Journal of Operations Management 32 (2014) 313–336
Contents lists available at ScienceDirect
Journal of Operations Management
journal homepage: www.elsevier.com/locate/jom
The effect of performance measurement systems on firm
performance: A cross-sectional and a longitudinal study
Xenophon Koufteros a,∗ , Anto (John) Verghese a , Lorenzo Lucianetti b
a
Business Administration Department of Information & Operations Management Mays Business School, Texas A & M University Department of Information
& Operations Management, College Station, TX 77843-4217, United States
b
Department of Management and Business Administration University of Chieti and Pescara, Viale Pindaro 42 - 65127, Pescara, Italy
a r t i c l e
i n f o
Article history:
Received 11 December 2013
Received in revised form 13 June 2014
Accepted 17 June 2014
Keywords:
Performance Measurement Systems
Operations management
Accounting
Financial performance
Empirical
Panel data
a b s t r a c t
Performance measurement (PM) systems have been popularized over the last 20 years and the operations management literature is replete with discussion of metrics and measurement systems. Yet, a
comprehensive nomological network relating types of PM system uses to organizational capabilities
and performance is lacking. Furthermore, there is scant empirical evidence attesting to the explanatory
efficacy of PM systems as it relates to organizational performance. We view PM system uses through
the lenses of the Resource Orchestration Theory (ROT) and explore specific relationships of underlying
variables by relying on the Organizational Information Processing Theory (OIPT). Resting on the extant
literature, we identify two types of uses which include Diagnostic Use (the review of critical performance
variables in order to maintain, alter, or justify patterns in an organizational activity) and interactive use (a
forward-looking activity exemplified by active and frequent involvement of top management envisioning new ways to orchestrate organizational resources for competitive advantage) and relate them along
with their interaction (i.e., dynamic tension) to organizational capabilities. We further link capabilities to
target performance, which subsequently impacts organizational performance (operationalized through
both perceptual and objective financial performance measures). The nomological network is tested via a
cross sectional study (386 Italian firms) while the efficacy of PM systems to explain organizational performance is examined by using longitudinal panel data approaches over a 10 year period. There is sufficient
evidence to suggest that the use of PM systems leads to improved capabilities, which then impact performance. Contrary to the extant literature, however, we discovered that Diagnostic Use appears to be
the most constructive explanatory variable for capabilities. On the other hand, in light of a longitudinal study, we also uncovered that Diagnostic Use experienced depreciating returns as far as objective
financial measures are concerned. Also, when high levels of Diagnostic Use were coupled with low levels of Interactive Use, they produced the lowest levels of organizational capabilities. Conversely, high
levels of both types of PM system use generated extraordinary high levels of capabilities. There is sufficient evidence to suggest that organizations cannot rely merely on Diagnostic Use of PM systems. We
also learned that the effects of PM systems (measured via adaptation) fade unless high learning rates
are applied. We offer detailed recommendations for future research which have theoretical as well as
empirical implications.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
The operations management literature has been a strong proponent of metrics and respective PM systems for quite some time.
For instance, Bititci et al. (1997) offer a developmental guide to
∗ Corresponding author. Tel.: +1 979 845 2254/fax: +1 979 845 5653.
E-mail addresses: [email protected] (X. Koufteros),
[email protected] (A.J. Verghese), [email protected] (L. Lucianetti).
http://dx.doi.org/10.1016/j.jom.2014.06.003
0272-6963/© 2014 Elsevier B.V. All rights reserved.
construct integrated PM systems and Gunasekaran et al. (2004) proposed a framework for supply chain performance measurement.
Melnyk et al. (2004) discuss metrics and performance measurement while Neely (1999; 2005) furnished a treatise on the evolution
of performance measurement research in operations management.
Later on, Gunasekaran and Kobu (2007) provided a literature
review of performance measures and metrics in logistics and supply chain management. While the topic is popular, what is vividly
missing from the literature is a judicious examination of how companies actually use PM systems to orchestrate their responses to
314
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
organizational challenges and whether such uses do in fact enhance
operational, strategic, and external stakeholder related capabilities
and performance over time.
PM systems are integral to resource orchestration processes
and over the last three decades many organizations have invested
enormous amounts of capital, time, and effort developing and
implementing such systems. Undoubtedly, one of the most popular paradigms is the Balanced Scorecard, first documented and
articulated by Kaplan and Norton (1992). There are nevertheless
numerous other measurement frameworks (Bititci and Turner,
2000) that have been proposed and implemented–e.g. the Performance Measurement Matrix (Keegan et al., 1989), the Result and
Determinants Model (Fitzgerald et al., 1991), the Performance Pyramid (Lynch & Cross, 1992), and the Performance Prism (Neely et al.,
2002). These formal performance systems are utilized as mechanisms that enable organizations to orchestrate their resources more
effectively.
The Resource-Based Theory (RBT) has long argued that possessing valuable, rare, inimitable, and non-substitutable resources is
vital to firm sustained advantage (Hitt et al., 2011; Wowak et al.,
2013). But as Hansen et al. (2004, p. 1280) state “what a firm does
with its resources is at least as important as which resources it
possesses.” Sirmon et al. (2011) similarly note that while possession of resources is essential, the ability of a firm to “orchestrate”
its resources is more fundamental as the firm bids to prosecute
its strategic objectives. Orchestration of resources however is also
subject to top management who mobilizes the vision to direct and
use firm resources to achieve objectives (Chirico et al., 2011; Crook
et al., 2008). Resource Orchestration Theory (ROT) is an emerging
theoretical stream of work which rests on the conceptual work of
Sirmon et al. (2007) and Helfat et al. (2007). Hitt et al. (2011) argue
that “Resources orchestration is concerned with the actions leaders
take to facilitate efforts to effectively manage the firm’s resources”
(p. 64). Such actions include for instance the structuring of the firm’s
resource portfolio, bundling resources into capabilities, and leveraging the capabilities to create value to customers (Hitt et al., 2011;
Sirmon and Hitt, 2003; Sirmon et al., 2007). According to Hitt et al.
(2011) and Holcomb et al. (2009), while each action and its particular nuances are vital, the synchronization of actions can contribute
positively towards performance.
To manage each action and to synchronize the orchestration
of resources, leaders rely on mechanisms such as PM systems
which yield information regarding the functioning of their resource
portfolio and bundle of capabilities (Hitt et al., 2011). Such information is critical for leaders because it enables them to make crucial
adjustments to their resources and mobilize requisite resources as
conditions change. Melnyk et al. (2004) recognized the orchestrating role of PM systems in operations management and assert that
the “performance measurement system is ultimately responsible
for maintaining alignment and coordination” (p. 213). Operations
management leaders, for instance, need information regarding
inventory performance to decide whether additional space to house
inventory is necessary in order to pursue a new “same-day delivery”
strategy that demands high service levels, such as the expansive
Amazon.com Local Express Delivery strategy. Via the diagnostic
attributes of a PM system, managers can focus attention on issues
of strategic significance, monitor performance, and detect whether
the desired service level can be achieved given the current level and
blend of resources. In addition, the active and personal engagement
of the leadership with performance measurement processes can
serve as a catalyst in orchestrating the acquisition and bundling
of essential resources and capabilities to meet delivery targets.
Melnyk et al. (2004) highlight that metrics and respective PM systems serve “as essential links between strategy, execution, and
ultimate value creation” (p. 209). A PM system can also be characterized as a management control system that incorporates a
structured framework specifying key financial and non-financial
performance metrics. From a theoretical point of view, a PM system
can be described as an ambidextrous system because it incorporates both mechanistic and organic elements.
As an orchestration mechanism, the organization can use the PM
system to control organizational behavior (alike to a mechanistic
use) but on the other hand it can use it to promote organizational
innovation and strategic renewal (resembling an organic use). The
literature however tends to focus more on the “mechanistic” use of
PM systems while the more “organic” use is in general neglected.
The mechanistic use is coined as the diagnostic use in the literature and it is primarily responsible for furnishing information. From
an organizational information processing theory (OIPT) perspective
(Galbraith, 1974), diagnostic use is liable to reduce uncertainty. On
the other hand, the organic use can be described as interactive in
nature and is deployed by top management to enact debate and
reduce equivocality. Simons (1995) acknowledges the complementary nature of the two systems, but only few studies explicitly test
relationships between types of uses of PM systems (e.g., Widener,
2007) or specify and account for their interactions (e.g., Henri,
2006a). However, these authors operationalize organizational performance only via perceptual measures rather than objective and
longitudinal financial performance measures.
Underscoring the importance of PM systems, Kaplan and Norton
(2008) made a rather revealing statement suggesting that “We
believe that if you don’t measure progress toward an objective, you
cannot manage and improve it” (p. 7). From an ROT perspective,
firms that deploy their PM systems should be capable of shaping
capabilities to meet or exceed target performance. Highlighting and
motivating interest in adopting a PM system is, therefore, the claim
that organizations with a PM system outperform their counterparts
without a PM system (Davis & Albright, 2004; Crabtree & DeBusk,
2008). Unfortunately, this claim is debatable in part because there is
only a handful of published empirical research studies investigating
the claim (e.g., Ahn, 2001; Chenhall & Langfield-Smith, 1998; Henri,
2006a; Hoque & James, 2000; Ittner & Larcker, 1998; Chenhall,
2005; Widener, 2007) and in part because the extant empirical literature has reported mixed results regarding the effects
of PM system usage on organizational performance (Chenhall &
Langfield-Smith, 1998; Ittner et al., 2003a). Henri (2006a) surmises
that prior work examined the role of PM systems toward strategy implementation and strategy formulation, but concedes there
is scant empirical evidence attesting to the professed virtues of
such systems. The operations management literature is especially
devoid of studies in this respect and much of what is currently
published can be ascribed as contributions from the accounting
discipline.
Henri (2006a) submits that the specific relationship between PM
systems and strategy is ambiguous and at times contradictory and
attributes such results to differences in definitions and operationalizations of the variables. Pavlov and Bourne (2011) add that the
mechanism relating a PM system to performance is poorly understood. Henri (2006a) notes that prior research has specified and
tested direct links between PM system usage and performance but
the effects may actually be reflected instead by the capabilities that
PM systems incite as orchestration mechanisms. Pavlov and Bourne
(2011) argue that it has not been demonstrated exactly how PM systems are linked to performance, thus leaving the gap between PM
systems and performance still unresolved. As Pavlov and Bourne
state, the power of a PM system can be seen as significant and yet
somewhat opaque. In other words, there is still a “black box.”
Overall, there are three gaps in the literature which merit investigation. The first gap relates to the specific types of uses of PM
systems: For what purposes do organizations use PM systems? The
second and related gap pertains to the modality by which different
types of uses of PM systems impact performance: How do specific
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
types of uses of PM systems actually affect performance? Stated otherwise, there is a lack of a nomological network that relates specific
types of uses of PM systems to performance. Exploring the effectiveness of each type of use is important because organizations
can direct resources and effort on specific types of uses in order
to maximize returns. The third gap concerns the efficacy of PM systems to explain variation in performance: Does the implementation
of PM systems benefit the organization financially over time? Both the
operations management and the accounting literatures are lacking of empirical studies that demonstrate the financial impact of
PM system implementations from a longitudinal perspective. Davis
and Albright (2004) examined the impact of PM systems on performance (i.e., key financial measures pertinent only to financial
institutions) but the study was constrained to a single firm (i.e., a
bank) and several of its branches, and assessed change in performance only across two periods.
This manuscript tackles the first two gaps in the literature by
identifying salient types of uses and by constructing a nomological
network of constructs. Then, based on literature (e.g., Henri, 2006a)
that posits that the effects of PM systems on performance are communicated via the capabilities they engender, it examines how
each type of use of a PM system impacts organizational capabilities across three dimensions which include operational capability,
strategic management capability, and external stakeholder relations capability. It articulates simple effects as well as interactions
amongst types of PM system usage, while controlling for an extensive list of demographics and tangible and intangible resources. In
essence, we are able to disentangle the orchestrating role of PM
systems from the mere possession of resources. Furthermore, the
relationship between organizational capabilities and target performance is tested. Finally, the link between target performance and
financial performance is tested utilizing both subjective and objective performance measures. To tackle the third gap in the literature,
the impact on four financial ratios is studied over time (before
and after implementation of a respective PM system) using several
methods. Analyses rely on primary data and secondary data over a
ten year time horizon. The longitudinal effects of PM system usage
are assessed using panel data analysis approaches and Stata12.
2. Literature review & theory development
A review of the operations and supply chain management literature suggests that there is no shortage of discussion regarding
metrics and PM systems (e.g., Neely, 1999; Bourne et al., 2000;
Bullinger et al., 2002; Melnyk et al., 2004; Neely, 2005; Hult
et al., 2008; Chia et al., 2009; Lehtinen & Ahola, 2010; Taylor &
Taylor, 2013; Garengo & Sharma, 2014). Neither there is scarcity of
frameworks for measuring strategic, tactical, and operational performance at the plant level or in a supply chain (Bititci et al., 1997;
Bititci & Turner, 2000; Neely et al., 2000; Gunasekaran et al., 2001;
Gunasekaran et al., 2004; Gunasekaran and Kobu, 2007; Estampe
et al., 2013). The implied assumption is made however that PM
systems are by their very nature conducive towards better performance because they can be used to benchmark operations against
targets and abet the organization orchestrate resources to enhance
future performance. Metrics and performance measurement are
seen as critical elements in translating an organization’s mission, or
strategy, into reality and for maintaining alignment and coordination (Melnyk et al., 2004). Thus, PM systems are typically ascribed
with positive evaluations.
The majority of studies that have surfaced in the literature
assess PM system effectiveness in relation to overall organizational performance (Braam & Nijssen, 2004; Crabtree & DeBusk,
2008; Davis & Albright, 2004; Ittner, Larcker, & Randall, 2003b),
thereby assuming a direct association between the PM system and
315
performance. In this respect, Stede et al. (2006) found that regardless of strategy, organizations with more extensive PM systems,
especially those that included objective and subjective nonfinancial measures, have better overall performance. However, the
literature suggests that the findings are inconsistent (Bourne et al.,
2013; de Leeuw & van den Berg, 2011; Hamilton & Chervany, 1981).
At the organizational level, Henri (2006a) and Widener (2007)
advocate that a more in-depth understanding of the engendering role of PM systems on organizational capabilities may help to
resolve some of the ambiguous findings. Recently, Grafton et al.
(2010) explored the process by which PM systems impact the way
managers orchestrate their resources via mobilization to achieve
competitive advantage and contribute to performance outcomes.
Relying on RBT, Grafton et al. (2010) argue that the use of performance measurement information for feedback and feed-forward
control influences the extent to which an organization is able to
exploit and identify its strategic capabilities, respectively. Modelling capabilities as a mediating variable would allow researchers
to explore the processes by which PM systems enhance organizational outcomes.
Also, the extant literature describes several PM system types
of uses (e.g., coordination, monitoring, diagnosis, problem solving,
legitimization, scorekeeping, focusing attention). Simons (1994)
work describing types of uses or purposes of PM systems has
been distinctively influential in the literature. Simons (1994) effectively covers the domain of PM system types or styles of uses
(Bisbe & Otley, 2004), yet he encapsulates them efficiently via only
two categories (i.e., diagnostic and interactive uses), which rely
on respective diagnostic systems and interactive systems. According to Simons (1994), diagnostic systems are “formal feedback
systems used to monitor organizational outcomes and correct deviations from pre-set standards of performance” (p. 170). Diagnostic
systems are in essence information systems that reduce uncertainty and thus aid strategy implementation (Bisbe & Otley, 2004).
Interactive systems are used “by top managers to regularly and
personally involve themselves in the decision activities of subordinates” (Simons, 1994, p. 171). Bisbe and Otley (2004), resting on the
work of Simons (1994), suggest that interactive systems help organizations adapt to changes in the environment by orchestrating
organizational responses.
The modality by which PM systems enable resource orchestration can be explained via the theoretical lenses of the organizational
information processing theory (OIPT) (Daft & Lengel, 1986;
Galbraith, 1974). As Melnyk et al. (2004) state, “An information
processing perspective offers yet another potentially rewarding
way to look at metrics” (p. 214). Choi et al. (2013) state that
PM systems “function as a framework that organizes the firm’s
information environment around its strategy” (p. 106). According to Grosswiele et al. (2012), “Performance measurement aims
to provide decision makers with information that enables them
to take effective actions and evaluate whether a company is progressing in line with its strategy” (p. 1017). Bisbe and Malagueño
(2012) suggest that PM systems “. . .support the decision-making
processes of an organization by gathering, processing, and analyzing quantified information about its performance, and presenting
it in the form of a succinct overview” (p. 297). Organizations however do not always have the requisite information and thus they
may face uncertainty and, sometimes, even when they do possess
information it may have multiple and conflicting interpretations
which creates equivocality (Daft & Lengel, 1986). In order to tackle
facets of both uncertainty and equivocality, organizations turned to
PM systems which can furnish relevant information and engender
debate and enactment of organizational action.
This study aims to examine how different types of uses of PM
system can fuel organizational capabilities leading to improved
organizational performance. Specifically, this research focuses on
316
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Diagnosc Use
Subjecve
Financial
Performance
Strategic
Management
Capability
ROTA
Dynamic Tension
(Diagnosc X Interacve)
Operaonal
Capability
Target
Performance
ROA
ROE
Interacve Use
External
Stakeholder
Relaons
Capability
EBIT
Fig. 1. Research Framework Note: Effects on ultimate dependent variables are controlled for industry, firm size, geographic diversification, R&D expenditures, Industrial
Patents & Intellectual Property Rights, Plant & Machinery, and Industrial & Commercial Equipment. These effects are displayed in Table 4.
the orchestrating roles of PM systems: “diagnostic use” and “interactive use.” According to Henri (2006a), “These two types of use
work simultaneously but for different purposes. Collectively, however, their power lies in the tension generated by their balanced
use, which simultaneously reflects a notion of competition and
complementarity” (p. 531). Fig. 1 displays the research framework.
2.1. Diagnostic PM system use
Diagnostic Use can be defined as a set of formalized procedures
that use information to maintain or alter patterns in an organizational activity (Henri, 2006a). These procedures may include
reporting systems, monitoring tools, and performance results being
communicated to all staff. Also, such procedures may describe a
conceptual model for understanding drivers of organizational success or how excellence may be attained (i.e., success map or strategy
map), or evaluate critical success factors to monitor performance
and specific targets. Widener (2007) states that diagnostic use
motivates employees not only to perform but also to align their
behavior with organizational goals. Diagnostic use comes in many
flavors however.
In this paper, we focus on three rather comprehensive diagnostic uses which are tasked to furnish information and aid resource
orchestration: “monitoring”, “focusing attention”, and “legitimization.” Monitoring enables managers to track performance against a
plan and can identify what, if any, is going awry (and needs correction) (Henri, 2006a,b). The monitoring element of diagnostic use
enables managers to benchmark against targets (Widener, 2007).
For instance, top management may set goals for customer service
level. Monitoring can help them establish whether goals are met
and pinpoint where performance is lacking. Focusing attention generates information which is pertinent in directing organizational
members on issues of strategic significance. It draws interest on
common issues, critical success factors, and integration. Widener
(2007) further notes that focusing attention sets boundaries or
constraints on employee behavior, alike to the boundary system
advocated by Simons et al. (2000). By focusing attention on critical
success factors, such as customer service level top management can
promote unity of effort across organizational units such as purchasing, operations, and distribution since many constituents influence
the capability of the organization to obtain and sustain high levels
of customer service.
Using a PM system for legitimization has the intention of justifying and qualifying actions and ensuring they are aligned with
strategy. In this sense, the legitimization process generates information to enhance the rightfulness and credibility of decisions
(Henri, 2006b) to heighten the legitimacy of current and future
organizational action. Legitimization can be seen as a political tool
(Henri, 2006b) that management can use to assess whether a specific action was judicious. One of the managers we interviewed
for this study commented that the methodological validity or the
legitimization of the planning process can only be ascertained if
it is verifiable ex post. Demonstrating that decisions are credible
can positively impact stakeholders’ support and heighten trust in
management, which should positively impact the ability to meet or
exceed performance targets. Top management can use the PM system to obtain data regarding customer service level to legitimize
strategic choices and garner employee support and commitment.
The three specific purposes are in essence the nuances that
reflect the diagnostic component of PM system use. Diagnostic Use
can be thought of as a higher-order variable composed of monitoring, focusing attention, and legitimization. Monitoring ensures
that managers know when standards are not being met and thus
can make corrections; focusing attention warrants that all levels
of the organization remain on the same page and in communication about where the organization is headed while legitimization
attests that those decisions are justified and are well-grounded in
the organization’s business strategy.
Diagnostic use can be used as a tool to quantify actions (Neely
et al., 1995) across a variety of metrics. Henri (2006a) suggests that
these metrics can assume a variety of orientations such as being
financial or non-financial, internal or external, short or long term as
well as ex post or ex ante (Franco-Santos et al., 2012). The diagnostic
use of PM system also “reflects two important features associated
with mechanistic controls: (i) tight control of operations and strategies, and (ii) highly structured channels of communication and
restricted flows of information” (Henri, 2006a, p.7). This formal use
of PM system provides a mechanistic approach to decision making which furnishes signals when issues such as productivity and
efficiency have fallen (Henri, 2006a) outside expectations.
However, as a mechanistic control mechanism, diagnostic use,
as Henri (2006a) states, has been associated with several dysfunctional behaviors in terms of smoothing, biasing, focusing, filtering,
and illegal acts (Birnberg et al., 1983). “These distortions constitute
defensive routines that aim to reduce potential embarrassment or
threat, or to improve personal interest” (Henri, 2006a, p.7). Given
some of the potential for adverse effects, Henri (2006a) in fact
labels diagnostic use as a negative force and even hypothesizes a
negative relationship between Diagnostic Use and organizational
capabilities. It is noted however that Henri (2006a) operationalized
Diagnostic Use by measures reflecting only monitoring.
Unlike Henri (2006a), we suspect diagnostic use may have a
net positive contribution especially in the presence of high interactive use. Diagnostic use is critical because it furnishes information
regarding deviations, positive or negative, against expectations.
While firms tend to focus on negative deviations rather than
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
positive deviations, the PM system itself is quite adept and furnishes valuable information that can reduce uncertainty and thus
improve organizational capabilities. A negative deviation prompts
action to steer the organizational energy and resources towards
the targets while a positive deviation can be utilized to reinforce
organizational efforts. PM systems via diagnostic use can also furnish information that can be consumed to improve relationships
with stakeholders. Information generated through PM systems
can be pertinent to customers who evaluate suppliers for future
business opportunities. Such information may also be germane to
the market, which determines the valuation of the firm and its
attractiveness to potential investors. Agreeably, diagnostic use may
somewhat limit the role of a PM system to a measurement tool
(Henri, 2006a), but even as a tool it is nevertheless essential because
it plays a significant role in orchestrating resources, especially when
deployed concurrently with interactive use.
2.2. Interactive PM system use
The interactive use of PM systems tends to have a more developmental role and represents a positive force (Henri, 2006a)
employed to expand opportunity seeking and learning throughout
the organization. Using a PM system interactively forces dialogue
and stimulates the development of new ideas and initiatives (Henri,
2006a; Widener, 2007). According to Henri (2006a), when a PM
system is used interactively, the information generated is a recurrent and important agenda item for senior leadership; frequent and
regular interest is fostered throughout the organization; data are
discussed and interpreted among managers from different hierarchical levels; and “continual challenge and debate occur concerning
data, assumptions, and action plans” (Henri, 2006a, p.5).
Bisbe and Otley (2004) state that senior managers, through their
personal involvement and commitment, can use interactive systems to orchestrate responses to environmental perturbations and
craft subsequent organizational renewal. Employing a PM system
from an interactive use perspective can generate pressure across
organizational levels and motivate further information gathering
while stimulating debate. Interactive use embraces institutional
dialogue and debate and stimulates information exchange and
communication (Henri, 2006a), which play a vital role in reducing equivocality. In addition, interactive PM system use equips
the organization with two attributes. It expands the organization’s information processing capacity and engenders interaction
amongst constituents (Henri, 2006a), leading to more eloquent
orchestration of resources. For example, top management can use
the PM system to critically evaluate discrepancies between targeted and actual customer service levels. Top management gets
intimately involved with the use of the PM system to analyse
root causes and orchestrate necessary resources to achieve targets.
Henri (2006a) labels interactive use as a “positive” force suggesting that interactive use promotes opportunity seeking behavior and
learning across organizational levels. According to Henri (2006a),
the interactive use of a PM system expands its role to a strategic
management tool.
2.3. Dynamic tension – diagnostic and interactive PM system
roles
PM systems have two complementary and interdependent
roles (Mundy, 2010) which are information based: (1) to exert
control over the attainment of organizational goals and (2) to
enable employees to search for opportunities and solve emergent problems via top management commitment and involvement
(Ahrens & Chapman, 2004; Chenhall & Morris, 1995; Simons, 1995;
Zimmerman, 2000). These roles necessitate a balance between
taking actions congruent with the organization’s goals while also
317
giving employees sufficient autonomy to make decisions (Roberts,
1990; Sprinkle, 2003). When combined, controlling and enabling
uses of a PM system create dynamic tension (Simons, 1994) which
harvests unique organizational capabilities and competitive advantages through organizational dialogue, mutual understanding and
better direction (Henri, 2006a). An organization’s inability to balance different uses of PM systems is associated with slower decision
making, wasted resources, instability and, ultimately, lower performance (Bisbe et al., 2007; Henri, 2006a).
Our theoretical model investigates whether the joint use of PM
systems for diagnostic and interactive purposes creates dynamic
tension that reflects complementarities (focusing on intended and
emergent strategies). According to Henri (2006a, p. 533), “. . . the
diagnostic use of PM systems may represent a mechanistic control tool used to track, review, draw attention, and support the
achievement of predictable goals. . .” via the provision of information. On the other hand, “. . .interactive use may be considered as
an organic control system supporting the emergence of communication processes and the mutual adjustment of organizational
actors. . .” which are used to reduce equivocality and stimulate
change. Also, diagnostic use of a PM system can be seen as a pure
measurement instrument, while interactive PM system use can
elevate its role to a strategic management tool (Kaplan & Norton,
2001).
Henri’s (2006a) empirical work concludes that there is evidence
to suggest that using the two levers of control generates tension
which then positively affects performance. Henri tested simple
effects and modelled tension as an interaction effect but Henri’s
operationalization of interactive use does not adhere to Simons
(1994) description. Henri (2006a) operationalizes interactive use
via measures of focusing attention while Simons (1994) states that
interactive use is described by nuances such as top management
commitment and engagement. In contrast to Henri, Widener (2007)
operationalized diagnostic use via measures of monitoring and
focusing attention while she operationalized interactive use via
measures of top management engagement. Widener also suggests
that the two levers are complementary and based on empirical evidence claims that the full benefits from a PM system can only be
derived when they are used from both diagnostic and interactive
perspectives. Widener (2007) however neither specified nor tested
interactions between the two systems.
2.4. Hypotheses: PM system USE→ ORGANIZATIONAL
CAPABILITIES
While much of the PM systems literature posits direct effects
on performance, this perspective has been challenged and empirical evidence is scarce. In essence, recent literature supports the
notion that diagnostic and interactive uses of PM systems advance
organizational capabilities which subsequently help the organization meet its targets. Organizational capabilities “refer to a firm’s
capacity to deploy Resources, usually in combination, using organizational processes, to effect a desired end. . ..Capabilities are based
on developing, carrying, and exchanging information through the
firm’s human capital” (Amit & Schoemaker, 1993, p. 35). The
PM system may affect organizational capabilities by increasing
communication between employees or encouraging employees to
analyze how resources should be combined to solve problems.
A PM system can encourage improvements across several capabilities such as in the areas of strategic management, operations,
and stakeholder relationships. Hence, varied PM system usage (i.e.,
diagnostic and interactive uses) may have an effect on firm performance via its impact on organizational capabilities and meeting
organizational targets.
Effective strategic management involves necessarily a variety of critical orchestrating activities. Strategic management
318
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
encompasses activities such as the prioritization of strategic actions
and objectives, obtaining information and feedback regarding the
existing strategy and forthcoming strategy, the alignment of organizational processes with the intended strategy, and the allocation
of budgets and capital expenditures among others (Porter, 1991;
Simons, 1991). These activities rely on feedback obtained from
diagnostic systems and direction from top management managers
who interactively obtain information, and discuss, challenge, and
debate courses of action with other organizational members across
levels. The information obtained from diagnostic and interactive
PM system uses can help organizations reduce uncertainty and ease
equivocality, enhancing strategic management capability. Without information derived from performance systems, organizations
cannot assess their current performance vis-à-vis their intended
strategy and may have a difficult time charting a new course of
action or direction without knowing the facts and without significant debate that can emerge from interactive use. The use of both
systems is necessary as the organization has to balance the needs
for control and renewal. We expect that PM system uses will have a
positive effect on strategic management capability but we also posit
that the interaction between the two types of uses will explain additional variation, above and beyond what the individual uses may
contribute.
– H1a : Diagnostic Use will have a positive impact on Strategic Management Capability.
– H1b : Interactive Use will have a positive impact on Strategic Management Capability.
– H1c :The Dynamic Tension between Diagnostic Use and Interactive Use will have a positive impact on Strategic Management
Capability.
Organizations also rely on their operational capabilities to garner advantage against competitors. Competitive advantage can
emerge for instance from more productive, innovative, and integrated processes, and higher employee productivity. To enhance
such processes, however, the organization relies on information
furnished by both diagnostic and interactive systems and the
tension created when both systems are used concurrently. For
instance, information regarding productivity or quality has to be
collected and shared in order for organizational constituents to take
corrective measures when necessary or for promotion of processes
that prove to be exemplary. Also, in order to integrate processes
and orchestrate resources, organizations need to discuss the value
proposition, identify costs and benefits, and set direction, all of
which need information and deliberation across organizational
members along with top management involvement. We expect that
diagnostic and interactive PM system uses along with the interaction between the two uses will have a boosting effect on operational
capabilities.
H2a :Diagnostic Use will have a positive impact on Operational
Capability.
H2b : Interactive Use will have a positive impact on Operational
Capability.
H2c : The Dynamic Tension between Diagnostic Use and Interactive
Use will have a positive impact on Operational Capability.
Organizations rely on external stakeholders for business, affirmation, and resources (Freeman, 1984; Donaldson & Preston,
1995, Agle et al., 1999; Berman et al., 1999). Relationships with
such stakeholders depend however on information. For instance,
relationships with customers depend on timely and accurate information generated via performance monitoring systems. Customers
can demand information regarding quality defects or costs. Similarly, suppliers seek feedback information furnished by monitoring
systems concerning their performance. Challenges and opportunities regarding customers, suppliers, and regulators require debate
and guidance by top management, which can interactively take an
interest to generate solutions and resolve equivocality. However,
when the two systems are used together, there is even greater
potential to improve external stakeholder relations because the
two systems can tackle uncertainty and equivocality only when
used jointly. Thus, we hypothesize that the two systems and their
interaction will have a positive effect on external stakeholder relations.
H3a : Diagnostic Use will have a positive impact on External Stakeholder Relations Capability.
H3b : Interactive Use will have a positive impact on External Stakeholder Relations Capability.
H3c : The Dynamic Tension between Diagnostic Use and Interactive
Use will have a positive impact on External Stakeholder Relations
Capability.
2.5. Hypotheses: ORGANIZATIONAL CAPABILITIES→TARGET
PERFORMANCE
Organizations capable of effective strategic management can
direct resources towards the attainment of organizational targets.
First, managing feedback regarding existing strategy can help the
organization align its operations with the respective strategy and
thus properly equip the organization to meet its goals. Second,
strategic management efforts can help the organization decide
on its strategic objectives and allocate resources to achieve those
objectives. In this form, strategic management is a protagonist but it
also plays an enabling role. Effective strategic management should
impact the attainment of organizational goals positively.
H4 : Strategic Management Capability has a positive effect on
Target Performance.
Organizations depend on their processes to add value and derive
healthy returns. These processes have to bolster the organizational operational capabilities which are responsible to propel the
organization to meet or exceed its targets. Operational capabilities can be reflected by a variety of nuances such as enhanced
productive processes, improved employee performance, integrated
processes, and innovation in working processes (Simons, 1991).
For example, improved employee performance, more integrated
processes, and more innovation in working processes can help customers derive better value from the relationship with the firm and
thus the organization can reap such benefits to meet its targets
regarding customer relations. Similarly, delivery performance can
be enhanced through operational capabilities and thus the organization can subsequently meet its operational performance targets.
Consequently, we expect that higher levels of operational capabilities will enable the organization to meet its targets more effectively.
H5 : Operational Capability has a positive effect on Target Performance.
Organizations depend on external stakeholders for business
opportunities, support, resources, acknowledgement, validation,
and rulings (Donaldson & Preston, 1995; Agle et al., 1999; Berman
et al., 1999). External stakeholders include customers, suppliers,
regulators or other governmental institutions as well as the market at large. Customers for instance provide business opportunities
while suppliers are assuming an ever growing importance in assuring the firm’s value chain. Organizations can improve profitability
by increasing revenues and/or cutting costs. Improving relationships with customers can potentially increase revenues while
working closer with suppliers can trim costs. Organizations cannot
ignore external stakeholders because they can help the organization meet or exceed its targets (Berman et al., 1999; Margolis &
Walsh, 2003).
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
H6 : External Stakeholder Relations Capability has a positive
effect on Target Performance.
2.6. Hypotheses: MEETING TARGET PERFORMANCE→FIRM
PERFORMANCE
When organizations meet their customer relations targets,
this generally manifests an ability to keep customers happy and
improve market share. Ultimately, this translates into higher
returns in terms of sales growth and profit growth, among other
financial measures. Similarly, when organizations meet operational
performance targets, they enjoy productivity improvements or
shorter lead times which then lead to better returns on investment
and higher profit to sales ratios. We expect that these effects will be
robust across both subjective as well as objective financial ratios.
We thus hypothesize that firms that meet their performance targets
will demonstrate better performance.
H7 : Target Performance will have a positive effect on firm performance.
2.7. Hypotheses: LONGITUDINAL EFFECTS
PM systems are expensive to implement and maintain. The
premise, however, is that the use of a PM system generates positive
financial returns over time. PM systems furnish information and
engender more effective resource orchestration which is critical for
the organization’s current well-being and future prosperity. Organizations obtain and process information for operational, relational,
and strategic use. We expect that firms that use a PM system would
outperform firms that do not deploy a PM system. They should
have better returns on assets and enjoy higher earnings. From a
static perspective, the performance of the firm can be assessed by
examining financial ratios before and after implementation. From
a dynamic view, performance improves over time as organizations
learn to adapt to a new system. The dynamic perspective acknowledges that time plays a vital role.
H8 : PM system use will have a positive effect on firm financial
performance.
3. Cross sectional study
Our first study can be described as cross sectional in nature.
We illustrate below the operationalization of constructs, research
design and research methods, and provide results based on subjective and objective data obtained for 2010. Objective data for
both the cross-sectional and longitudinal analyses were obtained
from Aida - Bureau van Dijk (https://aida.bvdinfo.com/), a source
of financial performance data and other information pertaining to
firms operating in Italy and many other countries.
3.1. Operationalization of constructs
In order to operationalize the variables, the researchers sought
to identify existing scales that can be employed or scales that can
be adapted to the specific context. Where existing scales were
not identified, the researchers relied on the extant literature to
specify the content domain of each construct. The review also
included an investigation of contextual and organizational factors that may affect performance. Contextual factors may involve
primary industry affiliation, firm size, and geographic ownership
diversification while organizational factors may embrace measures
of the resources the firm possesses.
To assure that the questions could be correctly understood by
respondents and easily answered by them, the initial survey questionnaire was carefully pre-tested. Thus, a preliminary draft of the
319
questionnaire was discussed with academic scholars to assess clarity, simplicity, and content validity. Afterwards, a pilot study was
also conducted with a group of accounting managers and controllers of six large organizations, where their inputs were used to
improve the clarity, comprehensiveness, and relevance of the survey questions. Finally, the emerged survey was reviewed once more
by a panel of three academics. The final survey items are discussed
below and appear in Table 2.
Diagnostic use was specified as a second-order construct and
was operationalized via three first-order variables which include
monitoring, focusing attention, and legitimization. All three rely
on measures developed by Henri (2006a,b) which he adapted from
Vandenbosch (1999). Monitoring is reflected via four manifest variables. Focusing Attention is operationalized by seven observed
variables. Legitimization includes nine observed items. Interactive
use relies for content on the work of Widener (2007). Strategic
Management Capability is a new scale and consists of six measures
that address whether the PM system in the organization supports
strategic management activities. These measures draw on the work
of Simons (1991) and his work on management control systems.
Operational Capability is also a new scale and was captured via
five indicators addressing productivity, employee performance,
innovation of working processes, and development of integrated
solutions. External Stakeholder Relations Capability is a multi-item
scale and relies on the work of Cousins et al. (2008) and Mahama
(2006). We operationalized Target Performance via a five item scale
which drew items from Ittner et al. (2003b). It addresses the extent
to which the organization meets the performance targets across a
variety of categories. Finally, Subjective Financial Performance is
measured using four financial indicators relative to industry average.
3.2. Research design
The sample population consisted of 1000 organizations operating in Italy and randomly selected from Amadeus - Bureau Van
Dijk (a database of public and private firms which include Italian firms as well as multinational firms filing separately for their
Italian operations). An appropriate sample of organizations covering different industry sectors and organizational characteristics
has been selected to maximize the generalizability of the findings. We contacted each firm’s management directly to select a
list of companies prepared to cooperate with the research. We
then identified the key figures in each company who would be
competent to fill in the survey. Names and e-mail addresses of targeted respondents were obtained during telephone contacts. The
research design enabled the researchers to target respondents that
possess substantive organizational level knowledge and specific
knowledge as it pertains to PM systems. We targeted high level
executives (i.e., CEOs, CFOs) as well as controllers and managers
(e.g., managing directors, operations managers). Given the positional levels of these individuals, we are confident that the target
population can serve as excellent respondents both in reference to
explanatory variables as well as performance variables.
The research has been carried out through a survey questionnaire using fax and e-mail during 2010. To increase the survey
response rate, we pre-notified specific respondents by phone to
seek their input regarding their availability to work on the survey.
This approach probably led to more involvement and commitment
by the respondents from the beginning of the project (Baldauf
et al., 1999). The survey instrument was preceded by an introductory letter clarifying the purposes and objectives of the entire
research project. Managers were also promised an overall PM system benchmark study to compare their responses to those of other
participating organizations. Respondents were alerted to respond
to the survey on three subsequent occasions.
320
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 1
Demographics.
Respondent’s job title
CEO
CFO
Controller
Managing Director
Operations Manager
HR Manager
Finance Manager
Others
Frequency
Percentage
Cumulative
percent
19
57
185
34
58
17
6
10
4.90
14.80
47.90
8.80
15.00
4.40
1.60
2.60
4.90
19.70
67.60
76.40
91.40
95.80
97.40
100.00
246
140
63.70
36.30
63.70
100.00
202
184
52.30
47.70
52.30
100.00
221
69
18
13
25
28
12
57.30
17.90
4.70
3.40
6.50
7.30
3.10
57.30
75.10
79.80
83.20
89.60
96.90
100.00
12
22
51
59
61
53
120
8
3.20
5.80
13.50
15.60
16.10
14.10
30.00
2.10
3.20
9.00
22.50
38.10
54.20
68.30
100.00
Industry
Manufacturing
Service
Geographic Diversification
Italian
Multinational
Employee Size
n < 500
500 ≤ n ≤ 999
1000 ≤ n ≤ 1499
1500 ≤ n ≤ ≤1999
2000 ≤ n ≤ 2999
3000 ≤ n ≤ 9999
n ≥ 10000
Organizational Age
Age ≤ 5
6 ≤ Age ≤ 10
11 ≤ Age ≤ 20
21 ≤ Age ≤ 30
31 ≤ Age ≤ 40
41 ≤ Age ≤ 50
Age ≥ 51
Not reported
To address the research questions and hypotheses, we obtained
useful responses from a cross-section of 386 firms operating in Italy,
which represents close to a 39% effective response rate. We note
that we obtained surveys from a total of 401 firms but responses
from 15 firms were discarded due to excessive missing data. There
were very rare instances of missing values for the retained firms
and those appeared to be missing at random; missing values were
mean replaced.
Demographics appear in Table 1. Roughly 20% of the respondents are C-level executives while about 48% of respondents
are controllers. Operations managers and managing directors
accounted for another 23% of participants. As far as firm size is
concerned, small, medium, and large organizations are represented
and reflect general industrial demographics in Italy. Around 57.30%
of the firms are considered small with employment levels below
500 while 17% of the firms have a size greater than 2,000. About
64% of the sample represents manufacturing firms and 36% of the
responding firms are service oriented. There is almost equal representation between pure Italian firms (52%) vis-à-vis multinationals
(48%). The firms tend to be older with relatively few firms (9%)
dating less than 10 years old. Overall, the demographics suggest a
broad representation of firms operating in Italy.
In order to identify whether the responding firms differ from
the population of firms operating in Italy, we randomly selected a
blend of another 581 firms operating in Italy which shared a similar
proportion of manufacturing and service firms as our sample, as
well as domestic and multinational firms operating in Italy, and
compared the two groups across many dimensions. Based on t-tests
(a = .05), the analysis suggests no differences in terms of employee
size, sales, profitability, and total assets. Thus, there is no evidence
to suggest that the sample is biased against the population of firms
operating in Italy.
Our research design included a variety of statistical and separation approaches (Craighead et al., 2011) to minimize the risk
of common method variance (CMV). In terms of statistical remedies, Harman’s single factor test using CFA failed to support a single
factor. We also used a marker variable technique (Malhotra et al.,
2006). According to Craighead et al. (2011), the marker variable
technique calls for the inclusion of an additional variable in the
study which is theoretically unrelated to at least one other variable
of interest (Harrison et al., 1996). Specifically, the marker variable
(i.e., My company has an operating top management philosophy
of a growth strategy primarily through external financing [borrowing, capital issues, etc.]; rated on a seven point scale where
1 = to a very great extent and 7 = not at all) exhibited low correlations with all focal variables: Diagnostic Use (−.008), Interactive
Use (.081), Strategic Management Capability (.010), Operational
Capability (.033), External Stakeholder Relations Capability (.094)
Target Performance (−.105), and Subjective Financial Performance
(−.094). We also lessened the risk of CMV via methodological
separation (Craighead et al., 2011). The independent and dependent perceptual variables were measured using different scales and
different formats. Subjective financial performance was operationalized using a different scale than target performance and all other
variables in the model. Importantly, the perceptual data set is
relying on a cross-sectional survey while objective financial performance was obtained from secondary sources for a ten year period,
including time periods both before and after the collection of perceptual data. Given the applied remedies, CMV is not expected to
impact our specified relationships.
3.3. Research methods and results
To address the measurement model through confirmatory factor analysis, we used a covariance matrix as input and tested it
via Mplus (Table 2). The measurement model includes first-order
factors as well as a second-order factor which operationalizes Diagnostic Use. To ascertain the efficacy of the second-order model
for Diagnostic Use, we examined four models within the context of the entire measurement model, relying on the guidance of
Koufteros et al. (2009). Details about the methods of examining for
the second-order model specification appear in Appendix A.
The overall CFA model, which includes the second-order specification for Diagnostic Use, has acceptable model fit as indicated
by the fit statistics (2 = 1985.576, df = 1090, 2 /df = 1.82, CFI = 0.93,
NNFI = 0.92, RMSEA = 0.046, and 90% CI of RMSEA = (.043, .049).
All item-factor loadings are greater than 0.50, and are statistically significant at the 0.01 level. We assessed discriminant validity
by comparing the squared correlation between each pair of firstorder constructs and their respective average variance extracted
(AVE) (Table 2) (Koufteros, 1999). A squared correlation greater
than individual AVEs provides evidence to suggest the absence
of discriminant validity. The highest squared correlation, which
can be derived from Table 3, is 0.40 and it is significantly lower
than the respective AVEs of the two specific first-order constructs,
i.e., Strategic Management Capability (AVE = 0.57) and Operational
Capability (AVE = 0.63). The composite reliabilities (Table 2) for the
focal constructs are greater than the threshold of 0.70. Overall, the
constructs appear to be reliable and valid.
We subsequently specified a structural model to partially examine the hypotheses advanced earlier. Before testing the structural
model, however, we examined the distribution of each variable
via measures of kurtosis and skewness, along with visual inspections. Each variable appeared to have an approximately normal
distribution. Also, the scores for Diagnostic Use and Interactive Use
were mean-centered before creating the interaction term which
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
321
Table 2
Measurement model and descriptive statistics.
Latent Variables and Respective Manifest Items
Mean/item
Standard
Deviation
Standardized
Coefficient
T-value
Response format for the following three constructs: Using the scale where 1 = To a very great extent, 2 = To a great extent, 3 = To some extent, 4 = To a little extent,
5 = To a very little extent, and 6 = Not at all, please rate the extent to which your top management team currently uses performance measures to:
Monitoring, Composite Reliability (CR) = .82, AVE = .54
Track progress towards goals
Review key measures
Monitor results
Compare outcomes to expectations
2.16
2.73
1.83
1.96
.923
1.007
.793
.888
.86
.72
.68
.67
39.39
24.69
21.48
20.57
3.03
2.86
2.44
2.98
2.76
2.92
2.55
1.002
.98
.971
1,088
1.030
1.018
1.016
.75
.80
.78
.67
.75
.66
.74
29.69
35.57
33.20
21.40
29.83
19.94
28.08
2.67
2.72
2.72
2.81
2.57
2.78
2.51
2.72
2.92
.928
.989
.901
.920
.866
.936
.978
.953
.970
.71
.57
.63
.75
.80
.76
.69
.69
.70
24.65
15.25
18.58
29.63
37.54
30.76
23.56
23.28
24.01
Focusing Attention, CR = .89, AVE = .54
Integrate the organisation- i.e. tie the organisation together
Enable the organisation to focus on common issues
Enable the organisation to focus on your critical success factors
Develop a common vocabulary in the organisation
Provide a common view of the organisation
Enable discussion in meetings of superiors, subordinates and peers
Enable continual challenge and debate underlying results, assumptions and action plans
Legitimization, CR = .90, AVE = .50
Increase your understanding of the business
Justify decisions
Verify assumptions
Maintain your perspectives
Support your actions
Reinforce your beliefs
Stay close to the business
Increase your focus
Validate your point of view
Response format for the following construct: In my organisation, senior managers have strong commitment to. . .Use the scale where 1 = To a very great extent, 2 = To
a great extent, 3 = To some extent, 4 = To a little extent, 5= To a very little extent, and 6 = Not at all.
Interactive Use, CR = .90, AVE = .64
Reviewing organisational performance based on our performance measurement systems
Using performance measures to analyse root problems
Using performance measures to achieve targets
Using performance measures to make decisions
Using performance measures to get information to support decision making
2.81
2.50
2.31
2.40
2.40
.977
.915
.871
.878
.865
.74
.87
.76
.82
.82
27.91
51.89
30.48
38.87
38.97
Response format for the following three constructs: The performance measurement systems in my organisation . . .Use the scale where 1 = To a very great extent,
2 = To a great extent, 3 = To some extent, 4 = To a little extent, 5 = To a very little extent, and 6 = Not at all.
Strategic Management Capability, CR = .89, AVE = .57
Support the achievement of key strategic objectives
Improve the prioritisation of actions, projects and objectives
Give feedback on the company strategy & its strategic direction
Give feedback on operational processes
Improve the alignment of strategy and operations
Enhance negotiation of capital expenditure, budget allocation and financial support to projects
2.36
2.45
2.48
2.61
2.69
2.80
.861
.911
.889
.906
.906
.940
.75
.80
.77
.72
.84
.62
30.01
37.10
32.09
26.49
45.09
17.97
2.91
2.89
2.66
2.76
2.97
.880
.907
.838
.816
.876
.85
.79
.84
.75
.73
45.78
34.91
45.20
27.15
27.84
2.95
3.17
2.80
3.32
.972
1.028
1.036
1.190
.80
.72
.71
.50
27.92
20.99
20.87
11.27
Operational Capability, CR = .89, AVE = .63
Increase the innovation of working practices
Enhance the development of integrated solutions
Promote operational improvements
Increase productivity
Improve employee performance in their operations
External Stakeholder Relations Capability, CR = .78, AVE = .49
Improve the overall company’s leadership in the market
Improve our relationship with suppliers
Improve our relationship with customers
Improve our relationship with regulators or government institutions
Response format for the following construct: To what extent does your organisation meet the targets set for the following categories of measures: Use the scale
where 1 = To a very great extent, 2 = To a great extent, 3 = To some extent, 4 = To a little extent, 5 = To a very little extent, and 6 = Not at all.
Target Performance, CR = .80, AVE = .49
Short term financial results – e.g. operating income, sales growth, etc.
Customer relations- e.g. market share, customer satisfaction, etc.
Employee relations- e.g. employee satisfaction, safety, etc.
Operational performance – e.g. productivity, lead times, etc.
Environmental performance – environmental compliances, etc.
2.25
2.59
3.06
2.41
2.94
.870
1.016
1.038
.911
1.276
.50
.69
.75
.71
.67
10.94
21.34
25.38
22.35
19.37
322
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 2 (Continued).
Latent Variables and Respective Manifest Items
Mean/item
Standard
Deviation
Standardized
Coefficient
T-value
Response format for the following construct: In comparison with the industry average, how would you rate the performance of your company over the last two years
in terms of the following indicators? Use the scale where: 1: Well Above Average, 2: Above Average, 3: Short of Above Average, 4: Average, 5: Short of Below Average,
6: Below Average, 7: Well Below Average
Subjective Financial Performance, CR = .93, AVE = .77
Rate of sales growth
Rate of profit growth
Return on investment (ROI)
Profit/sales ratio
3.11
3.28
3.35
3.30
1.257
1.358
1.289
1.266
.78
.91
.90
.92
35.03
75.31
72.86
87.89
.82
.87
.85
31.67
33.22
35.96
Diagnostic Use-Second Order Factor, CR = .88, AVE = .72
Monitoring
Focus Attention
Legitimization
Fit Indices: Chi-Square (df) = 1985.576 (1090), Chi-Square/df = 1.82, NNFI = .92, CFI = .93, RMSEA = .046, 90% CI RMSEA = (.043,.049)
operationalizes Dynamic Tension to mitigate potential multicollinearity. The structural model fit was evaluated using Mplus via
several criteria and structural paths were examined for statistical
significance based on t-tests and respective p-values. Correlations
with objective financial performance measures appear negative
due to the specific scaling (see Table 2) of all perceptual measures.
Based on ROT and RBT, we control for the possession of a variety
of both intangible and tangible resources. This was necessitous in
order to isolate the unique contributions of asset orchestration in
contrast to asset possession. The intangible assets we control for
include Research & Development Expenditures and Industrial Patents
& Intellectual Property Rights while the tangible assets encompass Plant & Machinery and Industrial & Commercial Equipment.
We also control for contextual variables such as firm size (number of employees in current year), industry (manufacturing versus
service) and whether the firm is multinational.
Table 4 presents standardized coefficients along with respective significance levels and t-values. The fit indices indicate that the
structural model exhibits a good model-to-data fit (2 = 2515.635,
df = 1666, 2 /df = 1.509, CFI = 0.91, NNFI = 0.90, RMSEA = 0.043, 90%
CI of RMSEA = [0.039, 0.046]). H1a and H1b predict respectively that
diagnostic use and interactive use have a positive effect on strategic management capability. There is ample statistical evidence to
support the hypotheses (p < .000 and p < .000 respectively) although
the effect size for Diagnostic Use ( = 0.625) is significantly larger
than the effect size of Interactive Use ( = 0.246). Furthermore, H1c
suggests that strategic management capability is subject to the
interaction between diagnostic use and interactive use, termed as
Dynamic Tension. Dynamic Tension did produce a positive relationship ( = 0.084) with Strategic Management Capability though
the effect was statistically moderate (p < .05). Similarly, H2a , H2b ,
and H2c posit respective positive effects on operational capability.
All three standardized effects were positive but clearly the effect of
diagnostic use (as opposed to interactive use and dynamic tension)
is more prominent from a substantive ( = 0.577 vs = 0.233 vs
= 0.113) and statistical perspective (p < .000 vs p < .002 vs p < .010).
As far as external stakeholder relations capability is concerned,
diagnostic use (H3a ) and interactive use (H3b ) play a significant role.
Diagnostic Use and Interactive Use have statistically significant positive effects ( = 0.535, p < .000 and = 0.320, p < .000 respectively)
on External Stakeholder Relations Capability while the impact of
Dynamic Tension (H3c ) was negative, albeit minimal ( = −0.032,
p < .280).
Collectively, all three explanatory variables contribute towards
organizational capabilities. While we expected that to be the case,
we were surprised that diagnostic use had such a strong positive
showing given Henri’s (2006a) conceptual and empirically based
designation of diagnostic use as a “negative force.” We note however that our operationalization of diagnostic use and our choices
for capabilities differ and thus direct comparisons may not be
fruitful. Henri operationalized diagnostic use only via measures
of monitoring, which may be perceived as surveillance instruments that can be used merely to exercise managerial control.
Instead, when monitoring is coupled with focusing attention and
legitimization, employees can more clearly see the purpose of
diagnostic use and verify that prior actions were credible, commanding higher levels of buy-in. We found that diagnostic use is a
mechanism organizations can rely on to improve a wide range of
capabilities.
Organizations strive to meet their target performance using
several means. The data analysis here illustrates that these capabilities can have a statistically significant and positive effect on the
ability of the organization to meet its target performance. Interestingly, the effect of external stakeholder relations capability was
more pronounced ( = 0.431, p < .000) than the effects of strategic
management capability ( = 0.174, p < .024) and operational capability ( = 0.163, p < .027). This was surprising as we expected that
operational capability, being more proximal to performance, would
probably be the most salient factor. However, we found some corroborating theoretical and empirical evidence in the literature that
can perhaps explain the prominence of stakeholder relations capability as an explanatory variable. For instance, Sarkis et al. (2010)
underscore the importance of stakeholders and according to Franco
et al. (2012) organizations are under a relentless pressure to deliver
value not only to shareholders but increasingly to other stakeholders. The first consequence is perhaps the inclusion of specific
organizational performance targets which are salient to stakeholders while the second consequence is the development of a top
leadership compensation structure which is tied to stakeholder
relations/corporate social performance.
The stakeholder theory (Donaldson & Preston, 1995) is considered to be rather instrumental for performance. As Donaldson
and Preston (1995) articulate, “. . .the principal focus of interest
here has been the proposition that corporations practicing stakeholder management will, other things being equal, be relatively
successful in conventional performance terms.” (p. 66–67). Berman
et al. (1999) present a strategic stakeholder management model
under which “. . .the nature and extent of managerial concern for a
stakeholder group is viewed as determined solely by the perceived
ability of such concern to improve firm financial performance” (p.
488). Berman et al. (1999) empirically demonstrated that after
controlling for the strategy and operational environment, stakeholder efforts are positively related to performance. A review of
the literature that examined the relationship between corporate
Table 3
Correlation matrix.
1
2
3
4
5
6
7
8
1. Monitoring
2. Focus Attention
3. Legitimization
4. Diagnostic Use1
5. Interactive Use
6. Dynamic Tension2
7. Strg. Mgmt Cap.
8. Operational Cap.
9. External Stakehol.
Relations Cap.
10. Target Perf.
11. Subjective
Financial Perf.
12. ROTA3
13. ROA4
14. ROE5
15. EBITM6
16. Employee Size7
17. Industry
18. Geographic Div.
19. R&DE8
20. IP&IPR9
21. P&M10
22. ICE11
1
.633**
.604**
.796**
.525**
.286**
.599**
.448**
9
10
1
.630**
.883**
.477**
.318**
.554**
.554**
1
.897**
.513**
.387**
.583**
.511**
1
.575**
.391**
.660**
.590**
1
.408**
.644**
.616**
1
.404**
.375**
1
.635**
1
.374**
.463**
.468**
.465**
.410**
.485**
.487**
.543**
.522**
.448**
.219**
.257**
.473**
.458**
.578**
.492**
1
.453**
1
.216**
−.113*
−.107*
−.081
−.135**
−.048
−.068
−.220**
.008
.019
.093
−.107
.225**
−.088
−.080
−.066
−.082
−.099
−.045
−.117*
−.021
−.053
.104
−.139*
.191**
−.102*
−.094
−.101*
−.121*
−.025
−.056
−.185**
.024
−.016
.224
−.096
.239**
−.113*
−.105*
−.096
−.125*
−.066
−.062
−.191**
.004
−.026
.095
−.131*
.241**
−.073
−.069
−.085
−.114*
−.066
−.074
−.156**
−.036
−.020
.747
−.115*
.101*
−.036
−.034
−.128*
−.048
−.039
−.120*
−.057
−.017
−.028
.779
−.011
.273**
−.048
−.040
−.035
−.066
−.098
−.027
−.101*
−.004
−.005
−.104
−.102
.246**
−.070
−.058
−.058
−.083
−.021
−.081
−.117*
−.047
−.034
−.008
−.132*
.247**
−.002
.002
−.074
.006
−.063
−.041
−.131**
−.084
−.055
−.013
−.106
.360**
−.124*
−.116*
−.178**
−.130*
−.083
−.099
−.262**
−.029
.002
−.078
−.103
2
11
12
13
14
15
16
17
18
19
20
21
22
1
−.094
−.085
−.168**
−.109*
−.111*
−.021
−.054
.009
.003
−.127*
−.051
1
.994**
.619**
.962**
−.130*
.233**
.228**
.009
.045
.048
.033
1
.618**
.957**
−.128*
.229**
.222**
.008
.037
.048
.015
1
.588**
−.041
.217**
.144**
.020
.035
.048
−.031
1
−.123*
.213**
.227**
.011
.180**
.430**
.202**
1
−.145**
.155**
.185**
.256**
.255**
.467**
1
.181**
.067
−.109*
−.134*
.001
1
.071
−.034
.043
.134*
1
.020
−.011
.297**
1
.153**
.566**
1
.377**
1
Diagnostic Use is a second-order Construct that includes Monitoring, Focus Attention, & Legitimization, Dynamic Tension= Diagnostic Use * Interactive Use,
ROTA = Return on Total Assets = EBIT/Total Net Assets where EBIT= Revenue–COGS - Operating Expenses - Depreciation & Amortization
ROA = Return on Assets = Net Income/Total Assets
5
ROE = Return on Equity = Net Income/Shareholders Equity
6
EBITM = Earnings before Interest and Taxes Margin = Last Four Quarters of Operating Earnings/Last Four Quarters of Sales
7
Employee size was measured on a scale where: 1: n < 500, 2: 500 < n < 999, 3: 999 < n < 1499, 4: 1500 < n < 1999, 5: 2000 < n < 2999, 6: 3000 < n < 9999, 7: n ≥ 10,000
8
Research & Development Expenditures (R&DE - in Euros): Investments in Research & Development (Based on the Balance Sheet). Count as Intangible Assets.
9
Industrial Patents and Intellectual Property Rights (IP&IPR - in Euros): Count as Intangible Assets.
10
Plant and Machinery (P&M - in Euros): Investments in Tangible Assets.
11
Industrial and Commercial Equipment (ICE - in Euros): Investments in Tangible Assets.
3
4
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
1
Variable
323
324
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 4
Hypotheses testing.
Hypothesis
Effect
Standardized Coefficient
t-Value
Significance
H1a
H1b
H1c
H2a
H2b
H2c
H3a
H3b
H3c
H4
H5
H6
Diagnostic Use→ Strategic Management Capability
Interactive Use→ Strategic Management Capability
Dynamic Tension→ Strategic Management Capability
Diagnostic Use→ Operational Capability
Interactive Use→ Operational Capability
Dynamic Tension→ Operational Capability
Diagnostic Use→ External Stakeholder Capability
Interactive Use→ External Stakeholder Capability
Dynamic Tension→ External Stakeholder Capability
Strategic Management Capability → Target Performance
Operational Capability → Target Performance
External Stakeholder Capability → Target Performance
Target Performance→ Subjective Financial Performance
Target Performance→ROTA
Target Performance→ROA
Target Performance→ROE
Target Performance→EBITM
Firm Size → Subj. Fin. Perf.
Industry → Subj. Fin. Perf.
Geog. Diversification → Subj. Fin. Perf.
R&D Expenditures → Subj. Fin. Perf.
Industrial Patents & Intell. Prop. Rights → Subj. Fin. Perf.
Plant & Machinery → Subj. Fin. Perf.
Industrial & Commercial Equipment → Subj. Fin. Perf.
Firm Size → ROTA
Industry → ROTA
Geog. Diversification → ROTA
R&D Expenditures → ROTA
Industrial Patents & Intell. Prop. Rights → ROTA
Plant & Machinery → ROTA
Industrial & Commercial Equipment → ROTA
Firm Size → ROA
Industry → ROA
Geog. Diversification → ROA
R&D Expenditures → ROA
Industrial Patents & Intell. Prop. Rights → ROA
Plant & Machinery → ROA
Industrial & Commercial Equipment → ROA
Firm Size → ROE
Industry → ROE
Geog. Diversification → ROE
R&D Expenditures → ROE
Industrial Patents & Intell. Prop. Rights → ROE
Plant & Machinery → ROE
Industrial & Commercial Equipment → ROE
Firm Size → EBITM
Industry → EBITM
Geog. Diversification → EBITM
R&D Expenditures → EBITM
Industrial Patents & Intell. Prop. Rights → EBITM
Plant & Machinery → EBITM
Industrial & Commercial Equipment → EBITM
.625
.246
.084
.577
.233
.113
.535
.320
−.032
.174
.163
.431
.414
−.199
−.169
−.225
−.115
−.090
−.010
.070
.026
−.031
−.102
.053
.088
.026
.007
.005
−.011
.054
−.080
.153
.001
−.016
.019
−.038
.073
−.127
.058
.093
−.096
.059
−.009
.105
−.147
.070
−.107
−.045
.025
−.097
.492
−.015
8.878
3.224
1.712
7.308
2.820
2.324
6.200
3.578
−.587
1.974
1.922
4.156
6.952
−3.035
−2.584
−3.526
−1.954
−1.345
−.174
1.152
.418
−.447
−1.393
.686
1.272
.435
.110
.084
-.149
.712
−1.020
2.232
.018
−.258
.291
-.531
.976
−1.623
.846
1.581
−1.563
.927
−.128
1.422
−1.904
1.141
−2.015
−.809
.429
−1.530
7.943
−.222
.000
.000
.043
.000
.002
.010
.000
.000
.280
.024
.027
.000
.000
.001
.005
.000
.025
.090
.431
.125
.338
.328
.082
.243
.102
.332
.456
.465
.441
.438
.154
.013
.493
.398
.355
.298
.165
.052
.198
.057
.059
.354
.449
.078
.028
.127
.022
.209
.334
.063
.000
.412
H7
Controls
Fit Indices: Chi-Square (df) = 2515.635 (1666), Chi-Square/df = 1.509, NNFI = .91, CFI = .90, RMSEA = .043,
90% CI RMSEA = (.039,.046). Statistically significant effects are noted in italics.
social performance and corporate financial performance over
1972-2002 identified 109 empirical studies, out of which 54
exhibited a positive association while only 7 suggest a negative
association (Margolis & Walsh, 2003).
In the context of PM systems, Cousins et al. (2008) and Mahama
(2006) empirically found that PM systems enhance perceived interfirm financial and nonfinancial performance indirectly by first
improving cooperation and socialization among firms. Mahama’s
(2006) research suggests that PM systems help to ensure that
performance information is distributed fairly among participants
in the supply chain relationship, which promotes learning and
problem solving. Furthermore, the sharing of information helps
to align the interests amongst relationship constituents, motivating them to adapt to changes without the need for coercive
measures. Cousins et al. (2008) demonstrate that the use of PM systems enhances communication, which improves socialization, an
essential element that can facilitate orchestration of resources
across organizational boundaries.
Given the perceived contributions that relations with external
stakeholders can beget, companies have incentivised their top leadership in order to improve relations with stakeholders (Deckop
et al., 2006). Towards this end, Deckop et al. (2006) suggest that CEO
pay structure can be used as an incentive to promote better relations with stakeholders. They found that CEO long-term pay focus
did have an impact on corporate social performance. We conjecture
that given the incentive regime where top leadership is rewarded
based on relations with stakeholders, the organization will be more
likely to measure performance against respective targets and then
direct resources toward the improvement of stakeholder relations,
which subsequently fuels better performance. Considering that
this is a relatively nascent topic, we expect that there is significant variance on how organizations use their resources to improve
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Diagnosc Use
325
Subjecve
Financial
Performance
Strategic
Management
Capability
ROTA
Dynamic Tension
(Diagnosc X Interacve)
Operaonal
Capability
Target
Performance
ROA
ROE
Interacve Use
External
Stakeholder
Relaons
Capability
EBIT
P - Values
p<.01
0.01 ≤ p<.05
Non - stascally Significant
Fig. 2. Test Results Note: Effects on ultimate dependent variables are controlled for industry, firm size, geographic diversification, R&D expenditures, Industrial Patents &
Intellectual Property Rights, Plant & Machinery, and Industrial & Commercial Equipment. These effects are displayed in Table 4.
stakeholder relations capabilities and this rationale explains the
striking positive effects of external stakeholder relations capabilities on target performance.
We next examined whether firms that meet their target performance actually do perform well financially, while controlling
for intangible and tangible firm resources as well as firm demographics. We demonstrate that meeting target performance does
have a significant impact on both subjective financial performance
as well as objective financial measures of performance. Subjective financial performance was operationalized via several manifest
variables while objective financial performance was reflected via
four measures: ROTA, ROA, ROE, and EBITM (details about formulations appear at the bottom of Table 3). As expected, the effect
of Target Performance on Subjective Financial Performance was
substantive (ˇ = .414, p < .000, R2 = .195). Though the effects on
objective financial performance measures were not as strong, they
are sizable given the plethora of other variables that can impact
the specific dimensions of financial performance. The coefficients
assume a negative sign due to the reverse scale coding of Target Performance (see Table 2). The respective impact was: ROA
(ˇ = −.169, p < .005, R2 = .050), ROTA (ˇ = −.199, p < .001, R2 = .048),
ROE (ˇ = −.225, p < .000, R2 = .079), and EBITM (␤=−.115, p<.025,
R2 = .252). Fig. 2 presents the hypothesized structural model and
respective significance levels. Although we suspected that the
relationship between target performance and subjective financial
performance will be statistically and substantively significant, we
were surprised that the effects on the objective financial measures
of performance were all relatively sizable and statistically significant. There is always the concern that results grounded on primary
data may differ from the results which are resting on secondary
data. In this particular case, the findings are consistent and thus
more credible.
3.3.1. Further review of interaction effects
The data analysis revealed two statistically significant interaction terms relating Dynamic Tension to Strategic Management
Capability and Operational Capability. In order to analyze and
interpret the statistically significant interaction effects (between
diagnostic and interactive use), the next step involved the construction of two separate factorial designs (one for each respective
statistically significant interaction effect). The differences in means
and associated p-values emerged from Univariate General Linear
Models and post hoc procedures using Fisher’s Least Significant
Distance (LSD). Details about specific methodology and results
appear in Appendix B. ANOVA results appear in Tables 5 and 6.
The first issue is the ANOVAs’ overall significance. As described
by the model R2 s, the models have significant explanatory power
since the minimum R2 -adjusted was .365 (for Operational Capability). The next issue concerns the significance of the factors and
interactions. The results demonstrate that for each of the baseline model paths: (1) Diagnostic Use as main effect was positive at
p < .001; (2) Interactive Use as a main effect was positive at p < .001
(confirming the baseline model SEM results); and (3) their interaction was positive and statistically significant, at least at p < .002.
Partial 2 for main or interaction effects in each table reveal that
substantial variation for each dependent variable is explained.
It is clear from Fig. 3a that at a low level of Diagnostic Use, a low
level of Interactive Use produces the lowest mean score for Strategic Management Capability. At a low level of Diagnostic Use, the
means produced by moderate and high Interactive Use are similar
but substantively higher than for low Interactive Use. At a moderate level of Diagnostic Use, the means produced by moderate and
high Interactive Use are still substantially higher than the mean
produced by low Interactive Use but there is a divergence in the
sense that high Interactive Use produces a mean score which is
appreciatively higher than the mean score generated by moderate
Interactive Use. The results for high Diagnostic Use are similar to
those for moderate Diagnostic Use except that the mean produced
by high Interactive Use is now substantially higher than the mean
produced by moderate Interactive Use.
Turning our attention on Interactive Use, at low levels of Interactive Use, higher levels of Diagnostic Use are associated with higher
means of Strategic Management Capability. The slope is positive
and the line appears to be rather linear. Similarly, at moderate levels
of Interactive Use, higher levels of Diagnostic Use produce higher
Strategic Management Capability means. At high levels of Interactive Use, the pattern changes significantly. Moving from low to
moderate levels of Diagnostic Use engenders higher mean levels
for strategic management capability but moving from moderate to
high levels produces an extraordinary response where the mean
value generated by high Diagnostic Use is substantially higher. It is
apparent from Fig. 3a that the low and moderate lines for Interactive Use are practically parallel and relatively linear albeit that the
moderate line is at a higher level. However, the high Interactive Use
line is non-linear and the slope is substantially greater, especially
when moving from a moderate to a high level of Interactive Use. In
essence, the returns from the combination of high Interactive Use
326
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 5
Factorial tests for strategic management capability.
Source
Type III Sum
of Squares
df
Mean Square
F
Sig.
Partial Eta
Squared
Noncent.
Parameter
Observed
Powerb
Corrected Model
Intercept
Interactive Use
Diagnostic Use
Interactive Use* Diagnostic Use
Error
Total
Corrected Total
3564.591a
35603.970
403.850
573.164
239.637
3549.339
98553.000
7113.930
8
1
2
2
4
377
386
385
445.574
35603.970
201.925
286.582
59.909
9.415
47.328
3781.746
21.448
30.440
6.363
.000
.000
.000
.000
.000
.501
.909
.102
.139
.063
378.620
3781.746
42.896
60.880
25.454
1.000
1.000
1.000
1.000
.990
a
b
R2 = .501 (Adjusted R2 = .490).
Computed using alpha = .05.
Fig. 3. (a) Estimated Marginal Means of strategic Management capacity, (b) Estimated Marginal Means of Operational capability.
and high Diagnostic Use appear to be more like exponential than
linear.
The results for Operational Capability are similar to the findings
for Strategic Management Capability in some respects but divergent in other. Fig. 3b does demonstrate that at higher levels of
Diagnostic or Interactive Use the means for Operational Capability
are always higher. At a low level of Diagnostic Use, high Interactive
Use produces the highest mean for Operational Capability but it
does not appear to be substantively different from the mean generated by moderate Interactive Use. At a moderate level of Diagnostic
Use, the mean produced by high Interactive Use is still the highest but the mean produced by a moderate level of Interactive Use
begins to converge to the mean produced by low Interactive Use.
This phenomenon is more pronounced at a high level of Diagnostic Use where the means generated by moderate and low levels
of Interactive Use converge while the mean for high Interactive
Use diverges upwards significantly. The convergence implies that
at high levels of Diagnostic Use, there is no difference between
the mean scores for Operational Capability whether there is low or
moderate level of Interactive Use. Focusing on the levels of Interactive Use, unlike the graph for Strategic Management Capability,
the low and moderate lines are not parallel though both appear to
be practically linear. The two lines converge when Diagnostic Use
is high. However, the line associated with high Interactive Use is
non-linear and the slope changes substantively above a moderate
level of Diagnostic Use.
Collectively, diverse combinations of Diagnostic and Interactive
Use produce significantly different means for Strategic Management Capability and Operational Capability. In general, high levels
of both Diagnostic and Interactive use engender high mean scores
across both dependent variables. However, the combination of high
Diagnostic Use and high Interactive Use produces mean scores
which are significantly higher than all other combinations. These
findings suggest that companies should be using PMS from both
diagnostic and interactive logics. Furthermore, the positive slopes
observed for all lines in Fig. 3 do imply that the effects on capabilities appear to be synergistic rather than compensatory. Practically,
Table 6
Factorial tests for operational capability.
Source
Type III Sum
of Squares
df
Mean Square
F
Sig.
Partial Eta
Squared
Noncent.
Parameter
Observed
Powerb
Corrected Model
Intercept
Interactive Use
Diagnostic Use
Interactive Use* Diagnostic Use
Error
Total
Corrected Total
1868.033a
31007.481
201.572
314.641
145.586
3076.394
82715.000
4944.427
8
1
2
2
4
377
386
385
233.504
31007.481
100.786
157.321
36.397
8.160
28.615
3799.845
12.351
19.279
4.460
.000
.000
.000
.000
.002
.378
.910
.061
.093
.045
228.920
3799.845
24.702
38.558
17.841
1.000
1.000
.996
1.000
.938
a. R2 = .378 (Adjusted R2 = .365), b. Computed using alpha = .05
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
if managers would like to attain the highest levels of capabilities,
they should invest in both types of PM systems.
4. Longitudinal analysis
To examine the longitudinal effects of PM systems on firm financial performance, we used several approaches. These approaches
involved the use of panel data over a ten year time horizon.
4.1. Model 1 – specification
Our first approach utilizes a panel data design using two time
periods. The first time period is the pre-implementation period and
is set as the year before the implementation of PM systems in order
to avoid contamination from potential implementation activities.
The second time period is set as the year 2010 which represents the
year we collected our primary data through a survey. The respondents to the survey furnished detailed information regarding the
number of years since their firm has been using a formal PM system. The intent of this approach is to examine the impact of the
adoption of PM systems on the change in firm performance while
controlling for time invariant factors and other variables. Towards
this end, two competing models could be adopted: a fixed effects
(FE) model or a random effects (RE) model. We opted for the random
effects model because fixed effects models remove time invariant
factors. Furthermore, the Hausman test which compares a FE model
and a RE model was statistically insignificant, favoring a RE model
for our analysis.
327
been using a PM system at the time of primary data collection (i.e., 2010). Research development expendituresit represents the
expenditures for a given firm for research and development and is
representative of the intangible resources possessed by the firm.
Similarly, Industrial pattents intellectual property rightsit also represents investments in intangible resources. Plant machineryit and
Industrial commercial machineryit reflect resources which are tangible in nature. The mean of the unobserved heterogeneity is a
constant term labeled as “˛.” The unobserved random firm specific effects are labeled as ui while εit is the disturbance term. The
analysis was performed with STATA12 using the xtreg (re) command.
4.2. Model 1 – results
The approach we followed was iterative in nature. The model
was first tested with all variables included and then iteratively
identified and excluded variables which do not contribute statistically toward each of the respective dependent variables, i.e., the
four financial ratios. Several control variables were retained though
variable retention varies by model.
Table 7 displays the coefficients for PM system adoption, the
R-square values, the Wald’s 2 statistic values, and the p-values
accompanied by their respective significance level for the final
model we obtained. Wald’s test is an omnibus test and it is used
to examine whether the variables collectively have a significant
impact on each dependent variable. Based on Wald’s 2 we find
Perfit = ˇ1 PMsystem diagostic useit + ˇ2 PMsystem interactive useit + ˇ3 PMsystem diagnostic useit
∗ PM system interactive useit + ˇ4 PM Years of system useit + ˇ5 firm sizei
+ˇ6 Geog Diversificationi +
j=16
ˇj indj + ˇ7 Research developement exp enditureit
j=1
+ˇ8 Industrial pattents intellectual property rightsit
+ˇ9 Plant machinery exp endituresit + +ˇ10 Industrial commercial equipmentit + ˛ + ui
+εit
The variable Perfit essentially represents the difference in financial performance of a firm i between two time periods. The first
period is calculated as the year at which the firm begun using
the formal PMS minus one. For instance, if a firm implemented
and begun using the PMS reportedly in 2007, the first period is
set as 2006. We set the second time as 2010, the year for which
we collected primary/survey data for diagnostic and interactive
types of usage. The financial performance is either measured in
terms of changes in ROA, ROTA, ROE, or EBITM and thus four models were estimated. The variables included in the model consist of
the manner in which PM systems are used (i.e., diagnostic use and
interactive use) along with their interaction. The values for diagnostic and interactive use of the PM system for the second time
period capture the extent and manner in which a PM system is
deployed within each firm at the time of cross-sectional data collection. The values for the diagnostic and interactive use of the PM
system are set at ‘0 in the first time period when firms had not yet
implemented a PM system. Covariates also include controls that we
identified as variables that can potentially influence our dependent variables. These include firm size (firm sizei , industry (indj ),
and geographical ownership diversification (Geog Diversificationi ).
Firm size is captured using a seven point Likert type scale, while the
geographical diversification is depicted using a binary variable that
denotes whether a firm has multinational presence or not. Specific
industry affiliation was reported in the survey. Furthermore, we
control for the duration of PM system use. PMYears of system useit
assumes a value of 0 for the first period while for the second
time period it is set as the number of years since the firm has
that all variables, including the time invariant factors, jointly had
an influence on each dependent variable (Wald’s 2 is statistically
significant at least at 0.001).The results presented in Table 7 suggest
that Diagnostic Use has a statistically significant negative impact on
the change in ROA (p < .05) and the change in EBITM (p < .05) (notice
that based on variable scaling for Diagnostic and Interactive Use, a
positive coefficient is indicative of negative effects and a negative
coefficient is reflective of positive effects), while the interactive use
of PM systems has a positive moderate impact on the change in
EBITM (p < .10). Dynamic Tension appears to have a strong positive
impact on the change in ROA (p < 0.05) and ROE (p < 0.01) and a
moderate positive impact on the change in ROTA (p < 0.10).
The findings suggest that the Diagnostic and Interactive uses
along with Dynamic Tension appear to explain changes in financial
performance between the pre-implementation period and 2010.
What the data suggests is rather interesting. Diagnostic use does
contribute to changes in financial performance (for ROA and EBITM)
but those companies reporting larger changes in diagnostic use
between the two periods experience smaller changes in financial performance across the two periods. This does not necessarily
intimate that diagnostic use has a negative effect on financial
performance. It suggests instead that as companies add to their
diagnostic use portfolio over time, they will experience smaller or
less dramatic changes in ROA and EBITM. In essence, the effects on
the two performance variables may still be positive but they get
smaller in magnitude for firms that use diagnostic systems more
intensely over time. On the other hand, companies that add to their
interactive use over the years engender larger changes in EBITM.
328
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 7
Longitudinal effects –model 1.
Dependent Variables
Independent Variables
Diagnostic Use
Interactive Use
Dynamic Tension
Years of PMS Use
Finance and Insurance
Manufacturing
Mining
Other Services
Real Estate, Rental & Leasing
Geog. Diversification
R&DEc
IP&IPRd
ICEe
Constant
Observations
Groups
Wald’s ␹2
R2
a
b
c
d
e
f
†
*
**
***
ROA
.0148a (.029b )*
−.021(.190)
−.002(.014)*
.041(.096)† ,f
−.639(.063)†
−.417(.006)**
−1.654(.000)***
ROTA
.004(.314)
−.005(.416)
−.001(.063)†
ROE
.008(.167)
.024(.176)
−.002(.015)*
.616(.075)†
−.402(.009)**
−1.620(.000)***
−1.582(.000)***
−.276(.026)*
-.370(.005)**
.119(.068)†
.151(.029)*
.097(.097)†
.504(.002)**
250
137
32.79(.000)***
.154
.606(.000)***
250
137
28.86(.000)***
.144
.097(0.097)†
245
137
30.89(.000)***
.124
−1.202(.001)***
EBITM
.014(.028)*
−.034(.077)†
−.000(.389)
−.555(.068)†
.556(.009)**
1.892(.000)***
−.157(.023)*
.136(.035)*
.261(.003)**
−.040(.289)
237
131
46.76(0.000)***
.224
Coefficient.
p-value.
R&DE: Research & Development Expenditures.
IP&IPR: Industrial Patents & Intellectual Property Rights.
ICE: Industrial & Commercial Equipment.
Only statistically significant effects at p < .10 are shown for control variables.
p-value < 0.1.
p-value < 0.05.
p-value < 0.01.
p-value < 0.001.
Similarly, firms that escalate both diagnostic use and interaction
use over time do experience larger changes in ROA, ROE, and ROTA
in contrast to firms that exhibit lower levels of both diagnostic and
interaction use.
4.3. Model 2 – specification
Model 2 utilizes a panel data design but includes several observations per individual firm. Again, we opted for the RE model
because the use of FE model tends to ignore time invariant factors. The technique followed here resembles the approach used
by Benner and Veloso (2008), but we use a RE model instead of
using a FE model. We use the following specification to further
examine whether the use of a PM system does lead to better firm
performance over time.
PM system while after adoption they received values 1, 2, 3 . . .
for each year until the final time period in our analysis (i.e., 2012).
For example, if a particular firm adopted a PM system in 2010, it
received a value of zero for the years 2003 until 2009, and subsequently it received values 1, 2, and 3 for the years 2010, 2011,
and 2012 respectively. A similar approach was carried to capture
ISO9000 adaptation trend by Benner and Veloso (2008). However,
we recognize that learning may diminish over time and thus we
tested the model also using three learning rates (80%, 60%, and 40%)
in order to have more robust results and respective conclusions.
The variables included in the model consist of PM system adaptation, and the covariates that we identified and specified in earlier
analysis. We also included time dummies (Year dumk ) to control
for the yearly shocks that can potentially influence the performances of all firms. We operationalize the mean of the unobserved
Perfit = ˇ1 PMsystem adaptationit + ˇ2 Geog Diversificationi + ˇ3 firm sizei +
j=16
k=10
j indj +
j=1
k Year dumk
k=1
+ˇ4 Research development exp endituresit
+ˇ5 Industrial pattents intellectual property rightsit + ˇ6 Plant machinery exp endituresit
+ˇ7 Industrial commercial equipmentit + ˛ + ui εit
Due to learning effects, we theorize that firms benefit while
adapting to a new system. The variable Perfit represents the financial performance of a firm i during time period t. The financial
performance is either measured in terms of ROA, ROTA, ROE,
or EBITM. Data from 2003–2012 are used in data analysis. Prior
empirical work on windows of opportunity after process changes
suggests that firms tend to adapt to a new system or technology
over a period of time (Heim & Ketzenberg, 2011; Tyre & Orlikowski,
1994). Thus, we created a variable related to longitudinal PM system adoption which is used to measure the trend of PM system
adaptation (PMsystem adaptationit ). This variable is built to capture
the improvement in performance as firms adapt to the new system
after implementing it. In our coding scheme, firms received a value
of zero for PMsystem adaptationit until such time they adopted a
heterogeneity as a constant term which we label as ␣, while ui
captures the unobserved firm specific random effects and ␧it is the
disturbance term. The analysis was performed with STATA12 using
the xtreg (re) command.
4.4. Model 2 – results
Table 8 (which includes one panel for each measure of financial
performance) produces the coefficients of PM system adaptation,
R-square values, Wald’s 2 statistics, and the p-values accompanied by their respective significance level. Across all four dependent
variables, the Wald’s 2 statistic is highly statistically significant for
the basic model as well across learning rates, except of two occasions for ROA (i.e., at a learning rate of 60% and 40%). This suggests
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
329
Table 8a
Longitudinal effects –model 2: Panel A (ROA).
Trend Since Adoption
Independent Variables
Adaptation
Geog. Diversification
Firm Size
Industry Dummiesd
Time Dummies
R&DE
IP&IPR
P&M
ICE
Constant
Observations
Groups
Wald’s ␹2
R2
a
b
c
d
†
*
**
***
ROA (Trend = +1/year)
.087a (.000b )***
.029(.376)
.053(.027)*
Included
Included
−.018(.220)
.018(.279)
.017(.309)
−.049(.075)†
−.769(.001)**
2361
315
50.88(.000)***
.067
ROA (Trend = 80% LRc )
.116(.000)***
.037(.346)
.054(.024)*
Included
Included
−.020(.191)
.016(.298)
.016(.319)
−.047(.088)†
−.685(.004)**
2361
315
45.04(.000)***
.065
ROA (Trend = 60% LR)
.081(.020)*
.065(.242)
.058(.018)*
Included
Included
−.022(.171)
.014(.320)
.017(.312)
−.042(.111)
−.362(.071)†
2361
315
34.77(.125)
0.054
ROA (Trend = 40% LR)
.017(.351)
.083(.182)
.061(.014)*
Included
Included
−.021(.177)
.014(.328)
.018(.297)
−.041(.119)
−.136(.276)
2361
315
30.61(.217)
0.042
Coefficient
p-value
LR = Learning Rate
Industry and Time dummies were included in the model but not displayed here.
p-value < 0.1.
p-value < 0.05.
p-value < 0.01.
p-value < 0.001.
that all variables, including time invariant factors and the time
dummies, jointly had a strong influence on the dependent variables. The individual coefficients for industry and time dummies
are not displayed in Table 8 for simplicity of exposition.
We find that the adaptation variable has a positive influence
on firm performance across all four performance measures for the
basic model (see Table 8-all four panels). The lowest value for R2
is roughly 6% while the highest value is around 20% and both are
rather substantive given the overabundance of variables than can
impact financial performance over time. At an 80% learning rate,
we observe similar results where all the coefficients are statistically significant for the four financial measures. At a 60% learning
rate, the adaptation coefficient for ROE appears (see panel C) to be
non-statistically significant. Nevertheless, the coefficients for ROA,
ROTA, and EBITM remain statistically significant and although Rsquare values sustain some losses, they are still substantive. Finally,
at a 40% learning rate, only the coefficients for ROTA and EBITM
(see panels B and D) remain statistically significant. As expected
the R-squares saw some decline.
The data analysis suggests here that the expansive use of PM
systems over time is indeed conducive towards firm performance.
We conjecture that the longer an organization uses its PM system
the more effectively it becomes at orchestrating its resources. However, we also demonstrated that the impressions of this adaptation
over time are subject to learning rate effects. When higher learning rates are applied, the effects of adaptation become more salient
towards performance. The practical consequences can be significant; if a firm does not maintain the momentum engendered by
the implementation of a PM system and continue adapting over
time at a high rate, it may see diminishing effects on performance.
Thus, management is urged to continue investing in PM systems
and advocate programs that can uphold high learning rates.
Table 8b
Longitudinal Effects–Model 2: Panel B (ROTA).
Trend Since Adoption
Independent Variables
Adaptation
Geog. Diversification
Firm Size
Industry Dummiesd
Time Dummies
R&DE
IP&IPR
P&M
ICE
Constant
Observations
Groups
Wald’s 2
R2
a
b
c
d
†
*
**
***
ROTA (Trend = +1/year)
.068a (.000b )***
.024(.399)
.062(.013)*
Included
Included
−.022(.187)
−.002(.481)
.022(.272)
−.027(.227)
−.526(.022)*
2336
315
45.95(.016)*
.069
ROTA (Trend = 80% LRc )
.104(.000)***
.024(.379)
.062(.013)*
Included
Included
−.024(.169)
−.003(.467)
.021(.281)
−.026(.241)
−.524(.023)*
2336
315
45.16(.019)*
.069
Coefficient
p-value
LR = Learning Rate
Industry and Time dummies were included in the model but not displayed here.
p-value < 0.1.
p-value < 0.05.
p-value < 0.01.
p-value < 0.001.
ROTA (Trend = 60% LR)
.106(.005)**
.040(.331)
.064(.011)*
Included
Included
−.025(.150)
−.004(.447)
.021(.280)
−.022(.273)
−.344(.085)†
2336
315
40.72(.046)*
0.063
ROTA (Trend = 40% LR)
.079(.047)*
.056(.271)
.066(.009)**
Included
Included
−.026(.145)
−.005(.437)
.022(.272)
−.020(.293)
−.162(.243)
2336
315
38.36(.091)†
0.054
330
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 8c
Longitudinal Effects–Model 2: Panel C (ROE).
Trend since adoption
Independent Variables
Adaptation
Geog. Diversification
Firm Size
Industry Dummiesd
Time Dummies
R&DE
IP&IPR
P&M
ICE
Constant
Observations
Groups
Wald’s 2
R2
a
b
c
d
†
*
**
***
ROE (Trend = +1/year)
.030a (.057b )†
−.059(.239)
.042(.048)*
Included
Included
−.027(.153)
.056(.048)*
.032(.190)
−.089(.009)**
−.113(.321)
2317
315
66.30(.000)***
.063
ROE (Trend = 80% LRc )
.038(.100)†
−.056(.252)
.043(.046)*
Included
Included
−.028(.146)
.055(.050)*
.032(.190)
−.088(.011)*
−.071(.385)
2317
315
65.36(.000)***
.062
ROE (Trend = 60% LR)
.029(.240)
−.048(.283)
.044(.042)*
Included
Included
−.028(.140)
.054(.053)†
.033(.188)
−.086(.011)*
−.024(.459)
2317
315
64.33(.000)***
0.059
ROE (Trend = 40% LR)
.007(.442)
−.041(.309)
.044(.039)*
Included
Included
−.028(.140)
.054(.054)†
.033(.184)
−.085(.012)*
.103(.318)
2317
315
63.92(.000)***
0.057
Coefficient.
p-value.
LR = Learning Rate.
Industry and Time dummies were included in the model but not displayed here.
p-value < 0.1.
p-value < 0.05.
p-value < 0.01.
p-value < 0.001.
4.5. Model 2 robustness check –specification
We also performed a robustness check for Model 2 using a
Propensity Score Matching (PSM) technique. Ideally, it would be optimal to assess whether the performance of the adopters would have
been different if they had not adopted PM systems. However, it
is not practically feasible to explicitly compare the performance
of adopters after adoption and under the assumption that they had
not adopted. One approach to overcome this challenge is to identify
non-adopters that are similar to adopters based on certain characteristics, and compare the performance of the adopters to these
similar firms that have not yet adopted PM systems. In essence,
a Propensity Score Matching technique enables us to compare the
treatment firms (i.e., early adopters of PM systems) and control
groups (i.e., firms that did not adopt a PM system) which share
similar characteristics. PSM allows us to evaluate firms based on a
single propensity score, which here is determined based on industry (Corbett et al., 2005), firm size, the geographic diversification of
firms (Peikes et al., 2008), as well as the four resource-based variables. Since adoption can be modelled as a dichotomous variable,
we use a logit model shown below to determine the propensity
score in STATA12. The propensity score for the observations in
our sample is the probability estimate obtained from our logistic
regression model.
j=16
Logit(p(Adoptiont )) = ˛ + ˇ1 Geog Diversificationi + ˇ2 firm sizei +
indj +
j=1 j
ˇ3Research development exp endituresit + ˇ4Industrial pattents intellectual property rightsit
ˇ5plant machinery exp endituresit + ˇ6Industrial commercial equipment it + ε
Table 8d
Longitudinal effects –model 2: panel D (EBITM).
Trend Since Adoption
Independent Variables
Adaptation
Geog. Diversification
Firm Size
Industry Dummiesd
Time Dummies
R&DE
IP&IPR
P&M
ICE
Constant
Observations
Groups
Wald’s 2
R2
a
b
c
d
†
*
**
***
EBITM (Trend = +1/year)
.044a (.009b )**
−.101(.124)
.066(.007)**
Included
Included
−.029(.093)†
−.016(.297)
.345(.000)***
−.021(.268)
−.543(.013)*
2311
312
176.28(.000)***
.195
EBITM (Trend = 80% LRc )
.081a (.002b )**
−.107(.112)
.065(.007)**
Included
Included
−.030(.086)†
−.016(.295)
.344(.000)***
−.021(.268)
−.619(.006)**
2311
312
178.87(.000)***
.197
Coefficient.
p-value.
LR = Learning Rate.
Industry and Time dummies were included in the model but not displayed here.
p-value < 0.1.
p-value < 0.05.
p-value < 0.01.
p-value < 0.001.
EBITM (Trend = 60% LR)
.108a (.002b )**
−.101(.124)
.066(.006)**
Included
Included
−.032(.074)†
−.017(.286)
.344(.000)***
−.019(.289)
−.557(.008)**
2311
312
179.38(.000)***
0.195
EBITM (Trend = 40% LR)
.111a (.004b )**
−.090(.150)
.068(.005)**
Included
Included
−.033(.066)†
−.018(.278)
.345(.000)***
−.016(.311)
−.434(.022)*
2311
312
178.12(.000)***
0.191
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 9
Longitudinal effects –robustness check for model 2.
Dependent
Variable
Adoption
Type
ROA
ROTA
ROE
EBITM
5. Summary and discussion
Performance 2012
PM System
Adopters
175
175
175
175
PM System
Non-Adopters
54
49
51
49
Average treatment effect
for treated (ATT) (95% C.I)
.607 (.143, .940)
.377 (.038, .817)
.466 (.022, .793)
.251 (.001, .631)
In our PSM method we employ the nearest neighbor matching
technique which in turn uses the computed propensity score to
match the treatment firms with the control group (Heckman et al.,
1997). In essence, the interest is to examine the average effect of
PM systems on the adopters by comparing them directly against
the most comparable non-adopters. This effect is termed as the
average treatment effect on the treated (ATT). The empirical form
for estimating ATT in the case of the nearest neighbor approach is
shown below.
ATT =
1
n1
y1,i − y0,j
331
i ∈ {D=1)
The outcome for the treated observation i which belongs to the
treated group D is represented by y1,i while y0,j denotes the outcome for the matched observation j in the control group and n1
is the number of observations in the treated group. The matched
observation from the control group is determined using the criteria
min||pi -pj || where Pi is the propensity score for a treated observation and Pj is the propensity score for the non-treated observation.
The nearest neighbor matching approach enables us to compute the effect of PM systems adoption on firm performance by
controlling for the propensity of adoption between adopters and
non-adopters. We choose 2005 to be the year we splice the data
to examine the difference between adopters and non-adopters
of PM systems. The choice of the specific year was governed by
the principle that we need a sufficiently large sample of firms
that had not yet adopted a PM system in order to identify a suitable control group while retaining a sufficiently large number of
firms that had adopted a PM system (>50%). We carried out PSM
using the nearest neighbor method on outcome variables in 2012
(Guo & Fraser, 2010). Recall that we demonstrated earlier that
performance improves with adaptation, rendering support for performance improvement over time across the whole study sample.
We label firms that have adopted PM systems by 2005 as early
adopters.
4.6. Model 2 robustness check results
Table 9 provides the results for the average treatment effect on
the treated group. The significance of the results is assessed based
on the 95% confidence interval (C.I) values for the ATT obtained by
bootstrapping the standard errors. We find significant evidence to
suggest that the early adopters of PM systems have better financial
performance when compared to the control group. We specifically
find that treated firms, when compared to others, have a significantly higher performance measured by all financial measures.
We find that this result corroborates our findings presented earlier
where we conclude that the adaptation variable has a significant
and positive impact on performance. The data analysis furnishes
additional evidence to suggest that longer usage of PM systems,
ceteris paribus, is rather conducive for firm performance.
Over the last two decades, firms have invested heavily in implementing PM systems under the premise that such systems will
deliver financially. Practitioners, consultants, societies, and academics alike ascribe positive evaluations to PM systems and tout
the benefits. However, there is scant empirical evidence to validate such claims. First, the literature is devoid of a comprehensive
framework that identifies specific uses of PM systems and examines their respective relationships to organizational capabilities
and firm performance. In essence, the extant literature lacks a
theoretically based comprehensive nomological network. Second,
there is no material evidence attesting that PM systems do actually
contribute positively to financial performance over time. We summarize our key findings and recommendations for future research
in Table 10.
As far the extant literature is concerned, our empirical findings
are rather interesting. First, we found that interactive use is a positive force just as Henri (2006a) articulated. However, diagnostic
use is also contributing positively towards organizational capabilities and performance. This was manifested by the instrumental
role of diagnostic PM system use in impacting positively all three
capabilities examined in the cross-sectional study. In fact, diagnostic use exhibited the most potent effects on capabilities from both
a statistical and a substantive perspective. Although we did not
hypothesize and test for indirect effects, there is significant evidence here to suggest that diagnostic use is also indirectly related
with the subjective measure of financial performance (p < .000) as
well as with the four objective measures of financial performance
(p-value for any indirect coefficient at least < 031). Furthermore, the
positive role of diagnostic use can be surmised when the impact
of dynamic tension is considered, as it is generated from the joint
forces of interactive and diagnostic use. Dynamic tension was crucial for explaining variability in strategic management capability
and operational capability. Also, from a longitudinal perspective,
dynamic tension demonstrated statistically significant positive
effects on three out of four measures of change in financial performance. On the aggregate, the evidence contradicts Henri’s (2006a)
claim that diagnostic use acts as negative force. As explained earlier
in the results section however, Henri’s operationalization of diagnostic use and choice of capabilities may have contributed to the
disparity between our findings. Also, it is noteworthy to point that
the correlations between diagnostic use, capabilities, and performance, as reported in Henri (2006a, p. 541), were all positive, albeit
one correlation with a capability was not statistically significant.
In order to elicit a more thorough interpretation regarding the
use of PM systems, we conducted retrospective interviews with
four managers that participated in the cross-sectional study. All
managers suggested that it would be most beneficial if a PMS is
used from both a diagnostic and interactive logic concurrently. In
one of the companies where diagnostic use dominated, a manager labeled his top management as “authoritarian” and “rigid”
based on how it deployed diagnostic use. He also suggested that
diagnostic use is akin to a “corporate thermometer.” But, while
top management could see the temperature rising, it could not
interact effectively with constituents to find out what was going
awry. Top management communicated with functional managers
via a top–down approach but this communication had the ulterior motive of control. Lower level managers were totally cut-off
from information exchange and thus a lot of opportunities for
identifying the root causes of challenges were relinquished. At
another company, a manager pointed that diagnostic use and interactive use work best when deployed in unison. For example, she
noted that actions undertaken by management via interactive use
can only be evaluated for their soundness by an ex post assessment afforded by the diagnostic system. Similarly, she stated that
332
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Table 10
Going forward: new research questions and implications.
Key Finding/Limitation
Diagnostic Use, Interactive Use, & Dynamic
Tension were found to impact a set of
competitive capabilities positively.
Diagnostic Use exhibited the most potent effects
in the cross-sectional study.
Diagnostic Use demonstrated diminishing
financial returns over time when examined via a
longitudinal study.
PM system use does improve financial
performance over time but learning rates do
matter. How firms adapt over time is important.
We did not examine the behavior of specific
types of PM use over time; rather we assessed
whether the presence or absence of PM systems
makes a difference in financial performance
overall.
We examined the impact of PM systems through
the theoretical lenses of the Resource
Orchestration Theory. We found that resource
orchestration, after accounting for resource
possession, does matter financially. We did not
examine how PM system uses affect specific
resource orchestration nuances: structuring the
resource portfolio, bundling resources, and
leveraging resources.
We studied the impact of PM system uses on
capabilities and performance variables. These
outcomes pertain to the measurement of
capabilities and performance of a firm. Given the
level of inter-firm relationships, it would be
fruitful to examine the saliency of specific PM
system uses on supply chain capabilities and
performance.
Our financial performance measures were
inclusive of cost considerations but we did not
specifically examine the implementation costs of
PM systems and subsequent usage costs.
We produced evidence suggesting that diverse
uses of PM systems engender positive
performance. However, it is possible that PM
systems also produce negative effects which
went unaccounted here. Based on our interviews
with practitioners, we suspect that mere use of
Diagnostic Use will produce negative outcomes.
We studied organizational level effects; we did not
examine the impact of different types of PM
system uses on the individual. It is not known
whether PM system use impacts for instance the
motivation or creativity of individuals.
The use of PM systems may impact individual
and organizational attributes. We do not know
whether the effects on the individual can explain
more or less variance in performance than the
effects on organizational attributes.
Also, we do not know whether the effects on the
individual and organization vary by the cultural
environment where the firm operates.
We studied firms operating in Italy. Although a
number of multinationals were included, we do
not know whether the findings stand in light of
differences in cultural environments.
Research Question
Discussion & Potential Contribution
Diagnostic versus Interactive PM system use
What is the idiosyncratic contribution of the two
types of use and dynamic tension on capabilities
and performance over time? Is the effect of
Diagnostic Use significantly more potent? What
are the effects on capabilities and performance
metrics which are most salient to operations and
supply chain management?
Do firms that merely rely on Diagnostic Use
perform well across a variety of capabilities and
performance measures?
Are the effects of Diagnostic Use and Interactive
Use sustainable over time?
What activities can the organizations engaged in to
maintain high learning rates?
PM Systems & resource orchestration
Do the effects of diverse PM system uses vary
across the three resource orchestration nuances?
Which specific PM system uses are more salient for
each nuance?
What is the relative efficacy of resource
orchestration vis-à-vis resource possession in
explaining performance over time?
Is the interaction between resource orchestration
and resource possession statistically significant?
Do the different types of PM system use foster
resource orchestration at the supply chain level?
PM system implementation costs & adverse effects
Do the costs of implementation and usage vary by
type of PM system use?
What type of negative effects does PM system use
produce?
Are the effects of Diagnostic Uses and Interactive
Uses equally salient when negative effects are
considered?
Effects on the organization vis-à-vis the individual
What is the impact of the different types of PM
system uses on the individual attributes (e.g.,
motivation, creativity)?
Do the effects on the individual explain more or
less variance in performance than the effects on
organizational attributes?
Do the effects on the individual and organization
vary by the cultural environment where the firm
operates?
Generalizability across cultural environments
Do the relationships studied here vary by cultural
environments?
diagnostic use can be seen as a mere futile exercise if actions
via interactive use are not carried out. Considering the response
from the first manager, the organization may observe the temperature rising on the “thermometer” but if no action is initiated to
Organizations possess limited resources and these
resources need to be orchestrated as effectively as
possible. Firms need to focus on systems that can
produce the best returns for the money and effort.
Examining the individual and collective role of
each type of PM system use over time can help
organizations properly apportion resources across
diagnostic and interactive uses of PM systems. The
organizations also need to know where specifically
to invest their resources in order to retain high
learning rates, which were found to be crucial for
sustaining high levels of financial performance.
Supply chain management professionals and
scholars also need to know whether PM systems
matter when respective metrics are considered.
Resource orchestration necessarily involves a lot of
activities. Some deal for instance with the
acquisition of resources while others involve the
mobilization of resources. Organizations need to
know which type of a PM system use is most
efficacious for each activity and invest fittingly.
Furthermore, organizations need to know whether
it is best to accumulate resources or exploit
existing resources or whether there needs to be a
balance in budget allocation across the two. Also,
considering the level of outsourcing that exists
today, organizations need to know whether their
specific PM systems and respective uses can
actually help them orchestrate resources across
the value chain. If adjustments are needed, they
need to know where those are needed.
Before organizations invest in PM systems, they
need to have an accurate estimate about the costs
of implementation and usage. They also need to
know the relative costs generated by diagnostic vs
interactive PM systems so they can make better
investments. Although there are some conjectures
and cursory evidence, we do not know much about
the negative consequences prompted by the use of
PM systems and which type of use is most
problematic (and thus needs to be addressed).
PM system uses may impact organizational
attributes but it is likely that they have a pervasive
impact on the individual attributes as well.
Organizations need to know what individual
attributes get affected and which specific type of
PM system use stimulates it. In order to invest
wisely, organizations need to know also whether it
is best to promote capabilities at the individual
level, organizational level, or both. Furthermore,
organizations need to know whether the effects
are sustainable across different environments so
they can make proper adjustements.
Organizations need to know whether the saliency
of specific types of PM system use is invariant
across environments. They need to identify which
type of PM system use works best for each
environment.
uncover the true causes of the surge in temperature, the organization’s performance may be at risk. Based on our analysis
and our interviews, it appears that diagnostic use is indispensable but management should be warned that if a PM system is
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
used solely from a diagnostic use motive, performance may be at
peril.
Second, the extant operations management or accounting literatures have produced no longitudinal studies regarding the effects
of PM systems that involve multiple companies over a considerable time. We initially utilized a panel data approach and specified
a random effects model to examine and contrast financial performance across two different periods (one of the periods was
pre-implementation and one after). We found that some of the
rates of change in performance are positively related with interactive use and dynamic tension while some of the rates of change
in performance are negatively related with diagnostic use, suggesting perhaps diminishing returns. We also examined a learning
effect due to adaptation by applying a variety of learning rates
over a ten year period. There is strong evidence to suggest that the
effects of adaptation on all four financial measures of performance
are positive across diverse learning rates. However, the effects of
adaptation on performance wane down when lower learning rates
are applied. Finally, a robustness check compared PM system early
adopters and PM system non-adopters using a propensity score
matching approach and produced evidence to suggest that PM system early adopters outperform PM system late and non-adopters
across all four financial measures. Admittedly the Propensity Score
Method is rather demanding because of the difficulty of identifying
“similar” firms, yet it still identified that early adopters perform the
best as far as financial performance is concerned. When we view
all the results jointly, we can assert that PM systems do positively
impact performance but organizations need to maintain high levels
of learning.
From a theoretical perspective, ROT suggests that organizations
need not only amass resources but they also need to orchestrate
those resources in order to garner competitive advantage (Hitt et al.,
2011; Sirmon et al., 2011). We position PM systems as mechanisms
that can help organizations manage and synchronize resource
orchestration (Hitt et al., 2011). Effective synchronization demands
information regarding the firm’s environment and a framework
that can be used to process that information. A PM system, through
its capacity as an information and control system, can facilitate top
management to evaluate prior actions and craft a new strategy
through the acquisition, bundling, and mobilization of resources.
Hitt et al. (2011) along with Sirmon et al. (2007,2011) are credited
with the advancement of the ROT tenets. They claim that the primary actions needed for resource orchestration include structuring
the resource portfolio (i.e., acquiring, accumulating, and divesting
resources), bundling resources (i.e., stabilizing existing capabilities,
enriching current capabilities, and pioneering new capabilities) and
leveraging resources (i.e., mobilizing capabilities to form capability
configurations, coordinating capability configurations, and deploying the configurations). Although we conceptually linked resource
orchestration at large with PM systems, a more in depth study
is needed to tie the three specific nuances of resource orchestration with PM systems. In other words, further research is needed
to understand how a PM system facilitates the constituent elements of structuring, bundling, and leveraging of resources. For
instance, how does diagnostic use of a PM system affect mobilization? Sirmon et al. (2007), resting on Hamel and Prahalad (1994),
assert that “The intent of mobilizing is to identify the capabilities
needed and to design the capability configurations necessary to
exploit opportunities in the market and gain competitive advantage” (p. 284). Diagnostic use of PM systems may be more conducive
towards the identification of capabilities needed through its monitoring attributes, which furnish information regarding deviations
from targets. On the other hand, design of capabilities may be
facilitated more effectively by the interactive use of PM systems
because the design of capabilities necessarily involves top management involvement. Similarly, diagnostic use of PM systems may
333
facilitate the enrichment of current capabilities while interactive
use may be more salient for pioneering new capabilities. We expect
that the different types of PM system uses and the tension they
create will have pervasive effects on the three primary resource
orchestration actions but their resonance may vary.
Our study adds to the body of literature by providing empirical evidence attesting to the value of resource orchestration. We
used four measures of resource endowments (both tangible and
intangible) as control variables in our empirical inquiry in order
to isolate the effects of resource orchestration from the mere possession of resources. In this respect, we provide empirical support
attesting to the efficacy of resource orchestration to explain performance. From both a theoretical and an empirical perspective, it
would be interesting to examine the relative role of resource possession vis-à-vis resource orchestration. Their relative contribution
towards performance can be examined via variance partitioning.
While both may be valuable however, we believe that the best performance level can be attained when valuable resources are used
effectively. In other words, we posit that the interaction between
resource possession and resource orchestration can explain performance variance more effectively. Organizations cannot orchestrate
resources they do not have or control, and possessing resources
without leveraging them can be futile.
The operations management literature and the supply chain
management literature have relied extensively on the resourcebased theory but there is scant research that examines the
orchestration of resources. The importance of PM systems is taken
for granted but more is needed to ascertain how specific PMS uses
can be deployed to orchestrate resources and add on firm capabilities. A recent exception is Ketchen et al. (2014) who examine
resource orchestration in the context of product returns. Given,
however, that firms now also rely on the resources bestowed by
their supply chain partners, it would be useful to examine how
PM systems facilitate the structuring, bundling, and mobilization
of resources across organizational boundaries. We are arguing in
essence that resource orchestration has to be considered in light
of supply chains that span multiple firms and multiple locations.
Although diagnostic PM system use can furnish valuable information pertaining to the organization, its efficacy may be curtailed
as far as the supply chain is concerned because the organization
may lack access to network-level information. Such information
may reside with suppliers and customers across the supply chain.
Obtaining relevant information and orchestrating action across the
network may require top management involvement. Top management can communicate with its counterparts to obtain information,
stimulate the development of new ideas, and deliberate on possible courses of action. In essence, interactive PM system use may be
more productive than diagnostic use.
Future research could also benefit from measuring the costs of
implementation and maintenance of PM systems and use these
costs to evaluate whether the benefits outweigh the costs. Our
measures of financial performance were net of costs but it would be
instrumental for both managers and academics to ascertain what
it specifically costs to implement and run PM systems. Widener
(2007) used management attention as a surrogate for cost and
hypothesized that diagnostic control systems make more efficient
use of management attention vis-à-vis interactive control systems.
Based on responses from 122 Chief Financial Officers, Widener
(2007) found support for her hypotheses. Widener explains that
challenges or concerns that are more routine in nature have smaller
information deficits and thus can be handled efficiently via diagnostic systems, without much intervention from management. On
the other hand, Widener credits the additional cost for interactive systems on the larger information deficits and more equivocal
tasks interactive systems are purported to tackle. Since interactive systems by definition personally involve individuals at the top
334
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
managerial ranks, they can devour managerial attention. While
diagnostic systems may not demand constant managerial attention, there is a compelling need however to invest in tangible
and intangible assets that foster PM diagnostic use; i.e., monitoring, focusing attention, and legitimization. For instance, the
organization may need to invest in information/surveillance systems (hardware, software, and human capital) that can furnish
information to monitor critical success factors. As Simons (1994)
states, “Any diagnostic control system can be made interactive by
continuing and frequent top management attention and interest”
(p. 171). Once the diagnostic system is in place, managers can use
it interactively. We assert that the implementation cost may be
higher for diagnostic systems than interactive systems but this is
an empirical question that future research can address. It would be
worthwhile to evaluate whether the costs of the two PM system
uses do vary and identify which one is the costlier. The conception
of cost has to be more encompassing however and include elements
beyond managerial attention.
Future research should also account for the potential negative
effects that may emerge from using PM systems. Some individuals
envision PM systems as bureaucratic, time consuming, and unnecessary. Voelpel et al. (2006) suggest that Balance Score Cards (BSCs),
which represent an exemplar of PM systems, are rigid and internally focused and are credited with static-ism. These “tyrannical”
effects, as Voelpel et al. coin them, may have adverse effects on
innovation. However, Voelpel et al. (2006) were assessing the role
of BSCs from a pure diagnostic PM system use perspective. Future
research should include variables connoting negative attributions,
such as rigidity and static-ism, as dependent variables to measure
and account for the role of diagnostic and interactive PM system
uses and dynamic tension as we evaluate performance. We suspect
that if an organization purely uses PM systems from a diagnostic
use perspective the effects on innovation may be indeed negative,
while the effects of interactive use and dynamic tension may be
positive.
PM systems can also be seen as translating a firm’s strategy
to its employees (Burney & Widener, 2013). Thus, the impact on
the individual actors should also be examined. For instance, how
does the use of PM systems impact employee motivation, effort,
and performance? Burney and Widener (2013) furnish empirical
evidence suggesting individual level effects but more research is
due. Specifically, Burney and Widener (2013) used responses from
242 financial services employees to demonstrate that the extent
to which a strategic performance measurement system is coupled with strategy impacts employee performance via perceived
self-efficacy and perceived psychological contract while controlling
for the effects of organizational citizenship behavior. Burney and
Widener relied on data from a single company, which generates
generalizability concerns, and they focused only on a few pertinent individual level variables. Furthermore, they did not study
the individual or collective role of PM system use (i.e., diagnostic vs interactive vs dynamic tension). It would be rather conducive
to assess the effects of PM system use on the individual by examining such effects across many organizations and cultures as well as
including more variables. For instance, we do not know how diagnostic PM use or interactive PM use or their collective use impact
creativity at the individual level. Our conjecture is that mere diagnostic use will stifle creativity as it suppresses experimentation and
elicits performance which is aligned to pre-established targets.
Furthermore, we do not know whether individual (employee)
performance, fuelled by PM system use, does impact capabilities
and firm performance. Rothaermel and Hess (2007) studied the
effects of individual, firm, and network level variables on innovation output in the pharmaceutical industry and hypothesized direct
effects as well as two competing interaction hypotheses (i.e., substitute versus complementary effects). They found that a significant
amount of variance in innovation was explained by individual-level
factors but they also uncovered that individual-level variables can
serve as substitutes for firm- and network-level antecedents to
innovation. Within the realm of PM system use, there are no studies that examine the relative or combined effect of individual and
firm level effects on organizational performance. While we expect
individual and firm level factors to contribute to organizational performance, it is likely that the effects are complementary. Creativity
and motivation reside with the individual and firms rely on human
capital to build their capabilities. At the same time, individuals rely
on the organizational capabilities to perform more effectively. Thus,
we postulate that individual (employee) performance and organizational level capabilities interact in a complementary fashion to
impact firm performance. Empirical research is however necessary
to address the research question.
The generalizability of our findings is somewhat limited in the
sense that data was obtained only from firms operating in Italy,
although a sizable number of multinationals were included. Both
the sociology and international business literatures have examined
cross-regional differences and their potential effects on organizational life (Deal and Kennedy, 1982; Hofstede, 1980). Performance
measurement systems are not immune to these effects; regional
differences in culture, political systems, and economic development can have pervasive effects on the organization (Rhody &
Tang, 1995). We suspect that three cultural variables, power distance (where authority is stratified and concentrated), collectivism
(promoting group goals over individual goals), and uncertainty
avoidance (comfort with ambiguity and change), exert influence
on the relationships between PM types of use and capabilities.
Our conjecture is that these variables may serve as moderators.
Cultural attributes derive from both formal and informal societal
structures and are reflective of belief and boundary systems. Belief
and boundary systems may act as catalysts or as hurtles when organizations use PM systems (Simons, 1994). Future research should
sample from other countries and respective geographical regions
in order to identify whether the effects of specific PM system uses
vary across cultural dimensions.
Appendix A. Appendix A Examination of Competing
Measurement Model Configurations
In order to examine the efficacy of a second-order model specification to explain variation for Diagnostic Use, a series of competing
models were examined. The first model specified one first-order
latent variable of Diagnostic Use, the second model specified three
uncorrelated first-order latent variables (i.e., monitoring, focus
attention, and legitimization) while the third model specified three
correlated first-order latent variables. Finally, the last model specified one second-order latent variable with three first-order latent
variables. Iteratively, the models are compared examining whether
there is a statistically significant difference in 2 values, where
lower 2 values are favored. Model 1 produced a 2 = 2470.76 (1094
df) while Model 2 had an 2 = 2402.00 (1082 df). The diff 2 is 68.76
and is statistically significant (P < .01) based on 12 df. Model 3 produced a significantly lower 2 = 1964.489 (1079 df) than Model
2 and the difference of 437.511 (3 df) is statistically significant
(p < .01). Finally, Model 4 generated a 2 = 1985.576 (1090 df) and
diff 2 = 21.087 (21 df), which is not statistically significant, suggesting model equivalence based on model fit. In fact, a comparison
of CFI and NNFI illustrates practically identical model fit between
Models 3 and 4. Statistically, first-order correlated models, such
as Model 3, will always generate a better model fit than secondorder models, such as Model 4, but a second-order model with
comparative model fit offers an attractive alternative when theory can support such model. Given the underlying characteristics
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
of a diagnostic performance measurement system, consideration
for parsimony, and based on statistical findings, the best model to
represent Diagnostic Use appears to be at the second-order level of
abstraction.
Appendix B. Appendix B Interaction Effects Explanations
For interpretability, the main effect constructs were recoded based on quartiles as follows: (1) Interactive Use into
“Low” (below 25% percentile) if Interactive Use < = 10 (N = 122),
“Moderate” (between 25% and 75% percentiles) if Interactive
Use = 10.01–15.00 (N = 202), and “High” (above 75% percentile)
if Interactive Use > = 15.01 (N = 62); and (2) Diagnostic Use into
“Low” (below 25% percentile) if Diagnostic Use < = 44 (N = 101),
“Moderate” (between 25% and 75% percentiles) if Diagnostic
Use = 44.01–59.00 (N = 203), and “High” (above 75% percentile) if
Diagnostic Use > = 59.01 (N = 82). Using Factorial ANOVA, the main
and interaction effects were examined using p-values and partial
2 (analogous to R2 , reflecting explanatory strength).
Fig. 3a and b display three lines representing Low/
Moderate/High Interactive use where Diagnostic Use is represented on the X-axis. The Y-axes represent Strategic Management
Capability and Operational Capability respectively. The graphs
can be interpreted in two ways: (1) given a particular level of
Diagnostic Use, do the Strategic Management Capability (or the
Operational Capability) means differ depending on the level of
Interactive Use? The answers to these types of questions can be
visualized in the graph by choosing a level of Diagnostic Use (e.g.,
low on the x-axis) and comparing the three Interactive Use means
graphed at that level of Diagnostic Use. (2) The second way to
interpret the results answers the question: given a particular level
of Interactive Use, do the Strategic Management Capability or
the Operational Capability means differ depending on the level
of Diagnostic Use? For example, given low Interactive Use, do
the Strategic Management Capability means differ depending on
whether Diagnostic Use is low vs. moderate? The answers to these
types of questions can be visualized in the graph by choosing
a level of Interactive Use (ex., the low Interactive Use line) and
comparing the three Diagnostic Use means graphed (e.g., Fig. 3a) at
the corresponding three levels of Diagnostic Use (low vs. moderate
vs. high on the Diagnostic Use x-axis).
References
Agle, B.R., Mitchell, R.K., Sonnenfeld, J.A., 1999. What matters to CEOs? An investigation of stakeholder attributes and salience, corporate performance and CEO
values. Acad. Manage. J. 42 (5), 507–525.
Ahn, H., 2001. Applying the balanced scorecard concept: an experience report. Long
Range Plann. 34 (4), 441–461.
Ahrens, T., Chapman, C.S., 2004. Accounting for flexibility and efficiency: a field study
of management control systems in a restaurant chain. Contemp. Account. Res.
21 (2), 271–301.
Amit, R., Schoemaker, P.J., 1993. Strategic assets and organizational rent. Strategic
Manage. J. 14 (1), 33–46.
Baldauf, A., Reisinger, H., Moncrief, W.C., 1999. Examining motivations to refuse in
industrial mail surveys. J. Market Res. Soc. 41 (3), 345–353.
Benner, M.J., Veloso, F.M., 2008. ISO 9000 practices and financial performance: a
technology coherence perspective. J. Oper. Manage. 26 (5), 611–629.
Berman, S.L., Wicks, A.C., Kotha, S., Jones, T.M., 1999. Does stakeholder orientation
matter? The relationship between stakeholder management models and firm
financial performance. Acad. Manage. J. 42 (5), 488–506.
Birnberg, J.G., Turopolec, L., Young, S.M., 1983. The organizational context of accounting. Account. Org. Soc. 8 (2), 111–129.
Bisbe, J., Batista-Foguet, J.M., Chenhall, R., 2007. Defining management accounting
constructs: a methodological note on the risks of conceptual misspecification.
Account. Org. Soc. 32 (7), 789–820.
Bisbe, J., Malagueño, R., 2012. Using strategic performance measurement systems for
strategy formulation: does it work in dynamic environments? Manage. Account.
Res. 23 (4), 296–311.
Bisbe, J., Otley, D., 2004. The effects of the interactive use of management control
systems on product innovation. Account. Org. Soci. 29 (8), 709–737.
335
Bititci, U.S., Carrie, A.S., McDevitt, L., 1997. Integrated performance measurement
systems: a development guide. Int. J. Oper. Prod. Manage. 17 (5), 522–534.
Bititci, U.S., Turner, T., 2000. Dynamics of performance measurement systems. Int.
J. Oper. Prod. Manage. 20 (6), 692–704.
Bourne, M., Mills, J., Wilcox, M., Neely, A., Platts, K., 2000. Designing, implementing
and updating performance measurement systems. Int. J. Oper. Prod. Manage. 20
(7), 754–771.
Bourne, M., Pavlov, A., Franco, M., Lucianetti, L., Mura, M., 2013. Generating organisational performance: the contributing effects of performance measurement and
human resource management practices. Int. J. Oper. Prod. Manage. 33 (11/12),
1599–1622.
Braam, G.J., Nijssen, E.J., 2004. Performance effects of using the balanced scorecard:
a note on the Dutch experience. Long Range Plann. 37 (4), 335–349.
Bullinger, H.J., Kuhner, M., Van Hoof, A., 2002. Analyzing supply chain peformance
using a balanced measurement method. Int. J. Prod. Res. 40 (15), 3533–3543.
Burney, L., Widener, S.K., 2013. Behavioral work outcomes of a strategic performance measurement system-based incentive plan. Behav. Res. Account. 25 (2),
115–143.
Chenhall, R.H., 2005. Integrative strategic performance measurement systems,
strategic alignment of manufacturing, learning and strategic outcomes: an
exploratory study. Account. Org. Soc. 30 (5), 395–422.
Chenhall, R.H., Langfield-Smith, K., 1998. The relationship between strategic priorities, management techniques and management accounting: an empirical
investigation using a systems approach. Account. Org. Soc. 23 (3), 243–264.
Chenhall, R.H., Morris, D., 1995. Organic decision and communication processes and
management accounting systems in entrepreneurial and conservative business
organizations. Omega 23 (5), 485–497.
Chia, A., Goh, M., Hum, S.H., 2009. Performance measurement in supply chain entities: balanced scorecard perspective. Benchmark Int. J. 16 (5), 605–620.
Chirico, F., Sirmon, D.G., Sciascia, S., Mazzola, P., 2011. Resource orchestration
in family firms: investigating how entrepreneurial orientation, generational
involvement, and participative strategy affect performance. Strat. Entr. J. 5 (4),
307–326.
Choi, J.W., Hecht, G.W., Tayler, W.B., 2013. Strategy selection, surrogation, and strategic performance measurement systems. J. Account. Res. 51 (1), 105–133.
Corbett, C.J., Montes-Sancho, M.J., Kirsch, D.A., 2005. The financial impact of ISO
9000 certification in the United States: an empirical analysis. Manage. Sci. 51
(7), 1046–1059.
Cousins, P.D., Lawson, B., Squire, B., 2008. Performance measurement in strategic
buyer-supplier relationships: the mediating role of socialization mechanisms.
Int. J. Oper. Prod. Manage. 28 (3), 238–258.
Crabtree, A.D., DeBusk, G.K., 2008. The effects of adopting the balanced scorecard on
shareholder returns. Adv. Account. 24 (1), 8–15.
Craighead, C.W., Ketchen, D., Dunn, K.S., Hult, G., 2011. Addressing common
method variance: guidelines for survey research on information technology,
operations, and supply chain management. Eng. Manage. IEEE Trans. 58 (3),
578–588.
Crook, T.R., Ketchen, D.J., Combs, J.G., Todd, S.Y., 2008. Strategic resources and performance: a meta- analysis. Strategic Manage. J. 29 (11), 1141–1154.
Daft, R.L., Lengel, R.H., 1986. Organizational information requirements, media richness and structural design. Manage. Sci. 32 (5), 554–571.
Davis, S., Albright, T., 2004. An investigation of the effect of balanced scorecard implementation on financial performance. Manage. Account. Res. 15 (2),
135–153.
Deckop, J.R., Merriman, K.K., Gupta, S., 2006. The effects of CEO pay structure on
corporate social performance. J. Manage. 32 (3), 329–342.
Deal, T.E., Kennedy, A.A., 1982. Corporate Cultures. Reading MA:. Addison-Wesley.
de Leeuw, S., van den Berg, J.P., 2011. Improving operational performance by influencing shopfloor behavior via performance management practices. J. Oper.
Manage. 29 (3), 224–235.
Donaldson, T., Preston, L.E., 1995. The stakeholder theory of the corporation: concept, evidence, and implications. Acad. Manage. Rev. 20 (1), 65–91.
Estampe, D., Lamouri, S., Paris, J.L., Brahim-Djelloul, S., 2013. A framework for
analysing supply chain performance evaluation models. Int. J. Prod. Econ. 142
(2), 247–258.
Fitzgerald, L., Brignall, S., Silvestro, R., Voss, C., Robert, J., 1991. Performance Measurement in Service Businesses: Chartered Institute of Management Accountants
London.
Franco-Santos, M., Lucianetti, L., Bourne, M., 2012. Contemporary performance measurement systems: a review of their consequences and a framework for research.
Manage. Account. Res. 23 (2), 79–119.
Freeman, R.E., 1984. Strategic Management: A Stakeholder Approach. Pitman,
Boston.
Galbraith, J.R., 1974. Organization design: an information processing view. Interfaces
4 (3), 28–36.
Garengo, P., Sharma, M.K., 2014. Performance measurement system contingency
factors: a cross analysis of Italian and Indian SMEs. Prod. Plan. Control 25 (3),
220–240.
Grafton, J., Lillis, A.M., Widener, S.K., 2010. The role of performance measurement and evaluation in building organizational capabilities and performance.
Account. Org. Soc. 35 (7), 689–706.
Grosswiele, L., Röglinger, M., Friedl, B., 2012. A decision framework for the consolidation of performance measurement systems. Decis. Support Syst. 54 (2),
1016–1029.
Gunasekaran, A., Patel, C., Tirtiroglu, E., 2001. Performance measures and metrics in
a supply chain environment. Int. J. Oper. Prod. Manage. 21 (1/2), 71–87.
336
X. Koufteros et al. / Journal of Operations Management 32 (2014) 313–336
Gunasekaran, A., Patel, C., McGaughey, R.E., 2004. A framework for supply chain
performance measurement. Int. J. Prod. Econ. 87 (3), 333–347.
Gunasekaran, A., Kobu, B., 2007. Performance measures and metrics in logistics
and supply chain management: a review of recent literature (1995–2004) for
research and applications. Int. J. Prod. Res. 45 (12), 2819–2840.
Guo, S., Fraser, M.W., 2010. Propensity score analysis. Stat. Methods Appl..
Hamel, G., Prahalad, C.K., 1994. Competing for the Future: Breakthrough Strategies
for Seizing Control of Your Industry and Creating the Markets of Tomorrow.
Harvard Business School Press.
Hamilton, S., Chervany, N.L., 1981. Evaluating information system effectiveness-Part
I: Comparing evaluation approaches. MIS Quart. 5 (3), 55–69.
Hansen, M.H., Perry, L.T., Reese, C.S., 2004. A Bayesian operationalization of the
resource-based view. Strategic Manage. J. 25 (13), 1279–1295.
Harrison, D.A., McLaughlin, M.E., Coalter, T.M., 1996. Context, cognition, and common method variance: psychometric and verbal protocol evidence. Organ.
Behav. Hum. Dec. 68 (3), 246–261.
Heckman, J.J., Ichimura, H., Todd, P.E., 1997. Matching as an econometric evaluation
estimator: Evidence from evaluating a job training programme. Rev. Econ. Stud.
64 (4), 605–654.
Heim, G.R., Ketzenberg, M.E., 2011. Learning and relearning effects with innovative
service designs: an empirical analysis of top golf courses. J. Oper. Manage. 29
(5), 449–461.
Helfat, C.E., Finkelstein, S., Mitchell, W., Peteraf, M., Singh, H., Teece, D., Winter, S.G.,
2007. Dynamic Capabilities: Understanding Strategic Change in Organizations.
Malden. MA: Blackwell.
Henri, J.F., 2006a. Management control systems and strategy: A resource-based perspective. Account. Org. Soc. 31 (6), 529–558.
Henri, J.-F., 2006b. Organizational culture and performance measurement systems.
Account. Org. Soc. 31 (1), 77–103.
Hitt, M.A., Ireland, D.R., Sirmon, D.G., Trahms, C.A., 2011. Strategic entrepreneurship: creating value for individuals, organizations, and society. Acad. Manage.
Perspect. 25 (1), 57–75.
Hofstede, G., 1980. Culture’s Consequences. Beverly Hills. Sage Publications.
Holcomb, T.R., Holmes, R.M., Connelly, B.L., 2009. Making the most of what you have:
managerial ability as a source of resource value creation. Strat. Manage. J. 30 (5),
457–485.
Hoque, Z., James, W., 2000. Linking balanced scorecard measures to size and market
factors: Impact on organizational performance. J. Manage. Account. Res. 12 (1),
1–17.
Hult, G.T.M., Ketchen, D.J., Adams, G.L., Mena, J.A., 2008. Supply chain orientation
and balanced scorecard performance. J. Manage. Issues 20 (4), 526–544.
Ittner, C.D., Larcker, D.F., 1998. Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction. J. Account. Res. 36 (1),
1–35.
Ittner, C.D., Larcker, D.F., Meyer, M.W., 2003a. Subjectivity and the weighting of
performance measures: evidence from a balanced scorecard. Account. Rev. 78
(3), 725–758.
Ittner, C.D., Larcker, D.F., Randall, T., 2003b. Performance implications of strategic
performance measurement in financial services firms. Account. Org. Soc. 28 (7),
715–741.
Kaplan, R.S., Norton, D.P., 1992. The balanced scorecard – measures that drive performance. Harvard Bus. Rev. 70 (1), 71–79.
Kaplan, R.S., Norton, D.P., 2001. Transforming the balanced scorecard from performance measurement to strategic management: Part I. Accounting Horizons 15
(1), 87–104.
Kaplan, R.S., Norton, D.P., 2008. Mastering the management system. Harvard Bus.
Rev. 86 (1), 1–16.
Keegan, D.P., Eiler, R.G., Jones, C.R., 1989. Are your performance measures obsolete?
Manage. Account. 70 (12), 45–50.
Ketchen, D.J., Wowak, K., Craighead, C.W., 2014. Resource gaps and resource orchestration shortfalls in supply chain management: The case of product recalls. J. Sup.
Chain Manage. 50 (3), 6–15.
Koufteros, X., 1999. Testing a model of pull production: a paradigm for manufacturing research using structural equation modeling. J. Oper. Manage. 17 (4),
467–488.
Koufteros, X., Babbar, S., Kaighobadi, M., 2009. A paradigm for examining secondorder factor models employing structural equation modeling. Int. J. Prod. Econ.
120 (2), 633–652.
Lehtinen, J., Ahola, T., 2010. Is performance measurement suitable for an extended
enterprise? Int. J. Oper. Prod. Manage. 30 (2), 181–204.
Lynch, R.L., Cross, K.F., 1992. Measure up!: The Essential Guide to Measuring Business
Performance: Mandarin.
Mahama, H., 2006. Management control systems, cooperation and performance in
strategic supply relationships: a survey in the mines. Manage. Account. Res. 17
(3), 315–339.
Malhotra, N.K., Kim, S.S., Patil, A., 2006. Common method variance in IS research: a
comparison of alternative approaches and a reanalysis of past research. Manage.
Sci. 52 (12), 1865–1883.
Margolis, J.D., Walsh, J.P., 2003. Misery loves companies: rethinking social initiatives
by business. Admin. Sci. Quart. 48, 268–305.
Melnyk, S.A., Stewart, D.M., Swink, M., 2004. Metrics and performance measurement
in operations management: dealing with the metrics maze. J. Oper. Manage. 22
(3), 209–218.
Mundy, J., 2010. Creating dynamic tensions through a balanced use of management
control systems. Account. Org. Soc. 35 (5), 499–523.
Neely, A.D., 1999. The performance measurement revolution: why now and what
next? Int. J. Oper. Prod. Manage. 19 (2), 205–228.
Neely, A.D., 2005. The evolution of performance measurement research. Int. J. Oper.
Prod. Manage. 25 (12), 1264–1277.
Neely, A.D., Gregory, M., Platts, K., 1995. Performance measurement system design: a
literature review and research agenda. Int. J. Oper. Prod. Manage. 15 (4), 80–116.
Neely, A.D., Mills, J., Platts, K., Richards, H., Gregory, M., Bourne, M., Kennerley,
M., 2000. Performance measurement system design: developing and testing a
process-based approach. Int. J. Oper. Prod. Manage. 20 (10), 1119–1145.
Neely, A.D., Adams, C., Kennerley, M., 2002. The Performance Prism: The Scorecard
for Measuring and Managing Business Success. Prentice Hall Financial Times
London.
Pavlov, A., Bourne, M., 2011. Explaining the effects of performance measurement on
performance: n organizational routines perspective. Int. J. Oper. Prod. Manage.
31 (1), 101–122.
Peikes, D.N., Moreno, L., Orzol, S.M., 2008. Propensity score matching. Am. Stat. 62
(3), 222–231.
Porter, M.E., 1991. Towards a dynamic theory of strategy. Strat. Manage. J. 12 (2),
95–117.
Rhody, J.D., Tang, T.L.P., 1995. Learning from Japanese transplants and American
corporations. Public Pers. Manage. 24 (1), 19–32.
Roberts, J., 1990. Strategy and accounting in a UK conglomerate. Account. Org. Soc.
15 (1), 107–126.
Rothaermel, F.T., Hess, A.M., 2007. Building dynamic capabilities: innovation driven
by individual-, firm-, and network-level effects. Organ. Sci. 18 (6), 898–921.
Sarkis, J., Gonzalez-Torre, P., Adenso-Diaz, B., 2010. Stakeholder pressure and the
adoption of environmental practices: the mediating effect of training. J. Oper.
Manage. 28 (2), 163–176.
Simons, R., 1991. Strategic orientation and top management attention to control
systems. Strat. Manage. J. 12 (1), 49–62.
Simons, R., 1994. How new top managers use control systems as levers of strategic
renewal. Strat. Manage. J. 15 (3), 169–189.
Simons, R., 1995. Control in an age of empowerment. Harvard Bus. Rev. 73 (2), 80–88.
Simons, R., Dávila, A., Kaplan, R.S., 2000. Performance Measurement & Control Systems for Implementing Strategy: Text & Cases. Prentice Hall Upper Saddle River,
NJ.
Sirmon, D.G., Hitt, M.A., Ireland, D.R., Gilbert, B.A., 2011. Resource orchestration to
create competitive advantage: breadth, depth, and life cycle effects. J. Manage.
37 (5), 1390–1412.
Sirmon, D.G., Hitt, M.A., 2003. Managing resources: Linking unique resources, management and wealth creation in family firms. Entrepreneurship Theory and
Practice 27 (4), 339–358.
Sirmon, D.G., Hitt, M.A., Ireland, D.R., 2007. Managing firm resources in dynamic
environments to create value: looking inside the black box. Acad. Manage. Rev.
32 (1), 273–292.
Sprinkle, G.B., 2003. Perspectives on experimental research in managerial accounting. Account. Org. Soc. 28 (2), 287–318.
Stede, W.A.V., Chow, d., Lin, C.W.T.W, 2006. Strategy, choice of performance measures, and performance. Behav. Res. Account. 18 (1), 185–205.
Taylor, A., Taylor, M., 2013. Antecedents of effective performance measurement system implementation: an empirical study of UK manufacturing firms. Int. J. Prod.
Res. 51 (18), 1–14.
Tyre, M.J., Orlikowski, W.J., 1994. Windows of opportunity: temporal patterns of
technological adaptation in organizations. Organ. Sci. 5 (1), 98–118.
Vandenbosch, B., 1999. An empirical analysis of the association between the use
of executive support systems and perceived organizational competitiveness.
Account. Org. Soc. 24 (1), 77–92.
Voelpel, S.C., Leibold, M., Eckhoff, R.A., 2006. The tyranny of the balanced scorecard
in the innovation economy. J. Intellect. Cap. 7 (1), 43–60.
Wowak, K.D., Craighead, C.W., Ketchen, D.J., Hult, G.T.M., 2013. Supply chain knowledge and performance: a meta-analysis. Decision Sci. 44 (5), 843–875.
Widener, S.K., 2007. An empirical analysis of the levers of control framework.
Account. Org. Soc. 32 (7), 757–788.
Zimmerman, J.L., 2000. Accounting for Decision Making and Control:.
Irwin/McGraw-Hill, Boston.