nancial support. We would also like
to thank participants at the American Accounting
Association Management Accounting Section Mid-
year Meeting, January 2006, Clearwater Beach,
Florida. In particular we would like to thank Anne
Lillis, Habib Mahama and Sally Widener for their
valuable comments, and Pekka Taipale for research
assistance. Finally, we also gratefully acknowledge
the many helpful comments and questions of two
anonymous reviewers.
References
Abernethy, M. A., & Brownell, P. (1999). The role of budgets in
organizations facing strategic change: An exploratory study.
Accounting, Organizations and Society, 24(3), 198–204.
Adler, P., & Borys, B. (1996). Two types of bureaucracy:
Enabling and coercive. Administrative Science Quarterly,
41(1), 61–90.
Ahrens, T., & Chapman, C. S. (2004). Accounting for ?exibility
and e?ciency: A ?eld study of management control systems in
a restaurant chain. Contemporary Accounting Research, 21(2),
271–301.
Barney, J., Wright, M., & Ketchen, D. J. (2001). The resource-
based view of the ?rm: Ten years after 1991. Journal of
Management, 27, 625–641.
Bashein, B., Markus, L., & Finley, J. (1997). Safety nets: Secrets
of e?ective information technology controls. Financial Execu-
tives Research Foundation.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 167
Bisbe, J., Batista-Foguet, J.-M., & Chenhall, R. (2007). De?ning
management accounting constructs: A methodological note
on the risks of conceptual misspeci?cation. Accounting,
Organizations and Society, 32(7–8), 789–820.
Blaine, J. (1999). Special edition using SAP R/3 (3rd ed.).
Caglio, A. (2003). Enterprise resource planning systems and
accountants: Towards hybridization?. European Accounting
Review 12(1), 123–154.
Carr, N. G. (2003). IT doesn’t matter. Harvard Business Review,
81(5), 41–49.
Chapman, C. S. (1998). Accountants in organizational networks.
Accounting, Organizations and Society, 23(8), 737–766.
Chapman, C. S. (2005). Not because they are new: Developing
the contribution of enterprise resource planning systems to
management control research. Accounting, Organizations and
Society, 30(7–8), 685–689.
Chapman, C. S., & Chua, W. F. (2003). Technology-driven
integration, automation and standardisation of business
processes: Implications for accounting. In A. Bhimani (Ed.),
Management accounting in the digital economy (pp. 74–94).
Oxford: Oxford University Press.
Chenhall, R. H. (2005). Integrative strategic performance mea-
surement systems: Strategy, strategic alignment of manufac-
turing, learning and organizational performance. Accounting,
Organizations and Society, 30(5), 395–422.
Chenhall, R. H., & Lang?eld-Smith, K. (1998). The relationship
between strategic priorities, management techniques and
management accounting: An empirical investigation using a
systems approach. Accounting, Organizations and Society,
23(3), 243–264.
Chenhall, R. H., & Morris, D. (1995). Organic decision and
communication processes and management accounting sys-
tems in entrepreneurial and conservative business organiza-
tions. Omega, International Journal of Management Science,
23(5), 485–497.
Chin, W. W. (2001). PLS graph user’s guide (Version 3.0.
February).
Chin, W. W., & Newsted, P. R. (1999). Structural equation
modelling analysis with small samples using partial least
squares. In R. H. Hoyle (Ed.), Statistical strategies for small
sample research. Thousand Oaks, CA: SAGE Publications.
Ciborra, C. U., & Lanzara, G. F. (1994). Formative contexts and
information technology. Accounting, Management and Infor-
mation Technologies, 4, 611–626.
Covaleski, M. A., Evans, H., III, Luft, J., & Shields, M. (2003).
Budgeting research: Three theoretical perspectives and criteria
for selective integration. Journal of Management Accounting
Research, 15, 3–49.
Cronbach, L. J. (1951). Coe?cient alpha and the internal
structure of tests. Psychometrika, 16, 297–334.
Davenport, T. (1998). Putting the enterprise into the enterprise
system. Harvard Business Review (July–August), 121–131.
Davenport, T. (2006). Competing on analytics. Harvard Business
Review (January), 99–107.
Dechow, N., & Mouritsen, J. (2005). On enterprise resource
planning systems, the quest for integration and management
control. Accounting, Organizations and Society, 30(7–8),
691–733.
Dillman, D. A. (2000). Mail and internet surveys: The tailored
design method. New York: Wiley.
Ekholm, B. G., & Wallin, J. (2000). Is the annual budget really
dead?. The European Accounting Review 9(4), 519–539.
Elmasri, R., & Navathe, S. B. (1989). Fundamentals of database
systems. Redwood, CA: Benjamin.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural
equation models with unobservable variables and measure-
ment error. Journal of Marketing Research, 18, 39–50.
Galbraith, J. R. (1973). Designing complex organizations. New
York, NY: Addison-Wesley.
Govindarajan, V. (1984). Appropriateness of accounting data in
performance evaluation: An empirical examination of envi-
ronmental uncertainty as an intervening variable. Accounting,
Organizations and Society, 9(2), 125–135.
Govindarajan, V., & Fisher, J. (1990). Strategy, control systems
and resource sharing: E?ects on business unit performance.
Academy of Management Journal, 33, 259–285.
Granlund, M., & Malmi, T. (2002). Moderate impact of ERPS on
management accounting: A lag or permanent outcome?.
Management Accounting Research 13(3), 299–321.
Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., &
Tatham, R. L. (2006). Multivariate data analysis (2nd ed.).
New Jersey: Pearson Prentice Hall.
Hall, M. (2008). The e?ect of comprehensive performance
measurement systems on role clarity, psychological empow-
erment and managerial performance. Accounting, Organiza-
tions and Society, 33(2–3), 141–163.
Hansen, S., Otley, D., & Stede, W. A. V. D. (2003). Practice
developments in budgeting: An overview and research
perspective. Journal of Management Accounting Research,
15, 49–70.
Hulland, J. (1999). Use of partial least squares (PLS) in strategic
management research: A review of four recent studies.
Strategic Management Journal, 20, 195–204.
Hunton, J., Lippincott, B., & Reck, J. (2003). Enterprise resource
planning systems: Comparing performance of adopters and
non-adopters. International Journal of Accounting Information
Systems, 4, 165–184.
Ittner, C., Larcker, D., & Randall, T. (2003). Performance
implications of strategic performance measurement in ?nan-
cial services ?rms. Accounting, Organizations and Society,
28(7/8), 715–741.
Kallinikos, J. (2004). Farewell to constructivism: Technology and
context-embedded action. In C. Avergrou, C. Ciborra, & F.
Land (Eds.), Social studies of information systems and
communication technologies (pp. 140–161). Oxford: Oxford
University Press.
Kallinikos, J. (2005). The order of technology: Complexity and
control in a connected world. Information and Organization,
15(3), 185–202.
Kihn, L. A. (2005). Comparing performance measurement
approaches in accounting research. Finnish Journal of Busi-
ness Economics, 5(2), 143–184.
Knights, D., & Vurdubakis, T. (2005). Information technology as
organization/disorganization. Information and Organization,
15(3), 181–184.
Kraut, R., Steinfeld, C., Chan, A., Butler, B., & Hoag, A. (1999).
Coordination and virtualisation: The role of electronic
networks and personal relationships. Organization Science,
10(6), 722–740.
Lee, Z., & Lee, J. (2000). An ERP implementation case study
from a knowledge transfer perspective. Journal of Information
Technology, 15(281–288).
Nelson, R., Todd, P., & Wixom, B. (2005). Antecedents of
information and system quality: An empirical examination
168 C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169
within the context of data warehousing. Journal of Manage-
ment Information Systems, 21(4), 199–235.
Newell, S., Huang, J., Galliers, R., & Pan, S. (2003). Implement-
ing enterprise resource planning and knowledge management
systems in tandem: Fostering e?ciency and innovation in
complementarity. Information and Organization, 13(1), 25–52.
Nunnally, J. C. (1978). Pscychometric theory (2nd ed.). New
York: McGraw-Hill.
Orlikowski, W. (1991). Integrated information environment or
matrix of control? The contradictory implications of infor-
mation technology. Accounting, Management and Information
Technology, 1(1), 9–42.
Poston, R., & Grabski, S. (2001). Financial impacts of enterprise
resource planning implementations. International Journal of
Accounting Information Systems, 2, 271–294.
Quattrone, P., & Hopper, T. (2001). What does organizational
change mean? Speculations on a taken for granted category.
Management Accounting Research, 12, 403–435.
Quattrone, P., & Hopper, T. (2005). A time–space odyssey:
Management control systems in two multinational organisa-
tions. Accounting, Organizations and Society, 30(7–8), 735–764.
Scapens, R., Jazayeri, M., & Scapens, J. (1998). SAP integrated
information systems and the implications for management
accountants. Management Accounting (September).
Shields, M. (1997). Research in management accounting by
North Americans in the 1990s. Journal of Management
Accounting Research, 9, 1–61.
Shields, J., & Shields, M. (1998). Antecedents of participative
budgeting. Accounting, Organizations and Society, 23(1),
49–76.
Simons, R. (1990). The role of management control systems in
creating competitive advantage: New perspectives. Account-
ing, Organizations and Society, 15(1/2), 127–143.
Smith, D., & Lang?eld-Smith, K. (2004). Structural equation
modelling in management accounting research: Critical anal-
ysis and opportunities. Journal of Accounting Literature, 23,
49–86.
Sotto, R. (1997). The virtual organization. Accounting, Manage-
ment and Information Technology, 7(1), 37–51.
Wold, H. (1980). Model construction and evaluation when
theoretical knowledge is scarce. In J. Kmenta & J. B. Ramsey
(Eds.), Evaluation of econometric models. New York: Aca-
demic Press.
Zubo?, S. (1988). In the age of the smart machine: The future of
work and power. New York: Basic books.
C.S. Chapman, L.-A. Kihn / Accounting, Organizations and Society 34 (2009) 151–169 169
doc_885623804.pdf