Description
Drawing on qualitative data collected during semi-structured interviews with 36profit centre managers in manufacturing
firms in Victoria, Australia, this study seeks to explore the mechanisms used to manage multiple manufacturing
performance dimensions arising from the pursuit of profit centre strategy. Where measures capture
potentially conflicting influences on the manufacturing cost function, strategy implementation is facilitated by loosening
control reactions to cost variances and through explicit attempts to integrate multiple measures.
Managing multiple dimensions of manufacturing
performance — an exploratory study
Anne M. Lillis*
Department of Accounting, University of Melbourne, Parkville 3010, Victoria, Australia
Abstract
Drawing on qualitative data collected during semi-structured interviews with 36 pro?t centre managers in manu-
facturing ?rms in Victoria, Australia, this study seeks to explore the mechanisms used to manage multiple manu-
facturing performance dimensions arising from the pursuit of pro?t centre strategy. Where measures capture
potentially con?icting in?uences on the manufacturing cost function, strategy implementation is facilitated by loosen-
ing control reactions to cost variances and through explicit attempts to integrate multiple measures. However, a joint
emphasis on performance dimensions relating to manufacturing e?ciency and customer responsiveness emerges as
problematic. In contrast, a joint emphasis on quality and e?ciency is relatively easily managed. It is suggested that in
the context of responsiveness strategies, the di?culty of designing complete measures inhibits the e?ectiveness of per-
formance measurement systems as a facilitator of strategy implementation. # 2002 Elsevier Science Ltd. All rights
reserved.
1. Introduction
The importance of designing performance
measurement systems that capture a range of
strategically important criteria in ?nancial and
non-?nancial terms is well established in the lit-
erature. Dominant themes in the performance
measurement literature relate to integration
(Nanni, Dixon, & Vollman, 1992), coherence (De
Haas & Kleingeld, 1999), the notion of a balanced
scorecard (Kaplan & Norton, 1992, 1996a, 2001),
and de?ning the performance variables that
represent important dimensions of a given strategy
(Simons, 1995). In all cases the message is that
e?ective performance measures must be able to
assess the ?rm’s progress on strategic initiatives
(Ittner & Larcker, 2001; Lang?eld-Smith, 1997;
Moon & Fitzgerald, 1996).
The literature that deals with the e?ective design
of performance measurement systems (PMSs)
incorporating multiple ?nancial and non-?nancial
measures emphasizes many attributes. These
include causal connections with strategy (Kaplan
& Norton, 1996a) integrating actions across func-
tional boundaries (De Haas & Kleingeld, 1999;
Nanni et al., 1992) and supporting critical strate-
gic measures with e?ective target setting and
reward systems (Moon & Fitzgerald, 1996; Otley,
1999). It has, however, been suggested that many
issues relating to implementation of multiple per-
formance measures remain unresolved (Hemmer,
1996) and that ‘‘research is needed on the treat-
ment of the inevitable trade-o?s that managers
0361-3682/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PI I : S0361- 3682( 01) 00032- 0
Accounting, Organizations and Society 27 (2002) 497–529
www.elsevier.com/locate/aos
* Tel.: +61-3-8344-5351; fax: +61-3-9349-2397.
E-mail address: [email protected] (A.M. Lillis).
will need to make among various ?nancial and
non-?nancial performance measures’’ (Ittner &
Larcker, 1998: 229).
This study focuses on the implementation issues
that arise when pro?t centre PMSs are dis-
aggregated to the functional subunit level. More
speci?cally, this study is designed to explore the
performance management challenges involved
when the implementation of pro?t centre strategy
requires manufacturing subunits to perform on
multiple and potentially con?icting dimensions.
Relevant performance dimensions at the manu-
facturing cost centre level may include, for exam-
ple, e?ciency, productivity, quality, response time
and delivery performance. While all are broadly
reconcilable at a pro?t centre level, these demands
are potentially con?icting performance parameters
at a functional subunit level. This paper reports
the ?ndings of an exploratory study into two
issues related to performance management of
manufacturing subunits:
1. the in?uence of pro?t centre strategy on the
formulation of PMSs for manufacturing
subunits; and
2. identi?cation of the ways in which multiple
manufacturing performance dimensions are
managed to facilitate the implementation of
pro?t centre strategy.
This study focuses on measures of manufactur-
ing cost centre performance utilized by pro?t cen-
tre managers as a key mechanism of performance
management in manufacturing ?rms. The perfor-
mance measures that are relied upon by pro?t
centre managers in evaluating manufacturing cost
centre performance signal strategic priorities and
performance expectations that in?uence cost cen-
tre decisions. The cost centre setting is invariably
one of high interdependencies. Strategies focused
on quality and responsiveness compound both
uncertainty and interdependence (Abernethy &
Lillis, 1995; Bowen, Siehl, & Schneider, 1989;
Fisher & Govindarajan, 1993; Nemetz & Fry,
1988; Van der Stede, 2000), and thus further
reduce the likelihood that complete subunit-level
performance measures can be constructed (Hirst,
1981). Pro?t centre managers are in a position to
discuss the rationale for the way multiple perfor-
mance dimensions are managed in such settings,
as well as any problems or unintended con-
sequences experienced. The ?ndings from this
study provide insights on issues of PMS imple-
mentation that do not feature in the pre-
dominantly prescriptive literature on e?ective
PMS design. This is a small sample study based on
36 case studies in a cross-section of ?rms. The data
are compiled and interpreted in the form of pat-
terns across cases. In turn, these patterns suggest
potential avenues for broader empirical research
to test their generalizability as well as raising
issues for theoretical consideration.
The remainder of the paper is organized as fol-
lows. The conceptual framework for this study is
outlined in Section 2. The research questions are
then identi?ed. This is followed by a description of
the study design and method, ?ndings and discus-
sion and some concluding comments.
2. Conceptual framework
2.1. Prior literature
In formulating PMS design characteristics to
e?ectively support strategy implementation, parti-
cular challenges have been apparent in the manu-
facturing industry. Rapid global change in the
bases of manufacturing competition has brought
new doctrines (e.g. JIT, TQM, and The Flexible
Factory), and required new approaches to perfor-
mance measurement (Abernethy & Lillis, 1995;
Banker, Potter, & Schroeder, 1993; Chenhall,
1997; Daniel & Rietsperger, 1991; Perera, Harri-
son, & Poole, 1997). The importance of quality,
?exibility and responsiveness has challenged the
relevance of conventional measures of manu-
facturing e?ciency (Abernethy & Lillis, 1995;
Dixon, Nanni, & Vollman, 1990; Kaplan, 1990;
Otley, 1994). One way in which this challenge has
been met is by PMS expansion. The literature has
established signi?cant associations between the
pursuit of speci?c strategies, such as TQM, JIT or
manufacturing ?exibility and the expansion of
traditional e?ciency-focused manufacturing
PMSs to embrace new manufacturing perfor-
mance measures (Abernethy & Lillis, 1995; Banker
498 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
et al., 1993; Chenhall, 1997; Daniel & Rietsperger,
1991; Perera et al., 1997). However, there have
been few positive performance outcomes asso-
ciated with such expansions (Ittner & Larcker,
1995; Perera et al., 1997; Young & Selto, 1993).
Ittner and Larcker (1998) suggest that the resul-
tant widespread use of multiple measures raises
several implementation issues including the like-
lihood of ine?ective spreading of managerial
e?ort.
The design of e?ective PMSs embracing multi-
ple measures has been considered in some depth as
the literature has sought to distinguish balanced,
integrated or coherent PMSs from a proliferation
of measures developed in an ad hoc way in
response to competitive pressures (Kaplan &
Norton, 1996b; Nanni et al., 1992; Norreklit,
2000; Simons, 1995). This literature emphasizes
the importance of explicitly embracing critical
dimensions of organizational strategy within
PMSs as well as the role of the PMS as a catalyst
for strategic dialogue throughout the organization
(Kaplan & Norton, 2001; Norreklit, 2000). While
much of the prescriptive performance measure-
ment literature is highly persuasive, implementa-
tion issues are beginning to feature in the
literature. Many of these issues are raised in the
context of the Balanced Scorecard because it is
arguably the most prescriptive and fully described
of contemporary frameworks. Several of these
challenges relate speci?cally to the multiplicity of
scorecard measures and their disaggregation.
Examples include the di?culty of attaining mutual
consistency among disaggregated measures of
interdependent processes (Sterman, Repenning, &
Kofman, 1997), the impact of cognitive limitations
on the processing of multiple performance mea-
sures (Lipe & Salterio, 2001) and the extent to
which the e?ectiveness of disaggregated PMSs
relies on a level of goal congruence and productive
strategic dialogue that may not be feasible (Nor-
reklit, 2000).
Kaplan and Norton (2001) address extensively
the disaggregation of corporate scorecards to the
subunit and individual level. While they consider
the potential for con?ict among measures of sub-
unit performance, they rely on the communication
strength of the balanced scorecard to generate
dialogue that will ultimately resolve di?erences in
perception and achieve goal congruence (Kaplan
& Norton, 2001; Norreklit, 2000). Related themes
in the literature refer to such con?ict as fruitful
dynamic tension (Simons, 1995) or an e?ective
signal of the need for strategic re-evaluation
(Kaplan & Norton, 1996b). Intraorganizational
con?ict has also been viewed as a productive e?ect
of the in?uence of uncontrollable factors on per-
formance measurement when there are signi?cant
interdependencies. In such settings, performance
metrics may result in productive cross-functional
pressure to improve shared outcomes (Merchant,
1987).
The presumption that con?ict induced by sub-
unit interdependencies and localized performance
measurement will be avoided or resolved through
dialogue, or alternatively that such con?ict will
represent an important signal may, however,
understate the direct in?uence of performance
measures on subunit performance (Otley, 1978;
Vagneur & Pieperl, 2000). The in?uence of speci?c
performance criteria relative to other less objective
in?uences on subunit performance is not clear,
and the preferences of evaluators in trading o?
performance measurement for dialogue remains
unexplored (Hartmann, 2000). There is, however,
general theoretical and empirical support for the
proposition that the presence of interdependencies
does not appear to limit reliance on subunit per-
formance measurement (Hartmann, 2000; MacIn-
tosh & Daft, 1987). This is evident despite the fact
that interdependency increases uncertainty
regarding subunit performance outcomes and
reduces controllability over those outcomes (Mer-
chant, 1987; Thomson, 1967). Furthermore, the
incompleteness of local measures may, in itself,
reduce the value of these measures as a catalyst for
productive dialogue. For example, the fact that
productivity and responsiveness have inherently
con?icting e?ects on a subunit cost function may
induce repetitive cross-functional dialogue to for-
mulate a ‘complete’ ?nancial picture of the trans-
action, but the productivity of such dialogue is
questionable (Dent, 1987), and there is always the
risk of unilateral subunit decision making without
engaging in cross-functional dialogue. The risk
associated with incomplete measures of perfor-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 499
mance is that they may ‘‘mobilise individual ener-
gies and motivations [resulting] in a transfer of
interest from the wider organizational purposes
which they were designed to serve onto the speci?c
behaviours which are necessary to improve the
indices of performance’’ (Hopwood, 1973: 16).
While the prior literature addressed this perfor-
mance management issue by focusing generally on
attributes of PMSs (such as controllability) and
in?uences on PMS design (such as inter-
dependencies and uncertainty), contemporary
performance management frameworks focus more
on the broader domain of strategic information
systems in which PMSs are just one element.
There are solutions evident in the literature that
rely less on con?ict avoidance and resolution
through dialogue. These include approaches to
con?ict avoidance that rely on changes to the per-
formance measures themselves. Examples include
the technical integration of multiple measures by
modifying internal measures to capture the impact
of multiple performance goals (McNair et al.,
1990), the use of explicit or implicit weightings on
measures to focus managerial e?ort (Ittner &
Larcker, 1998), or the tolerance of slack and the
less rigid use of budgetary controls (Van Der
Stede, 2000). Alternatively, Abernethy and Lillis
(1995) highlight the importance of integrative
structural forms such as cross-functional teams as
a means of avoiding the need to rely on PMSs to
promote co-ordination.
2.2. Summary
Fig. 1 illustrates the issue of concern in this
study. Issues related to the mapping of perfor-
mance measures to strategy at the pro?t centre
level are not considered in this study. Instead, it
focuses on the disaggregation of pro?t-centre level
performance measures to the functional subunit
level. Subunits X and Y have their own PMSs, but
they are highly interdependent. This would be the
case if, for example, X and Y are manufacturing
and sales departments and the pro?t centre is
pursuing a strategy of customer responsiveness. In
such a setting performance measurement in the
manufacturing cost centre may include customer-
focused measures of quality, response time and
delivery performance and business process or
?nancial measures of e?ciency and productivity.
1
At the pro?t centre level, these customer-focused
and business process measures are integrated
e?ectively as part of the causal chain linking
strategy and pro?t-focused ?nancial outcomes,
but e?ciency and customer responsiveness still
represent con?icting in?uences on the manu-
facturing cost function.
2
Even in the context of an
organizationally pro?table product customization,
disruption caused by customization is almost cer-
tain to reduce e?ciency and productivity at the
manufacturing cost centre level. It is also likely to
reduce overall performance on response time and
delivery as other customer orders are set aside to
accommodate the customization. The gains are
clearly evident elsewhere in the ?rm—primarily in
the sales subunit. The important point here is that
even expanded performance measures, at some
level of disaggregation, become internally-focused
and incomplete. Pro?t centre outcomes are a
function of the outcomes of both functional sub-
units independently and also of the way their
strategic interdependencies are managed. While
this may occur through speci?c mechanisms that
systematically manage multiple performance
dimensions throughout the pro?t centre, we have
little empirical evidence as to how these mechan-
isms work in practice.
We know empirically that multiple measures of
manufacturing subunit performance are common
and that the breadth of measures is increasing to
capture measures of e?ciency, productivity, qual-
1
While Balanced Scorecard terminology is used here, it is
not suggested that this problem is speci?c to Balanced Scor-
ecard implementations. It may arise whenever strategy set at a
higher organizational level is translated into disaggregated
multiple performance measures.
2
This is somewhat similar to the argument of Bushman,
Indjejikian, and Smith (1995) that where there is signi?cant
subunit interdependency, highly disaggregated, local perfor-
mance measures become less informative in the context of
compensation contracts. It is consistent also with the explana-
tions o?ered by Sterman et al. (1997) in their attempts to
explain the lack of mutual consistency among operational and
?nancial gains at Analog Devices. Sterman et al. (1997) argue
that much of the paradox is attributable to attempts to dis-
aggregate subunit improvement e?orts when processes are
inherently tightly coupled across the organization.
500 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
ity and responsiveness. We have normative
descriptions of the characteristics of balanced,
integrated and coherent PMSs that distinguish
these from ad hoc collections of multiple measures,
but there is no evidence that these contemporary
frameworks avoid the use of incomplete measures
at the subunit level. In this largely prescriptive lit-
erature, the mechanisms required to manage highly
disaggregated, incomplete measures of functional
subunit performance are not obvious. We know
little empirically of the features of PMSs at this
level, the nature of the con?icts created by dis-
aggregation, whether integration of performance
measures is sought and the way it is attempted. In
the absence of extensive empirical examination of
these phenomena, this study seeks to compile and
interpret the insights o?ered by a small sample of
pro?t centre managers into these issues.
3. Research questions
Using a cross-sectional case study research
design and qualitative data this study seeks to:
1. examine the in?uence of pro?t centre strat-
egy on the formulation of PMSs for manu-
facturing subunits; and
2. identify the ways multiple performance mea-
sures are managed to facilitate strategy
implementation.
Fig. 2 places the research questions within the
broad domain of Fig. 1. This study focuses ?rst on
identifying manufacturing subunit performance
measures used by pro?t centre managers against a
context of pro?t centre strategy. Second, it focuses
on the mechanisms used by pro?t centre managers
to implement pro?t centre strategy through man-
ufacturing subunits, given the in?uence of both
local performance measures and cross-functional
interdependencies.
4. Study design and method
Underlying the data collection and analysis
method for this study is the desire to explore ‘how’
and ‘why’ questions relating to management prac-
tices, their rationale, and expected outcomes (Yin,
1994). Rather than studying these phenomena in
depth within a single case, the aim is to link con-
textual and performance measurement character-
istics through the exploration of patterns among
cases. In order to observe patterns in the connec-
tions among these variables the data collected are
Fig. 1. Diagrammatic representation of the problem of disaggregation of performance measures when there are strategic inter-
dependencies.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 501
cross-sectional and qualitative. To explore issues
related to the management of multiple perfor-
mance dimensions this study draws on managerial
re?ections captured through semi-structured
interviews conducted in a small sample of pro?t
centres. This approach limits the depth of data
available on individual cases but provides greater
cross-sectional comparison than is typically avail-
able in ?eld research studies.
4.1. Description of the sample
Data were collected from 36 pro?t centre man-
agers in manufacturing ?rms in Victoria, Australia.
A general database of Victorian manufacturers
was used to select ?rms with at least 200 employ-
ees. Firms were selected across a range of indus-
tries with the aim of selecting participants with a
range of strategic orientations. Broad character-
istics of the sample and the managerial partici-
pants are given in Appendix A. ‘Firms’ in this
context may be fully independent or subunits of
other larger ?rms. They all, however, appeared as
separate entries in the Victorian Manufacturing
Directory. Given the size of most of the ?rms in
the sample, the general manager was the ?rst level
linkage for functional units. Thus, these managers
had the equivalent of ‘pro?t centre’ responsi-
bility.
3
While seven pro?t centres (independently
listed in the database) were drawn from two large
diversi?ed companies, each was from a separate
autonomous division. The sample did not include
Fig. 2. Placement of this study within the general domain of Fig. 1.
502 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
more than one pro?t centre from any one division
of a company. All of the remaining participants
were either independent ?rms or the only pro?t
centres included from the relevant company. The
response rate was 65% with 55 e?ective contacts
yielding 36 interviews with pro?t centre managers
and only 19 refusals.
4
With one exception, all
interviews were tape recorded and transcribed
verbatim. The sections of the interview guide rele-
vant to this paper are reproduced in Appendix B.
4.2. Data analysis
This paper reports ?ndings in the form of pat-
terns across cases. These patterns are a highly
summarized product of the researcher’s detailed
analysis of more extensive data. To limit the ana-
lytical bias inherent in the analysis of qualitative
data, a systematic, auditable process was used.
The systematic approach adopted here attempts to
provide an audit trail from interview transcript
through to theoretical proposition. In common
with other forms of data analyses, the process
involves data reduction or summarization, classi-
?cation and interpretation. Data were coded using
NUD
.
IST qualitative analysis software and then
analyzed using Miles and Huberman’s (1994) the-
matic conceptual matrix data displays. While the
?ndings reported in this paper are summarized in
a matrix format, the actual analytical matrices
used by the researcher were much more detailed
and frequently focused on examining patterns
between two variables rather than among many.
5
Throughout the analysis, an audit trail was main-
tained. As a result, all matrix cell entries are audi-
table back to transcripts through the use of
identifying case and text-unit numbers. Thus, the
link between data and ?ndings should be obser-
vable by others working from the same database.
The following section identi?es patterns in the
data. These patterns are then discussed in Section
6.
5. Findings—patterns in the data
5.1. The strategic pro?le of the sample
In order to explore the in?uence of pro?t centre
strategy on the implementation of multiple cost
centre performance measures, it was necessary ?rst
to identify the strategic pro?le of the sample.
Questions focused on the following attributes of
pro?t centre strategy and how these translate into
manufacturing strategy:
1. the pro?t centre’s competitive edge;
2. the nature of the markets within which the
pro?t centre operates;
3. the extent to which the pro?t centre attracts
price premiums consistent with di?erentia-
tion; and
Fig. 3. Classi?ed summary of strategic orientations.
3
In two cases it became apparent during an interview that
the general manager was not at an appropriate level, and in
both cases interviews were conducted subsequently in the same
?rms with managers at a lower, pro?t centre level. These latter
interviews are included. For simplicity, respondents are referred
to as ‘pro?t centre managers’.
4
From 66 letters sent to General Managers, 11 ?rms were
excluded without conducting an interview because they had
ceased manufacture in Australia, the General Manager was
located interstate or similar reason. Twelve managers refused to
participate, generally for reasons related to time pressure. In
seven cases, I gave up trying to make contact after approxi-
mately 10 phone calls. These seven cases have been classi?ed as
refusals.
5
These matrices are too detailed and numerous to include in
this paper. However, the patterns that emerged from that ana-
lysis ought to be evident when the data are reported in the full
case summary display format that is included in Appendices C,
D , E and referred to in the discussion of ?ndings.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 503
4. whether the pro?t centre faces demands for
product customization, and how these
demands are dealt with.
A summary of the strategic pro?le of the sample
is illustrated in Fig. 3.
6
Three strategic groupings
emerged from the analysis of these data. Two of
these groupings re?ect di?erentiation strategies
relating to either quality (based on stated com-
mitment to competing on quality, product design,
technical capability, product features, brand
names) or responsiveness (in terms of customer
service, product customization, and/or depend-
ability
7
). The third classi?cation relates to a stra-
tegic priority of cost leadership. Two important
features of the strategic pro?le of the sample
emerged from this analysis. First, there was the
prevalance of di?erentiation as a strategic priority.
There was an overwhelming emphasis on di?er-
entiation (29) with a very limited emphasis on low
cost as a strategic priority (three). Second, the
overlapping areas in Fig. 3 illustrate the presence
of widespread competition on multiple criteria
(23). The patterns observed here are consistent
with prior literature that has recognized the
decline in low cost mass production, the increasing
emphasis on product design innovation and cus-
tomer responsiveness as platforms of di?erentia-
tion, and the necessity for ?rms to compete on
multiple strategies (Belohlav, 1993; Drucker, 1990;
Hill, 1988; Jones & Butler, 1988; Nemetz & Fry,
1988; Otley, 1994).
5.2. The PMS pro?le of the sample
Fig. 4 gives an overview of the manufacturing
subunit PMS characteristics of the pro?t centres in
this study.
8
Performance measures used by pro?t
centre managers to measure manufacturing cost
centre performance are categorized into three
groups: cost and e?ciency measures, quality mea-
sures, and customer responsiveness measures. The
?ndings indicate that 34 of the pro?t centres use
?nancial and non-?nancial e?ciency and/or pro-
ductivity measures of manufacturing performance.
Of these, 21 use multiple performance measures
capturing performance on e?ciency, productivity,
quality and/or customer service. The measures
used to capture manufacturing performance
appear to be relatively generic.
9
The extent of
emphasis on e?ciency and productivity perfor-
mance measurement criteria in manufacturing
Fig. 4. Classi?ed summary of performance measures used to
measure manufacturing performance.
6
This diagram is based on data in Appendix C, which
summarizes and classi?es strategic orientations by case.
7
The ‘correct’ placement of ‘dependability’ is not clear.
Traditionally, dependability was associated with a defender-
type strategy (Miles & Snow, 1978). Frequently ‘reliability’,
which is akin to dependability refers to product reliaiblity and
is viewed as an element of a quality strategy. Dependability
may also refer to the ability to deliver on promises to custo-
mers, which is an element of a customer responsiveness or ser-
vice strategy. Its placement here with service and
responsiveness is based on the context in which the term was
used by the managers in this study. The term ‘dependability’ is
used in only four cases.
8
Fig. 4 is developed from the data in Appendix D. Appen-
dix D contains a listing of the measures identi?ed by each pro?t
centre manager as the measures relied upon in evaluating the
performance of the manufacturing subunit. In the right-hand
side column of Appendix D the measures used are classi?ed as
cost variance, non-?nancial e?ciency and productivity, exter-
nal quality and customer service measures.
9
While it is tempting to try to link Figs. 2 and 3 at the ?rm
level, such an analysis of the mapping of performance measures
to strategy is not the focus of this paper. It is evident at a glance
that the match between strategy and PMS composition is not
strong. It is obvious, for example, that the emphasis on cost
and non-?nancial e?ciency/productivity measures cannot be
matched with strategic orientations. Thus the PMS composi-
tion is best described as generic and somewhat inconsistent with
the overall view of strategic orientations.
504 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
suggests that an emphasis on low cost is almost
universally necessary, despite the fact that low cost
is a strategic priority in very few pro?t centres in
this sample. While this in itself is not surprising, it
creates a performance management challenge.
Manufacturing performance is measured based on
e?ciency and productivity criteria while at the
same time these subunits are focused on strategies
relating to responsiveness, service, dependability
or quality. The next section examines the chal-
lenges in managing multiple performance dimen-
sions that arise in this context.
5.3. Managing multiple performance dimensions
The qualitative data base for this study con-
tains, on a case-by-case basis, data relating to
patterns in a range of variables that combine to
‘tell the story’ of e?ective or problematic manage-
ment of the relations among multiple dimensions
of manufacturing performance. Given the focus of
this study on the management of multiple manu-
facturing performance dimensions, cases are clas-
si?ed as ‘e?ective’ in their implementation of
strategic priorities when pro?t centre managers
indicate that manufacturing subunits are suppor-
tive rather than resistant to the implementation of
pro?t centre strategy. While these cases do not
always suggest absolute compliance without dia-
logue, they indicate a level of acceptance within
manufacturing subunits of the need to perform on
multiple strategic dimensions. In contrast, cases
are classi?ed as ‘problematic’ in their imple-
mentation of strategic priorities when pro?t centre
managers indicate a perception that manufactur-
ing subunits inappropriately trade o? some per-
formance dimensions against others, engage in
non-productive con?ict, or resist expectations to
perform on multiple dimensions. Patterns in the
data are used to identify mechanisms that are used
by pro?t centre managers to facilitate strategy
implementation. It is possible to distinguish from
the data whether these facilitators result in
e?ective outcomes (i.e. are able to facilitate the
implementation of multiple strategic priorities)
or whether strategic implementation remains
problematic. These ?ndings are summarized in
Fig. 5.
10
Analysis of the data indicates that there are pri-
marily ?ve mechanisms used to facilitate the
implementation of multiple strategic priorities:
loosening of control reactions to variances, struc-
tural changes, innovative performance measure-
ment systems, and two speci?c adaptations to
PMSs to enable integration of multiple perfor-
mance measures, namely technical integration and
use of weightings. Fig. 5 contains several paths
representing a small number of occurrences in the
data. The paths with three or more occurrences
are bold for visibility. The following discussion
attempts to make sense of the patterns observed in
the data.
Fig. 5 tracks links between the use of facilitators
and pro?t centre managers’ perceptions of e?ec-
tive or problematic strategy implementation. In
the data, the managers’ perceptions are classi?ed
from both general and speci?c comments. More
general descriptions of the e?ective management
of all performance dimensions are captured in the
category ‘e?ective implementation of all strategic
priorities’. In other cases, descriptions of e?ective
outcomes are related to speci?c dimensions of
strategy. Thus the individual pairings of quality/
e?ciency, responsiveness/e?ciency and quality/
responsiveness are included to capture these ?ner
distinctions.
The pro?t centres that e?ectively manage the
demands for manufacturing e?ciency and pro-
ductivity with demands for quality and respon-
siveness do so by loosening the response to
e?ciency and productivity variances or by actively
employing other mechanisms that facilitate man-
agement of multiple performance dimensions.
Many (12) manage potential con?ict by taking a
10
Figs. 5 and 6 are developed from data in Appendix E,
which includes summary descriptions for strategy, PMS com-
position, reactions to e?ciency variances, mechanisms for
integrating multiple performance measures, and con?icts
experienced. There is a tension in reporting the results of qua-
litative analysis between conveying the rich detail in the data,
and excessive data reduction. Attempting to report rich detail
across 36 cases is of limited value. The form of display adopted
here is a compromise based on the need to convey the basis on
which the cross-case patterns discussed in the paper have been
identi?ed, while remaining relatively readable. Figs. 5 and 6 can
be audited against Appendix E via the nominated network
paths in the right-hand side column of Appendix E.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 505
Fig. 5. Network diagram of patterns in the data re?ecting the implications of using mechanisms that facilitate the implementation of
multiple strategic performance dimensions.
506 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
less stringent reaction to variances. These man-
agers play down the importance of individual
variances at the subunit level.
The next most common facilitative approach
reported in the data is in the nature of a ‘technical’
integration of e?ciency and other performance
dimensions by restating e?ciency measures to
accommodate the impact of other performance
expectations. Interestingly, this approach applies
almost exclusively to the integration of quality and
e?ciency expectations. This theme is evident in 10
descriptions of ‘e?ective’ management of multiple
strategic dimensions.
Few other facilitative approaches are evident in
the data. Value-added management, MRP II
(both represented by the Fig. 5 network path ori-
ginating at ‘‘innovative performance measure-
ment’’), weighted performance indices and
structural mechanisms such as teams or small fac-
tory-based ‘strategic business units’ are suggested
in individual cases to promote a performance
focus that enhances the e?ective management of
multiple performance expectations. As illustrated
in Fig. 5, there are only seven cases in total using
these three facilitative mechanisms. It is notable
also that there are seven cases of problematic
strategy implementation arising in spite of the
presence of facilitative mechanisms. These are all
focused on the management of performance
expectations relating to responsiveness in con-
junction with either e?ciency or quality.
The patterns in the data are also informative in
identifying the implications of a failure to develop
explicit integrative mechanisms. Using the same
strategy implementation outcome categories that
were used in Fig. 5, Fig. 6 tracks the implications
of a lack of facilitative mechanisms. There are 14
cases where there is no evidence of facilitators
speci?cally implemented to manage multiple per-
formance dimensions. All but two of these cases
experience di?culties in strategy implementation.
All problematic implementations in the data relate to
speci?c pairings of quality/e?ciency, responsiveness/
e?ciency or quality/responsiveness. Furthermore,
eight of these relate to the combination of respon-
siveness and e?ciency performance dimensions.
Overall, the data suggest that performance
management problems are those that require the
management of responsiveness in conjunction with
either quality or e?ciency performance dimen-
sions. This pattern is apparent for all of the seven
problematic cases reported by pro?t centres using
facilitative approaches (Fig. 5), and nine cases out
of the 12 problematic cases without facilitators
(Fig. 6). In the case of the pro?t centres using
facilitative approaches, it is notable that respon-
siveness pressures present a source of con?ict with
other manufacturing performance dimensions in
six cases even when there is a reduced emphasis on
variances in performance measurement (Fig. 5).
6. Discussion
Each individual case has its unique and complex
story that consists of elements of strategy, leader-
ship style, performance measurement practice,
strength of communication, functional isolation
and integration. It was not the purpose of this
study to explore these stories in depth, but rather
to identify and interpret patterns in the cross-sec-
tional data. The ?ndings demonstrate the exis-
tence of patterns among the variables of interest
and these patterns can be interpreted within the
exploratory con?nes of this study. Ultimately they
may provide a foundation for further research.
It is evident in the ?ndings of this study that
virtually all of the ?rms in this sample translate
pro?t centre strategy into multiple measures of
manufacturing performance. A variety of approa-
ches are used to manage these disaggregated mea-
sures and the potential con?icts that may occur
among them. Reduced emphasis on variances has
the most signi?cant impact on facilitating the
implementation of multi-dimensional strategy.
Consistent with the ?ndings reported by Van Der
Stede (2000), less rigid budgetary control enhances
the management of interdependencies by reducing
speci?c cost centre accountability for e?ciency
against pre-set cost targets. It allows room for
trade-o?s from the costs associated with imple-
menting responsiveness within a manufacturing
cost centre to be evaluated more holistically across
the ?rm’s value chain. However, loose budgetary
control is not universally e?ective in supporting
the management of multiple performance dimen-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 507
Fig. 6. Network diagram of patterns in the data re?ecting the implications of not using mechanisms that facilitate the implementation
of multiple strategic performance dimensions.
508 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
sions. There are six cases where performance
management remains problematic in spite of less
rigid reactions to variances.
E?ective integration of multiple performance
dimensions appears to be frequently associated
with managing quality with the demands for
manufacturing e?ciency and productivity. In
contrast perceived problems with strategy imple-
mentation are commonly associated with the pur-
suit of customer responsiveness. Closer reading of
the data in Appendix E, and the more detailed
data that lie behind this summary, indicates that
management of quality and e?ciency performance
is relatively easy. These data also suggest that it is
the management of customer responsiveness stra-
tegies with other manufacturing performance
dimensions that is most di?cult (i.e. responsive-
ness with e?ciency and/or quality). Management
appears to be able to anticipate and measure the
performance impacts of quality to a much greater
extent than they are able to with responsive man-
ufacturing. Product quality attributes have cost
consequences that can be identi?ed. Quality can
be readily costed into cost targets so there is no
incentive to compromise quality to meet these
targets. The capacity to integrate quality dimen-
sions into cost targets increases the completeness
of e?ciency and productivity criteria at the man-
ufacturing subunit level as these are ‘restated’ for
the impact on cost targets of quality levels, leaving
no incentive to compromise on quality to achieve
cost targets. On the other hand, the impact of
‘quick response’ on e?ciency performance criteria
and quality is, by nature, di?cult to predict and
adjust for. In fact, the most di?cult control pro-
blem evident in the cases examined here exists
where there is a strategic emphasis on responsive-
ness, signi?cant market-based competition on cost
and thus the need to manage constant friction
between e?cient and responsive manufacturing.
It is notable that all of the cases with loose
reactions to variances and problematic perfor-
mance management were focused on responsive-
ness not quality. In these settings, the e?ects of
disruption and product mix changes still produce
performance variances within the manufacturing
subunit. The evidence presented here suggests that
even when the reaction to variances at the pro?t
centre level is deliberately loose, cost centre man-
agers may still resist responsiveness to avoid
showing performance variances. Reference to the
broader data set suggests that pro?t centre man-
agers intervene and use communication mechan-
isms to reduce resistance but, as discussed below,
it remains a source of frustration that such inter-
ventions should be required repetitively.
There are few articulated mechanisms even
among the ‘e?ective’ cases in this study that sup-
port e?ective management of performance along
dimensions of e?ciency and responsiveness
jointly. Given the widespread importance of man-
ufacturing e?ciency, this combination is expected
to arise commonly when pro?t centres pursue a
customer-responsive strategy. A small number of
cases utilize structural mechanisms such as teams.
It is of interest that so few cases discussed struc-
tural adjustments to enhance the management of
interdependencies when it has been noted exten-
sively in the literature that such adjustments
would be required to implement strategies depen-
dent on cross-functional cooperation (Abernethy
& Lillis, 1995; Bowen et al., 1989; Parthasarthy &
Sethi, 1992). This result is, however, consistent
with Davila’s (2000) ?nding of limited use of
cross-functional teams in the control of product
development. Similarly there was little evidence of
PMS innovation beyond the adoption of non-
?nancial performance measures. While value-
added management was cited in only one case, the
discussion surrounding this approach focused on
the ability to integrate multiple performance
dimensions including responsiveness. The value-
added measurement in this case is similar to that
described by Lind (2001). Consistent with Lind’s
observation, the use of value-added per man-hour
rather than more conventional productivity mea-
sures of standard output per man-hour focuses
attention on the whole process of a manufactured
product rather than the output from an individual
operation, thus potentially integrating more ele-
ments of the value chain into disaggregated per-
formance measures.
Beyond these few cases there is little evidence of
e?ective means of developing PMSs to support
strategies focused on responsiveness. Elaborated
responses suggest that the performance manage-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 509
ment requirements imposed by responsiveness
strategies are met by greater pro?t centre level
management involvement and intervention in day-
to-day manufacturing operations in an attempt to
manage the inherent friction between responsive-
ness and e?ciency. While this intervention may be
the sort of productive strategic dialogue that is
envisaged by Norreklit (2000) and Kaplan and
Norton (2001) the data suggest that pro?t cen-
tre managers are unconvinced at the productiv-
ity and e?ectiveness of such interventions, as
well as being wary of the direct impact of
incomplete PMSs on subunit decisions. The fol-
lowing comments illustrate common themes from
the interviews:
Production is so intent on meeting their
weekly targets, if a special order comes in
they tend to say ‘‘Oh no what a nuisance’’,
rather than looking at the opportunity pre-
sented. And that’s fair enough, that’s where
they’re valued at. . .That’s their whole reward
system. Yes [the special order does get done],
but it takes a lot of management e?ort to tell
people that they are going to do it. (Case 11)
Setters and leading hands are imbued with
this view that the line must not stop and try as
we may we cannot get that out of their thinking.
The trouble is we have far too many long ser-
ving employees and they knowthat they have to
get 25,000 products o? that line this shift and
they will do it. . .They’ll believe that they have
done a good job and in fact there may be even
some of the management mechanisms that tell
them they’re doing a good job. But it might
not match the customer service angle and
that’s what’s wrong. (Case 30)
The manager in Case 30, was not con?dent that
PMS change would solve the problem which he saw
as relating to communication and cultural di?culties.
It is also of interest that none of the pro?t centre
managers referred to the deliberate use of con?ict
or incomplete local performance measures as a
catalyst to productive cross-functional or strategic
dialogue. It may of course be that the design of
this study and the researcher’s functionalist per-
spective induced greater re?ection on system
characteristics than deeper, more intuitive control
mechanisms. However, many managers discussed
con?ict and frustration so the opportunity to discuss
its positive value was present also. To the extent that
the pro?t centre manager orientation can be descri-
bed in general terms, it seems to consistently re?ect
an expectation that quantitative control systems will
prevail over more ?exible styles of management
simply ‘‘because they exist’’ and thus in?uence func-
tional subunit behaviour signi?cantly.
7. Conclusion
This study has sought to contribute empirical
insights relating to the management of multiple
manufacturing performance dimensions to a pre-
dominantly prescriptive literature on e?ective
PMS design. It was argued that PMSs consisting
of an array of ?nancial and non-?nancial perfor-
mance measures are consistent with literature
linking strategy with PMS design, and the norma-
tive balanced scorecard and integrated perfor-
mance measurement literature. The ?ndings of this
exploratory study suggest that mutual consistency
among multiple performance dimensions may be
problematic when these multiple measures are
disaggregated into partial sets of measures in
functional subunits. This study examined speci?-
cally the potentially con?icting performance
dimensions that arise in the disaggregation of
pro?t centre performance dimensions to manu-
facturing subunits. There is some preliminary evi-
dence from this study that performance
management problems appear to occur more in
the context of a strategic emphasis on customer
responsiveness, than quality. Furthermore the
data suggest that the problem is not one of lack of
strategic commitment. Rather, the di?erential dif-
?culties experienced with managing customer
responsiveness, along with the comments of the
managers, suggest that it is a friction created by
the failure to determine and adjust for the impli-
cations of pro?t centre strategy on the manu-
facturing cost function.
Given the apparent ease with which quality
performance is managed along with general
510 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
expectations for manufacturing e?ciency and
productivity, it would seem promising to pursue
similar performance management approaches in
the context of responsiveness. In order to elicit
commitment to customer responsiveness from a
manufacturing cost centre it would seem advanta-
geous to integrate multiple measures by quantifying
the impact of pursuit of quality and responsiveness
strategies on e?ciency and productivity. Such quan-
ti?cation would enhance the completeness of manu-
facturing subunit performance measures. The data
presented here suggest that pro?t centre managers
perceive less problems in managing the performance
of cost centres when they are able to construct more
complete measures of performance, and that they
experience strategy implementation problems when
measures are incomplete. In situations where per-
formance measures are local and disaggregated but
strategies require the management of functional
interdependencies the design of complete measures is
likely to be very challenging. The costs of being
responsive are re?ected in disruption, changeovers,
shortened lead times, product mix changes which
are di?cult to estimate, situation-dependent, and
not traceable to product speci?cations. Thus it
remains di?cult to develop technically complete
measures by restating cost targets for the antici-
pated costs of responsiveness.
The failure of traditional management account-
ing systems to quantify costs and bene?ts of cus-
tomization, coordination or the e?ects of multiple
goals have been noted in the literature (Fisher,
1992; Hergert & Morris, 1989). The accountant’s
role in developing and monitoring the costs of pro-
duct and process characteristics also has been high-
lighted (Bromwich, 1990) but there are few
documented models for identifying customization
costs and managing inherent tradeo?s (Srinidhi,
1992 is an example). More widespread documenta-
tion of these performance management di?culties
may encourage future research into the development
and application of models of con?ict reduction in the
context of managing a joint emphasis on e?ciency
and responsiveness in manufacturing cost centres.
The ?ndings reported here should be interpreted
in the light of some potential limitations. First, the
results are subject to signi?cant limitations
because of sample size, ?rm size and limited geo-
graphical spread. This is an exploratory study
drawing on data collected from a small sample of
pro?t centre managers. Further research is
required to test the generalizability of the ?ndings
reported here. Furthermore, more in-depth analy-
sis of attempts at managing multiple performance
dimensions in particular cases may identify
mechanisms to facilitate strategy implementation
that were overlooked in this study. Drawing on
data from 36 pro?t centres, this study compro-
mises on depth of analysis compared with ?eld
studies, and on breadth, generalizability and
reliability of analysis compared with cross-sec-
tional surveys. The issues examined in this paper
could have been examined in either greater depth
or breadth, with potentially di?erent results. Sin-
gle case studies may elicit di?erent rationales for
the design of performance measurement systems,
their link with strategy, the role and management
of con?ict and the managerial mechanisms that
facilitate strategy implementation. Examination of
political agendas, the role of inertia, contagion
e?ects and the contrast between detrimental task
con?ict and fruitful dynamic tension are all possi-
ble within in-depth ?eld studies, but not possible
in a study spanning 36 ?rms. In contrast, broad-
based survey studies would be able to examine a
broader, more representative strategic pro?le
within a sample and, by covering a greater range
of di?erences, draw more reliable generalizations
about relations between strategy, performance
measurement system design and issues relating to
performance management.
Within the con?nes of this study, analyses are
built on interpretations of qualitative data and are
therefore potentially subject to considerable bias.
Attempts have been made to minimize analytical
bias by maintaining a disciplined, systematic ana-
lytical protocol. Nonetheless, others may view the
data di?erently and interpret patterns that are not
drawn out in this paper. The reporting of patterns
observable in a limited set of cross-sectional qua-
litative data in this paper attempts to provide
‘food’ for further work. In particular, the ?ndings
of this exploratory study make a small but non-
trivial contribution to literature examining the
in?uence of manufacturing strategy on PMS com-
position, the tension between developing complete
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 511
measures of subunit performance and fostering
interdependencies, and the unresolved managerial
challenges relating to the management of multiple,
disaggregated dimensions of performance.
Acknowledgements
I am grateful to Margaret Abernethy, Frank
Selto, Anthony Hopwood and two anonymous
referees for their constructive contributions to this
paper. I also acknowledge helpful comments on
earlier versions of this paper contributed by the
late Peter Brownell, participants at the University
of New South Wales Fifth Biennial Management
Accounting Research Conference and the Amer-
ican Accounting Association Annual Meeting
(August, 1997). Finally, I thank the Australian
Centre for Management Accounting Development
(ACMAD) for ?nancial assistance.
Appendix A. Demographic data relating to the sample
1. Industry classi?cation of participating ?rms
Industry No. ?rms in
sampling frame
No. ?rms
contacted
No. ?rms
that agreed
to participate
23 Textiles 23 11 8
24 Clothing and Footwear 29 13 7
26 Paper, paper products 23 8 4
31 Fabricated metal products 31 14 5
34 Miscellaneous Manufacturing 37 20 12
143 66 36
2. Size of participant pro?t centres (by number of employees)
Size No. pro?t centres
Drawing on qualitative data collected during semi-structured interviews with 36profit centre managers in manufacturing
firms in Victoria, Australia, this study seeks to explore the mechanisms used to manage multiple manufacturing
performance dimensions arising from the pursuit of profit centre strategy. Where measures capture
potentially conflicting influences on the manufacturing cost function, strategy implementation is facilitated by loosening
control reactions to cost variances and through explicit attempts to integrate multiple measures.
Managing multiple dimensions of manufacturing
performance — an exploratory study
Anne M. Lillis*
Department of Accounting, University of Melbourne, Parkville 3010, Victoria, Australia
Abstract
Drawing on qualitative data collected during semi-structured interviews with 36 pro?t centre managers in manu-
facturing ?rms in Victoria, Australia, this study seeks to explore the mechanisms used to manage multiple manu-
facturing performance dimensions arising from the pursuit of pro?t centre strategy. Where measures capture
potentially con?icting in?uences on the manufacturing cost function, strategy implementation is facilitated by loosen-
ing control reactions to cost variances and through explicit attempts to integrate multiple measures. However, a joint
emphasis on performance dimensions relating to manufacturing e?ciency and customer responsiveness emerges as
problematic. In contrast, a joint emphasis on quality and e?ciency is relatively easily managed. It is suggested that in
the context of responsiveness strategies, the di?culty of designing complete measures inhibits the e?ectiveness of per-
formance measurement systems as a facilitator of strategy implementation. # 2002 Elsevier Science Ltd. All rights
reserved.
1. Introduction
The importance of designing performance
measurement systems that capture a range of
strategically important criteria in ?nancial and
non-?nancial terms is well established in the lit-
erature. Dominant themes in the performance
measurement literature relate to integration
(Nanni, Dixon, & Vollman, 1992), coherence (De
Haas & Kleingeld, 1999), the notion of a balanced
scorecard (Kaplan & Norton, 1992, 1996a, 2001),
and de?ning the performance variables that
represent important dimensions of a given strategy
(Simons, 1995). In all cases the message is that
e?ective performance measures must be able to
assess the ?rm’s progress on strategic initiatives
(Ittner & Larcker, 2001; Lang?eld-Smith, 1997;
Moon & Fitzgerald, 1996).
The literature that deals with the e?ective design
of performance measurement systems (PMSs)
incorporating multiple ?nancial and non-?nancial
measures emphasizes many attributes. These
include causal connections with strategy (Kaplan
& Norton, 1996a) integrating actions across func-
tional boundaries (De Haas & Kleingeld, 1999;
Nanni et al., 1992) and supporting critical strate-
gic measures with e?ective target setting and
reward systems (Moon & Fitzgerald, 1996; Otley,
1999). It has, however, been suggested that many
issues relating to implementation of multiple per-
formance measures remain unresolved (Hemmer,
1996) and that ‘‘research is needed on the treat-
ment of the inevitable trade-o?s that managers
0361-3682/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PI I : S0361- 3682( 01) 00032- 0
Accounting, Organizations and Society 27 (2002) 497–529
www.elsevier.com/locate/aos
* Tel.: +61-3-8344-5351; fax: +61-3-9349-2397.
E-mail address: [email protected] (A.M. Lillis).
will need to make among various ?nancial and
non-?nancial performance measures’’ (Ittner &
Larcker, 1998: 229).
This study focuses on the implementation issues
that arise when pro?t centre PMSs are dis-
aggregated to the functional subunit level. More
speci?cally, this study is designed to explore the
performance management challenges involved
when the implementation of pro?t centre strategy
requires manufacturing subunits to perform on
multiple and potentially con?icting dimensions.
Relevant performance dimensions at the manu-
facturing cost centre level may include, for exam-
ple, e?ciency, productivity, quality, response time
and delivery performance. While all are broadly
reconcilable at a pro?t centre level, these demands
are potentially con?icting performance parameters
at a functional subunit level. This paper reports
the ?ndings of an exploratory study into two
issues related to performance management of
manufacturing subunits:
1. the in?uence of pro?t centre strategy on the
formulation of PMSs for manufacturing
subunits; and
2. identi?cation of the ways in which multiple
manufacturing performance dimensions are
managed to facilitate the implementation of
pro?t centre strategy.
This study focuses on measures of manufactur-
ing cost centre performance utilized by pro?t cen-
tre managers as a key mechanism of performance
management in manufacturing ?rms. The perfor-
mance measures that are relied upon by pro?t
centre managers in evaluating manufacturing cost
centre performance signal strategic priorities and
performance expectations that in?uence cost cen-
tre decisions. The cost centre setting is invariably
one of high interdependencies. Strategies focused
on quality and responsiveness compound both
uncertainty and interdependence (Abernethy &
Lillis, 1995; Bowen, Siehl, & Schneider, 1989;
Fisher & Govindarajan, 1993; Nemetz & Fry,
1988; Van der Stede, 2000), and thus further
reduce the likelihood that complete subunit-level
performance measures can be constructed (Hirst,
1981). Pro?t centre managers are in a position to
discuss the rationale for the way multiple perfor-
mance dimensions are managed in such settings,
as well as any problems or unintended con-
sequences experienced. The ?ndings from this
study provide insights on issues of PMS imple-
mentation that do not feature in the pre-
dominantly prescriptive literature on e?ective
PMS design. This is a small sample study based on
36 case studies in a cross-section of ?rms. The data
are compiled and interpreted in the form of pat-
terns across cases. In turn, these patterns suggest
potential avenues for broader empirical research
to test their generalizability as well as raising
issues for theoretical consideration.
The remainder of the paper is organized as fol-
lows. The conceptual framework for this study is
outlined in Section 2. The research questions are
then identi?ed. This is followed by a description of
the study design and method, ?ndings and discus-
sion and some concluding comments.
2. Conceptual framework
2.1. Prior literature
In formulating PMS design characteristics to
e?ectively support strategy implementation, parti-
cular challenges have been apparent in the manu-
facturing industry. Rapid global change in the
bases of manufacturing competition has brought
new doctrines (e.g. JIT, TQM, and The Flexible
Factory), and required new approaches to perfor-
mance measurement (Abernethy & Lillis, 1995;
Banker, Potter, & Schroeder, 1993; Chenhall,
1997; Daniel & Rietsperger, 1991; Perera, Harri-
son, & Poole, 1997). The importance of quality,
?exibility and responsiveness has challenged the
relevance of conventional measures of manu-
facturing e?ciency (Abernethy & Lillis, 1995;
Dixon, Nanni, & Vollman, 1990; Kaplan, 1990;
Otley, 1994). One way in which this challenge has
been met is by PMS expansion. The literature has
established signi?cant associations between the
pursuit of speci?c strategies, such as TQM, JIT or
manufacturing ?exibility and the expansion of
traditional e?ciency-focused manufacturing
PMSs to embrace new manufacturing perfor-
mance measures (Abernethy & Lillis, 1995; Banker
498 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
et al., 1993; Chenhall, 1997; Daniel & Rietsperger,
1991; Perera et al., 1997). However, there have
been few positive performance outcomes asso-
ciated with such expansions (Ittner & Larcker,
1995; Perera et al., 1997; Young & Selto, 1993).
Ittner and Larcker (1998) suggest that the resul-
tant widespread use of multiple measures raises
several implementation issues including the like-
lihood of ine?ective spreading of managerial
e?ort.
The design of e?ective PMSs embracing multi-
ple measures has been considered in some depth as
the literature has sought to distinguish balanced,
integrated or coherent PMSs from a proliferation
of measures developed in an ad hoc way in
response to competitive pressures (Kaplan &
Norton, 1996b; Nanni et al., 1992; Norreklit,
2000; Simons, 1995). This literature emphasizes
the importance of explicitly embracing critical
dimensions of organizational strategy within
PMSs as well as the role of the PMS as a catalyst
for strategic dialogue throughout the organization
(Kaplan & Norton, 2001; Norreklit, 2000). While
much of the prescriptive performance measure-
ment literature is highly persuasive, implementa-
tion issues are beginning to feature in the
literature. Many of these issues are raised in the
context of the Balanced Scorecard because it is
arguably the most prescriptive and fully described
of contemporary frameworks. Several of these
challenges relate speci?cally to the multiplicity of
scorecard measures and their disaggregation.
Examples include the di?culty of attaining mutual
consistency among disaggregated measures of
interdependent processes (Sterman, Repenning, &
Kofman, 1997), the impact of cognitive limitations
on the processing of multiple performance mea-
sures (Lipe & Salterio, 2001) and the extent to
which the e?ectiveness of disaggregated PMSs
relies on a level of goal congruence and productive
strategic dialogue that may not be feasible (Nor-
reklit, 2000).
Kaplan and Norton (2001) address extensively
the disaggregation of corporate scorecards to the
subunit and individual level. While they consider
the potential for con?ict among measures of sub-
unit performance, they rely on the communication
strength of the balanced scorecard to generate
dialogue that will ultimately resolve di?erences in
perception and achieve goal congruence (Kaplan
& Norton, 2001; Norreklit, 2000). Related themes
in the literature refer to such con?ict as fruitful
dynamic tension (Simons, 1995) or an e?ective
signal of the need for strategic re-evaluation
(Kaplan & Norton, 1996b). Intraorganizational
con?ict has also been viewed as a productive e?ect
of the in?uence of uncontrollable factors on per-
formance measurement when there are signi?cant
interdependencies. In such settings, performance
metrics may result in productive cross-functional
pressure to improve shared outcomes (Merchant,
1987).
The presumption that con?ict induced by sub-
unit interdependencies and localized performance
measurement will be avoided or resolved through
dialogue, or alternatively that such con?ict will
represent an important signal may, however,
understate the direct in?uence of performance
measures on subunit performance (Otley, 1978;
Vagneur & Pieperl, 2000). The in?uence of speci?c
performance criteria relative to other less objective
in?uences on subunit performance is not clear,
and the preferences of evaluators in trading o?
performance measurement for dialogue remains
unexplored (Hartmann, 2000). There is, however,
general theoretical and empirical support for the
proposition that the presence of interdependencies
does not appear to limit reliance on subunit per-
formance measurement (Hartmann, 2000; MacIn-
tosh & Daft, 1987). This is evident despite the fact
that interdependency increases uncertainty
regarding subunit performance outcomes and
reduces controllability over those outcomes (Mer-
chant, 1987; Thomson, 1967). Furthermore, the
incompleteness of local measures may, in itself,
reduce the value of these measures as a catalyst for
productive dialogue. For example, the fact that
productivity and responsiveness have inherently
con?icting e?ects on a subunit cost function may
induce repetitive cross-functional dialogue to for-
mulate a ‘complete’ ?nancial picture of the trans-
action, but the productivity of such dialogue is
questionable (Dent, 1987), and there is always the
risk of unilateral subunit decision making without
engaging in cross-functional dialogue. The risk
associated with incomplete measures of perfor-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 499
mance is that they may ‘‘mobilise individual ener-
gies and motivations [resulting] in a transfer of
interest from the wider organizational purposes
which they were designed to serve onto the speci?c
behaviours which are necessary to improve the
indices of performance’’ (Hopwood, 1973: 16).
While the prior literature addressed this perfor-
mance management issue by focusing generally on
attributes of PMSs (such as controllability) and
in?uences on PMS design (such as inter-
dependencies and uncertainty), contemporary
performance management frameworks focus more
on the broader domain of strategic information
systems in which PMSs are just one element.
There are solutions evident in the literature that
rely less on con?ict avoidance and resolution
through dialogue. These include approaches to
con?ict avoidance that rely on changes to the per-
formance measures themselves. Examples include
the technical integration of multiple measures by
modifying internal measures to capture the impact
of multiple performance goals (McNair et al.,
1990), the use of explicit or implicit weightings on
measures to focus managerial e?ort (Ittner &
Larcker, 1998), or the tolerance of slack and the
less rigid use of budgetary controls (Van Der
Stede, 2000). Alternatively, Abernethy and Lillis
(1995) highlight the importance of integrative
structural forms such as cross-functional teams as
a means of avoiding the need to rely on PMSs to
promote co-ordination.
2.2. Summary
Fig. 1 illustrates the issue of concern in this
study. Issues related to the mapping of perfor-
mance measures to strategy at the pro?t centre
level are not considered in this study. Instead, it
focuses on the disaggregation of pro?t-centre level
performance measures to the functional subunit
level. Subunits X and Y have their own PMSs, but
they are highly interdependent. This would be the
case if, for example, X and Y are manufacturing
and sales departments and the pro?t centre is
pursuing a strategy of customer responsiveness. In
such a setting performance measurement in the
manufacturing cost centre may include customer-
focused measures of quality, response time and
delivery performance and business process or
?nancial measures of e?ciency and productivity.
1
At the pro?t centre level, these customer-focused
and business process measures are integrated
e?ectively as part of the causal chain linking
strategy and pro?t-focused ?nancial outcomes,
but e?ciency and customer responsiveness still
represent con?icting in?uences on the manu-
facturing cost function.
2
Even in the context of an
organizationally pro?table product customization,
disruption caused by customization is almost cer-
tain to reduce e?ciency and productivity at the
manufacturing cost centre level. It is also likely to
reduce overall performance on response time and
delivery as other customer orders are set aside to
accommodate the customization. The gains are
clearly evident elsewhere in the ?rm—primarily in
the sales subunit. The important point here is that
even expanded performance measures, at some
level of disaggregation, become internally-focused
and incomplete. Pro?t centre outcomes are a
function of the outcomes of both functional sub-
units independently and also of the way their
strategic interdependencies are managed. While
this may occur through speci?c mechanisms that
systematically manage multiple performance
dimensions throughout the pro?t centre, we have
little empirical evidence as to how these mechan-
isms work in practice.
We know empirically that multiple measures of
manufacturing subunit performance are common
and that the breadth of measures is increasing to
capture measures of e?ciency, productivity, qual-
1
While Balanced Scorecard terminology is used here, it is
not suggested that this problem is speci?c to Balanced Scor-
ecard implementations. It may arise whenever strategy set at a
higher organizational level is translated into disaggregated
multiple performance measures.
2
This is somewhat similar to the argument of Bushman,
Indjejikian, and Smith (1995) that where there is signi?cant
subunit interdependency, highly disaggregated, local perfor-
mance measures become less informative in the context of
compensation contracts. It is consistent also with the explana-
tions o?ered by Sterman et al. (1997) in their attempts to
explain the lack of mutual consistency among operational and
?nancial gains at Analog Devices. Sterman et al. (1997) argue
that much of the paradox is attributable to attempts to dis-
aggregate subunit improvement e?orts when processes are
inherently tightly coupled across the organization.
500 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
ity and responsiveness. We have normative
descriptions of the characteristics of balanced,
integrated and coherent PMSs that distinguish
these from ad hoc collections of multiple measures,
but there is no evidence that these contemporary
frameworks avoid the use of incomplete measures
at the subunit level. In this largely prescriptive lit-
erature, the mechanisms required to manage highly
disaggregated, incomplete measures of functional
subunit performance are not obvious. We know
little empirically of the features of PMSs at this
level, the nature of the con?icts created by dis-
aggregation, whether integration of performance
measures is sought and the way it is attempted. In
the absence of extensive empirical examination of
these phenomena, this study seeks to compile and
interpret the insights o?ered by a small sample of
pro?t centre managers into these issues.
3. Research questions
Using a cross-sectional case study research
design and qualitative data this study seeks to:
1. examine the in?uence of pro?t centre strat-
egy on the formulation of PMSs for manu-
facturing subunits; and
2. identify the ways multiple performance mea-
sures are managed to facilitate strategy
implementation.
Fig. 2 places the research questions within the
broad domain of Fig. 1. This study focuses ?rst on
identifying manufacturing subunit performance
measures used by pro?t centre managers against a
context of pro?t centre strategy. Second, it focuses
on the mechanisms used by pro?t centre managers
to implement pro?t centre strategy through man-
ufacturing subunits, given the in?uence of both
local performance measures and cross-functional
interdependencies.
4. Study design and method
Underlying the data collection and analysis
method for this study is the desire to explore ‘how’
and ‘why’ questions relating to management prac-
tices, their rationale, and expected outcomes (Yin,
1994). Rather than studying these phenomena in
depth within a single case, the aim is to link con-
textual and performance measurement character-
istics through the exploration of patterns among
cases. In order to observe patterns in the connec-
tions among these variables the data collected are
Fig. 1. Diagrammatic representation of the problem of disaggregation of performance measures when there are strategic inter-
dependencies.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 501
cross-sectional and qualitative. To explore issues
related to the management of multiple perfor-
mance dimensions this study draws on managerial
re?ections captured through semi-structured
interviews conducted in a small sample of pro?t
centres. This approach limits the depth of data
available on individual cases but provides greater
cross-sectional comparison than is typically avail-
able in ?eld research studies.
4.1. Description of the sample
Data were collected from 36 pro?t centre man-
agers in manufacturing ?rms in Victoria, Australia.
A general database of Victorian manufacturers
was used to select ?rms with at least 200 employ-
ees. Firms were selected across a range of indus-
tries with the aim of selecting participants with a
range of strategic orientations. Broad character-
istics of the sample and the managerial partici-
pants are given in Appendix A. ‘Firms’ in this
context may be fully independent or subunits of
other larger ?rms. They all, however, appeared as
separate entries in the Victorian Manufacturing
Directory. Given the size of most of the ?rms in
the sample, the general manager was the ?rst level
linkage for functional units. Thus, these managers
had the equivalent of ‘pro?t centre’ responsi-
bility.
3
While seven pro?t centres (independently
listed in the database) were drawn from two large
diversi?ed companies, each was from a separate
autonomous division. The sample did not include
Fig. 2. Placement of this study within the general domain of Fig. 1.
502 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
more than one pro?t centre from any one division
of a company. All of the remaining participants
were either independent ?rms or the only pro?t
centres included from the relevant company. The
response rate was 65% with 55 e?ective contacts
yielding 36 interviews with pro?t centre managers
and only 19 refusals.
4
With one exception, all
interviews were tape recorded and transcribed
verbatim. The sections of the interview guide rele-
vant to this paper are reproduced in Appendix B.
4.2. Data analysis
This paper reports ?ndings in the form of pat-
terns across cases. These patterns are a highly
summarized product of the researcher’s detailed
analysis of more extensive data. To limit the ana-
lytical bias inherent in the analysis of qualitative
data, a systematic, auditable process was used.
The systematic approach adopted here attempts to
provide an audit trail from interview transcript
through to theoretical proposition. In common
with other forms of data analyses, the process
involves data reduction or summarization, classi-
?cation and interpretation. Data were coded using
NUD
.
IST qualitative analysis software and then
analyzed using Miles and Huberman’s (1994) the-
matic conceptual matrix data displays. While the
?ndings reported in this paper are summarized in
a matrix format, the actual analytical matrices
used by the researcher were much more detailed
and frequently focused on examining patterns
between two variables rather than among many.
5
Throughout the analysis, an audit trail was main-
tained. As a result, all matrix cell entries are audi-
table back to transcripts through the use of
identifying case and text-unit numbers. Thus, the
link between data and ?ndings should be obser-
vable by others working from the same database.
The following section identi?es patterns in the
data. These patterns are then discussed in Section
6.
5. Findings—patterns in the data
5.1. The strategic pro?le of the sample
In order to explore the in?uence of pro?t centre
strategy on the implementation of multiple cost
centre performance measures, it was necessary ?rst
to identify the strategic pro?le of the sample.
Questions focused on the following attributes of
pro?t centre strategy and how these translate into
manufacturing strategy:
1. the pro?t centre’s competitive edge;
2. the nature of the markets within which the
pro?t centre operates;
3. the extent to which the pro?t centre attracts
price premiums consistent with di?erentia-
tion; and
Fig. 3. Classi?ed summary of strategic orientations.
3
In two cases it became apparent during an interview that
the general manager was not at an appropriate level, and in
both cases interviews were conducted subsequently in the same
?rms with managers at a lower, pro?t centre level. These latter
interviews are included. For simplicity, respondents are referred
to as ‘pro?t centre managers’.
4
From 66 letters sent to General Managers, 11 ?rms were
excluded without conducting an interview because they had
ceased manufacture in Australia, the General Manager was
located interstate or similar reason. Twelve managers refused to
participate, generally for reasons related to time pressure. In
seven cases, I gave up trying to make contact after approxi-
mately 10 phone calls. These seven cases have been classi?ed as
refusals.
5
These matrices are too detailed and numerous to include in
this paper. However, the patterns that emerged from that ana-
lysis ought to be evident when the data are reported in the full
case summary display format that is included in Appendices C,
D , E and referred to in the discussion of ?ndings.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 503
4. whether the pro?t centre faces demands for
product customization, and how these
demands are dealt with.
A summary of the strategic pro?le of the sample
is illustrated in Fig. 3.
6
Three strategic groupings
emerged from the analysis of these data. Two of
these groupings re?ect di?erentiation strategies
relating to either quality (based on stated com-
mitment to competing on quality, product design,
technical capability, product features, brand
names) or responsiveness (in terms of customer
service, product customization, and/or depend-
ability
7
). The third classi?cation relates to a stra-
tegic priority of cost leadership. Two important
features of the strategic pro?le of the sample
emerged from this analysis. First, there was the
prevalance of di?erentiation as a strategic priority.
There was an overwhelming emphasis on di?er-
entiation (29) with a very limited emphasis on low
cost as a strategic priority (three). Second, the
overlapping areas in Fig. 3 illustrate the presence
of widespread competition on multiple criteria
(23). The patterns observed here are consistent
with prior literature that has recognized the
decline in low cost mass production, the increasing
emphasis on product design innovation and cus-
tomer responsiveness as platforms of di?erentia-
tion, and the necessity for ?rms to compete on
multiple strategies (Belohlav, 1993; Drucker, 1990;
Hill, 1988; Jones & Butler, 1988; Nemetz & Fry,
1988; Otley, 1994).
5.2. The PMS pro?le of the sample
Fig. 4 gives an overview of the manufacturing
subunit PMS characteristics of the pro?t centres in
this study.
8
Performance measures used by pro?t
centre managers to measure manufacturing cost
centre performance are categorized into three
groups: cost and e?ciency measures, quality mea-
sures, and customer responsiveness measures. The
?ndings indicate that 34 of the pro?t centres use
?nancial and non-?nancial e?ciency and/or pro-
ductivity measures of manufacturing performance.
Of these, 21 use multiple performance measures
capturing performance on e?ciency, productivity,
quality and/or customer service. The measures
used to capture manufacturing performance
appear to be relatively generic.
9
The extent of
emphasis on e?ciency and productivity perfor-
mance measurement criteria in manufacturing
Fig. 4. Classi?ed summary of performance measures used to
measure manufacturing performance.
6
This diagram is based on data in Appendix C, which
summarizes and classi?es strategic orientations by case.
7
The ‘correct’ placement of ‘dependability’ is not clear.
Traditionally, dependability was associated with a defender-
type strategy (Miles & Snow, 1978). Frequently ‘reliability’,
which is akin to dependability refers to product reliaiblity and
is viewed as an element of a quality strategy. Dependability
may also refer to the ability to deliver on promises to custo-
mers, which is an element of a customer responsiveness or ser-
vice strategy. Its placement here with service and
responsiveness is based on the context in which the term was
used by the managers in this study. The term ‘dependability’ is
used in only four cases.
8
Fig. 4 is developed from the data in Appendix D. Appen-
dix D contains a listing of the measures identi?ed by each pro?t
centre manager as the measures relied upon in evaluating the
performance of the manufacturing subunit. In the right-hand
side column of Appendix D the measures used are classi?ed as
cost variance, non-?nancial e?ciency and productivity, exter-
nal quality and customer service measures.
9
While it is tempting to try to link Figs. 2 and 3 at the ?rm
level, such an analysis of the mapping of performance measures
to strategy is not the focus of this paper. It is evident at a glance
that the match between strategy and PMS composition is not
strong. It is obvious, for example, that the emphasis on cost
and non-?nancial e?ciency/productivity measures cannot be
matched with strategic orientations. Thus the PMS composi-
tion is best described as generic and somewhat inconsistent with
the overall view of strategic orientations.
504 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
suggests that an emphasis on low cost is almost
universally necessary, despite the fact that low cost
is a strategic priority in very few pro?t centres in
this sample. While this in itself is not surprising, it
creates a performance management challenge.
Manufacturing performance is measured based on
e?ciency and productivity criteria while at the
same time these subunits are focused on strategies
relating to responsiveness, service, dependability
or quality. The next section examines the chal-
lenges in managing multiple performance dimen-
sions that arise in this context.
5.3. Managing multiple performance dimensions
The qualitative data base for this study con-
tains, on a case-by-case basis, data relating to
patterns in a range of variables that combine to
‘tell the story’ of e?ective or problematic manage-
ment of the relations among multiple dimensions
of manufacturing performance. Given the focus of
this study on the management of multiple manu-
facturing performance dimensions, cases are clas-
si?ed as ‘e?ective’ in their implementation of
strategic priorities when pro?t centre managers
indicate that manufacturing subunits are suppor-
tive rather than resistant to the implementation of
pro?t centre strategy. While these cases do not
always suggest absolute compliance without dia-
logue, they indicate a level of acceptance within
manufacturing subunits of the need to perform on
multiple strategic dimensions. In contrast, cases
are classi?ed as ‘problematic’ in their imple-
mentation of strategic priorities when pro?t centre
managers indicate a perception that manufactur-
ing subunits inappropriately trade o? some per-
formance dimensions against others, engage in
non-productive con?ict, or resist expectations to
perform on multiple dimensions. Patterns in the
data are used to identify mechanisms that are used
by pro?t centre managers to facilitate strategy
implementation. It is possible to distinguish from
the data whether these facilitators result in
e?ective outcomes (i.e. are able to facilitate the
implementation of multiple strategic priorities)
or whether strategic implementation remains
problematic. These ?ndings are summarized in
Fig. 5.
10
Analysis of the data indicates that there are pri-
marily ?ve mechanisms used to facilitate the
implementation of multiple strategic priorities:
loosening of control reactions to variances, struc-
tural changes, innovative performance measure-
ment systems, and two speci?c adaptations to
PMSs to enable integration of multiple perfor-
mance measures, namely technical integration and
use of weightings. Fig. 5 contains several paths
representing a small number of occurrences in the
data. The paths with three or more occurrences
are bold for visibility. The following discussion
attempts to make sense of the patterns observed in
the data.
Fig. 5 tracks links between the use of facilitators
and pro?t centre managers’ perceptions of e?ec-
tive or problematic strategy implementation. In
the data, the managers’ perceptions are classi?ed
from both general and speci?c comments. More
general descriptions of the e?ective management
of all performance dimensions are captured in the
category ‘e?ective implementation of all strategic
priorities’. In other cases, descriptions of e?ective
outcomes are related to speci?c dimensions of
strategy. Thus the individual pairings of quality/
e?ciency, responsiveness/e?ciency and quality/
responsiveness are included to capture these ?ner
distinctions.
The pro?t centres that e?ectively manage the
demands for manufacturing e?ciency and pro-
ductivity with demands for quality and respon-
siveness do so by loosening the response to
e?ciency and productivity variances or by actively
employing other mechanisms that facilitate man-
agement of multiple performance dimensions.
Many (12) manage potential con?ict by taking a
10
Figs. 5 and 6 are developed from data in Appendix E,
which includes summary descriptions for strategy, PMS com-
position, reactions to e?ciency variances, mechanisms for
integrating multiple performance measures, and con?icts
experienced. There is a tension in reporting the results of qua-
litative analysis between conveying the rich detail in the data,
and excessive data reduction. Attempting to report rich detail
across 36 cases is of limited value. The form of display adopted
here is a compromise based on the need to convey the basis on
which the cross-case patterns discussed in the paper have been
identi?ed, while remaining relatively readable. Figs. 5 and 6 can
be audited against Appendix E via the nominated network
paths in the right-hand side column of Appendix E.
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 505
Fig. 5. Network diagram of patterns in the data re?ecting the implications of using mechanisms that facilitate the implementation of
multiple strategic performance dimensions.
506 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
less stringent reaction to variances. These man-
agers play down the importance of individual
variances at the subunit level.
The next most common facilitative approach
reported in the data is in the nature of a ‘technical’
integration of e?ciency and other performance
dimensions by restating e?ciency measures to
accommodate the impact of other performance
expectations. Interestingly, this approach applies
almost exclusively to the integration of quality and
e?ciency expectations. This theme is evident in 10
descriptions of ‘e?ective’ management of multiple
strategic dimensions.
Few other facilitative approaches are evident in
the data. Value-added management, MRP II
(both represented by the Fig. 5 network path ori-
ginating at ‘‘innovative performance measure-
ment’’), weighted performance indices and
structural mechanisms such as teams or small fac-
tory-based ‘strategic business units’ are suggested
in individual cases to promote a performance
focus that enhances the e?ective management of
multiple performance expectations. As illustrated
in Fig. 5, there are only seven cases in total using
these three facilitative mechanisms. It is notable
also that there are seven cases of problematic
strategy implementation arising in spite of the
presence of facilitative mechanisms. These are all
focused on the management of performance
expectations relating to responsiveness in con-
junction with either e?ciency or quality.
The patterns in the data are also informative in
identifying the implications of a failure to develop
explicit integrative mechanisms. Using the same
strategy implementation outcome categories that
were used in Fig. 5, Fig. 6 tracks the implications
of a lack of facilitative mechanisms. There are 14
cases where there is no evidence of facilitators
speci?cally implemented to manage multiple per-
formance dimensions. All but two of these cases
experience di?culties in strategy implementation.
All problematic implementations in the data relate to
speci?c pairings of quality/e?ciency, responsiveness/
e?ciency or quality/responsiveness. Furthermore,
eight of these relate to the combination of respon-
siveness and e?ciency performance dimensions.
Overall, the data suggest that performance
management problems are those that require the
management of responsiveness in conjunction with
either quality or e?ciency performance dimen-
sions. This pattern is apparent for all of the seven
problematic cases reported by pro?t centres using
facilitative approaches (Fig. 5), and nine cases out
of the 12 problematic cases without facilitators
(Fig. 6). In the case of the pro?t centres using
facilitative approaches, it is notable that respon-
siveness pressures present a source of con?ict with
other manufacturing performance dimensions in
six cases even when there is a reduced emphasis on
variances in performance measurement (Fig. 5).
6. Discussion
Each individual case has its unique and complex
story that consists of elements of strategy, leader-
ship style, performance measurement practice,
strength of communication, functional isolation
and integration. It was not the purpose of this
study to explore these stories in depth, but rather
to identify and interpret patterns in the cross-sec-
tional data. The ?ndings demonstrate the exis-
tence of patterns among the variables of interest
and these patterns can be interpreted within the
exploratory con?nes of this study. Ultimately they
may provide a foundation for further research.
It is evident in the ?ndings of this study that
virtually all of the ?rms in this sample translate
pro?t centre strategy into multiple measures of
manufacturing performance. A variety of approa-
ches are used to manage these disaggregated mea-
sures and the potential con?icts that may occur
among them. Reduced emphasis on variances has
the most signi?cant impact on facilitating the
implementation of multi-dimensional strategy.
Consistent with the ?ndings reported by Van Der
Stede (2000), less rigid budgetary control enhances
the management of interdependencies by reducing
speci?c cost centre accountability for e?ciency
against pre-set cost targets. It allows room for
trade-o?s from the costs associated with imple-
menting responsiveness within a manufacturing
cost centre to be evaluated more holistically across
the ?rm’s value chain. However, loose budgetary
control is not universally e?ective in supporting
the management of multiple performance dimen-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 507
Fig. 6. Network diagram of patterns in the data re?ecting the implications of not using mechanisms that facilitate the implementation
of multiple strategic performance dimensions.
508 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
sions. There are six cases where performance
management remains problematic in spite of less
rigid reactions to variances.
E?ective integration of multiple performance
dimensions appears to be frequently associated
with managing quality with the demands for
manufacturing e?ciency and productivity. In
contrast perceived problems with strategy imple-
mentation are commonly associated with the pur-
suit of customer responsiveness. Closer reading of
the data in Appendix E, and the more detailed
data that lie behind this summary, indicates that
management of quality and e?ciency performance
is relatively easy. These data also suggest that it is
the management of customer responsiveness stra-
tegies with other manufacturing performance
dimensions that is most di?cult (i.e. responsive-
ness with e?ciency and/or quality). Management
appears to be able to anticipate and measure the
performance impacts of quality to a much greater
extent than they are able to with responsive man-
ufacturing. Product quality attributes have cost
consequences that can be identi?ed. Quality can
be readily costed into cost targets so there is no
incentive to compromise quality to meet these
targets. The capacity to integrate quality dimen-
sions into cost targets increases the completeness
of e?ciency and productivity criteria at the man-
ufacturing subunit level as these are ‘restated’ for
the impact on cost targets of quality levels, leaving
no incentive to compromise on quality to achieve
cost targets. On the other hand, the impact of
‘quick response’ on e?ciency performance criteria
and quality is, by nature, di?cult to predict and
adjust for. In fact, the most di?cult control pro-
blem evident in the cases examined here exists
where there is a strategic emphasis on responsive-
ness, signi?cant market-based competition on cost
and thus the need to manage constant friction
between e?cient and responsive manufacturing.
It is notable that all of the cases with loose
reactions to variances and problematic perfor-
mance management were focused on responsive-
ness not quality. In these settings, the e?ects of
disruption and product mix changes still produce
performance variances within the manufacturing
subunit. The evidence presented here suggests that
even when the reaction to variances at the pro?t
centre level is deliberately loose, cost centre man-
agers may still resist responsiveness to avoid
showing performance variances. Reference to the
broader data set suggests that pro?t centre man-
agers intervene and use communication mechan-
isms to reduce resistance but, as discussed below,
it remains a source of frustration that such inter-
ventions should be required repetitively.
There are few articulated mechanisms even
among the ‘e?ective’ cases in this study that sup-
port e?ective management of performance along
dimensions of e?ciency and responsiveness
jointly. Given the widespread importance of man-
ufacturing e?ciency, this combination is expected
to arise commonly when pro?t centres pursue a
customer-responsive strategy. A small number of
cases utilize structural mechanisms such as teams.
It is of interest that so few cases discussed struc-
tural adjustments to enhance the management of
interdependencies when it has been noted exten-
sively in the literature that such adjustments
would be required to implement strategies depen-
dent on cross-functional cooperation (Abernethy
& Lillis, 1995; Bowen et al., 1989; Parthasarthy &
Sethi, 1992). This result is, however, consistent
with Davila’s (2000) ?nding of limited use of
cross-functional teams in the control of product
development. Similarly there was little evidence of
PMS innovation beyond the adoption of non-
?nancial performance measures. While value-
added management was cited in only one case, the
discussion surrounding this approach focused on
the ability to integrate multiple performance
dimensions including responsiveness. The value-
added measurement in this case is similar to that
described by Lind (2001). Consistent with Lind’s
observation, the use of value-added per man-hour
rather than more conventional productivity mea-
sures of standard output per man-hour focuses
attention on the whole process of a manufactured
product rather than the output from an individual
operation, thus potentially integrating more ele-
ments of the value chain into disaggregated per-
formance measures.
Beyond these few cases there is little evidence of
e?ective means of developing PMSs to support
strategies focused on responsiveness. Elaborated
responses suggest that the performance manage-
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 509
ment requirements imposed by responsiveness
strategies are met by greater pro?t centre level
management involvement and intervention in day-
to-day manufacturing operations in an attempt to
manage the inherent friction between responsive-
ness and e?ciency. While this intervention may be
the sort of productive strategic dialogue that is
envisaged by Norreklit (2000) and Kaplan and
Norton (2001) the data suggest that pro?t cen-
tre managers are unconvinced at the productiv-
ity and e?ectiveness of such interventions, as
well as being wary of the direct impact of
incomplete PMSs on subunit decisions. The fol-
lowing comments illustrate common themes from
the interviews:
Production is so intent on meeting their
weekly targets, if a special order comes in
they tend to say ‘‘Oh no what a nuisance’’,
rather than looking at the opportunity pre-
sented. And that’s fair enough, that’s where
they’re valued at. . .That’s their whole reward
system. Yes [the special order does get done],
but it takes a lot of management e?ort to tell
people that they are going to do it. (Case 11)
Setters and leading hands are imbued with
this view that the line must not stop and try as
we may we cannot get that out of their thinking.
The trouble is we have far too many long ser-
ving employees and they knowthat they have to
get 25,000 products o? that line this shift and
they will do it. . .They’ll believe that they have
done a good job and in fact there may be even
some of the management mechanisms that tell
them they’re doing a good job. But it might
not match the customer service angle and
that’s what’s wrong. (Case 30)
The manager in Case 30, was not con?dent that
PMS change would solve the problem which he saw
as relating to communication and cultural di?culties.
It is also of interest that none of the pro?t centre
managers referred to the deliberate use of con?ict
or incomplete local performance measures as a
catalyst to productive cross-functional or strategic
dialogue. It may of course be that the design of
this study and the researcher’s functionalist per-
spective induced greater re?ection on system
characteristics than deeper, more intuitive control
mechanisms. However, many managers discussed
con?ict and frustration so the opportunity to discuss
its positive value was present also. To the extent that
the pro?t centre manager orientation can be descri-
bed in general terms, it seems to consistently re?ect
an expectation that quantitative control systems will
prevail over more ?exible styles of management
simply ‘‘because they exist’’ and thus in?uence func-
tional subunit behaviour signi?cantly.
7. Conclusion
This study has sought to contribute empirical
insights relating to the management of multiple
manufacturing performance dimensions to a pre-
dominantly prescriptive literature on e?ective
PMS design. It was argued that PMSs consisting
of an array of ?nancial and non-?nancial perfor-
mance measures are consistent with literature
linking strategy with PMS design, and the norma-
tive balanced scorecard and integrated perfor-
mance measurement literature. The ?ndings of this
exploratory study suggest that mutual consistency
among multiple performance dimensions may be
problematic when these multiple measures are
disaggregated into partial sets of measures in
functional subunits. This study examined speci?-
cally the potentially con?icting performance
dimensions that arise in the disaggregation of
pro?t centre performance dimensions to manu-
facturing subunits. There is some preliminary evi-
dence from this study that performance
management problems appear to occur more in
the context of a strategic emphasis on customer
responsiveness, than quality. Furthermore the
data suggest that the problem is not one of lack of
strategic commitment. Rather, the di?erential dif-
?culties experienced with managing customer
responsiveness, along with the comments of the
managers, suggest that it is a friction created by
the failure to determine and adjust for the impli-
cations of pro?t centre strategy on the manu-
facturing cost function.
Given the apparent ease with which quality
performance is managed along with general
510 A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529
expectations for manufacturing e?ciency and
productivity, it would seem promising to pursue
similar performance management approaches in
the context of responsiveness. In order to elicit
commitment to customer responsiveness from a
manufacturing cost centre it would seem advanta-
geous to integrate multiple measures by quantifying
the impact of pursuit of quality and responsiveness
strategies on e?ciency and productivity. Such quan-
ti?cation would enhance the completeness of manu-
facturing subunit performance measures. The data
presented here suggest that pro?t centre managers
perceive less problems in managing the performance
of cost centres when they are able to construct more
complete measures of performance, and that they
experience strategy implementation problems when
measures are incomplete. In situations where per-
formance measures are local and disaggregated but
strategies require the management of functional
interdependencies the design of complete measures is
likely to be very challenging. The costs of being
responsive are re?ected in disruption, changeovers,
shortened lead times, product mix changes which
are di?cult to estimate, situation-dependent, and
not traceable to product speci?cations. Thus it
remains di?cult to develop technically complete
measures by restating cost targets for the antici-
pated costs of responsiveness.
The failure of traditional management account-
ing systems to quantify costs and bene?ts of cus-
tomization, coordination or the e?ects of multiple
goals have been noted in the literature (Fisher,
1992; Hergert & Morris, 1989). The accountant’s
role in developing and monitoring the costs of pro-
duct and process characteristics also has been high-
lighted (Bromwich, 1990) but there are few
documented models for identifying customization
costs and managing inherent tradeo?s (Srinidhi,
1992 is an example). More widespread documenta-
tion of these performance management di?culties
may encourage future research into the development
and application of models of con?ict reduction in the
context of managing a joint emphasis on e?ciency
and responsiveness in manufacturing cost centres.
The ?ndings reported here should be interpreted
in the light of some potential limitations. First, the
results are subject to signi?cant limitations
because of sample size, ?rm size and limited geo-
graphical spread. This is an exploratory study
drawing on data collected from a small sample of
pro?t centre managers. Further research is
required to test the generalizability of the ?ndings
reported here. Furthermore, more in-depth analy-
sis of attempts at managing multiple performance
dimensions in particular cases may identify
mechanisms to facilitate strategy implementation
that were overlooked in this study. Drawing on
data from 36 pro?t centres, this study compro-
mises on depth of analysis compared with ?eld
studies, and on breadth, generalizability and
reliability of analysis compared with cross-sec-
tional surveys. The issues examined in this paper
could have been examined in either greater depth
or breadth, with potentially di?erent results. Sin-
gle case studies may elicit di?erent rationales for
the design of performance measurement systems,
their link with strategy, the role and management
of con?ict and the managerial mechanisms that
facilitate strategy implementation. Examination of
political agendas, the role of inertia, contagion
e?ects and the contrast between detrimental task
con?ict and fruitful dynamic tension are all possi-
ble within in-depth ?eld studies, but not possible
in a study spanning 36 ?rms. In contrast, broad-
based survey studies would be able to examine a
broader, more representative strategic pro?le
within a sample and, by covering a greater range
of di?erences, draw more reliable generalizations
about relations between strategy, performance
measurement system design and issues relating to
performance management.
Within the con?nes of this study, analyses are
built on interpretations of qualitative data and are
therefore potentially subject to considerable bias.
Attempts have been made to minimize analytical
bias by maintaining a disciplined, systematic ana-
lytical protocol. Nonetheless, others may view the
data di?erently and interpret patterns that are not
drawn out in this paper. The reporting of patterns
observable in a limited set of cross-sectional qua-
litative data in this paper attempts to provide
‘food’ for further work. In particular, the ?ndings
of this exploratory study make a small but non-
trivial contribution to literature examining the
in?uence of manufacturing strategy on PMS com-
position, the tension between developing complete
A.M. Lillis / Accounting, Organizations and Society 27 (2002) 497–529 511
measures of subunit performance and fostering
interdependencies, and the unresolved managerial
challenges relating to the management of multiple,
disaggregated dimensions of performance.
Acknowledgements
I am grateful to Margaret Abernethy, Frank
Selto, Anthony Hopwood and two anonymous
referees for their constructive contributions to this
paper. I also acknowledge helpful comments on
earlier versions of this paper contributed by the
late Peter Brownell, participants at the University
of New South Wales Fifth Biennial Management
Accounting Research Conference and the Amer-
ican Accounting Association Annual Meeting
(August, 1997). Finally, I thank the Australian
Centre for Management Accounting Development
(ACMAD) for ?nancial assistance.
Appendix A. Demographic data relating to the sample
1. Industry classi?cation of participating ?rms
Industry No. ?rms in
sampling frame
No. ?rms
contacted
No. ?rms
that agreed
to participate
23 Textiles 23 11 8
24 Clothing and Footwear 29 13 7
26 Paper, paper products 23 8 4
31 Fabricated metal products 31 14 5
34 Miscellaneous Manufacturing 37 20 12
143 66 36
2. Size of participant pro?t centres (by number of employees)
Size No. pro?t centres