Henri2006c sách hệ thống kiểm soát

Furthermore, some evidence suggeststhe inXuence of dynamic tension resulting from the balanced use of PMS in a diagnostic and interactive fashion oncapabilities and performance.. This st

manage-contribute speci Wcally to the creation and maintenance of capabilities leading to strategic choices? (ii) To what extent do

the diagnostic and interactive uses of MCS act in combination to produce dynamic tension which contributes to the

cre-ation and maintenance of these capabilities? (iii) To what extent does the use of MCS contribute to organizcre-ational formance? The results suggest that an interactive use of PMS fosters the four capabilities by focusing organizationalattention on strategic priorities and stimulating dialogue Also, by creating constraints to ensure compliance withorders, the diagnostic use of PMS exerts negative pressure on these capabilities Furthermore, some evidence suggeststhe inXuence of dynamic tension resulting from the balanced use of PMS in a diagnostic and interactive fashion oncapabilities and performance

per-© 2005 Elsevier Ltd All rights reserved

Introduction

In the current business environment

charac-terized by fast changes in customers, technologies

and competition, organizations need to

continu-ously renew themselves to survive and prosper

(Danneels, 2002) Innovativeness, organizational

learning, market orientation and entrepreneurshipare recognized as primary capabilities to reachcompetitive advantage (Hult & Ketchen, 2001;Hurley & Hult, 1998; Ireland, Hitt, Camp, & Sex-ton, 2001) Over the past 15 years, the resource-based view (RBV) of the Wrm on the origins ofcompetitive advantage has become a very inXuen-tial framework and one of the standard theories inthe Weld of strategy (Barney, Wright, & Ketchen,2001; Hoopes, Madsen, & Walker, 2003) The RBV

* Tel.: +1 418 656 7737; fax: +1 418 656 7746.

E-mail address: jean-francois.henri@ctb.ulaval.ca

is based on the principle that competitiveness is a

function of distinctive and valuable resources and

capabilities controlled by a Wrm Despite

consider-able interest in the relationship between

manage-ment control systems (MCS) and strategy, the

MCS literature has devoted scant attention to the

RBV This study seeks to extend the research at

the interface between MCS and strategy with the

application of an RBV framework

So far, a signiWcant body of literature has

explored the eVects of strategy on MCS and, to a

lesser extent, the eVects of MCS on strategy (Dent,

1990; LangWeld-Smith, 1997; Shields, 1997) A Wrst

line of research has emphasized the eVects of

strat-egy on MCS The concept of stratstrat-egy has been

gen-erally examined at a strategic-choice level: (i)

market positioning: cost leadership versus

diVeren-tiation (e.g Bruggeman & Stede, 1993;

Govindara-jan, 1988; Govindarajan & Fisher, 1990), (ii)

strategic pattern: prospector versus defender (e.g

Abernethy & Guthrie, 1994; Hoque, 2004; Simons,

1987), (iii) strategic mission: build, hold, harvest

(e.g Govindarajan & Gupta, 1985; Merchant,

1985), or (iv) strategic priorities: customization,

quality, Xexibility, etc (e.g Abernethy & Lillis,

1995; Chenhall & LangWeld-Smith, 1998; Baines &

LangWeld-Smith, 2003; Ittner, Larcker, & Randall,

2003)

These conceptualizations generally take

strat-egy as a given, consider it from a content

perspec-tive (Fahey & Christensen, 1986), and restrict its

scope to the notion of intended strategy (

Mintz-berg & Waters, 1985).1 In these studies, MCS are

considered for the most part to be

strategy-imple-mentation systems and the last step in the

strate-gic-management process This conceptualization of

MCS follows a structural approach whereby the

perspective is static and the focus is placed on such

issues as the presence or absence of speciWc

sys-tems, their technical properties and their design

(Chapman, 1997, 1998; Dent, 1987)

A second line of research has emphasized the

eVects of MCS on strategy The concept of strategyhas also been examined at a strategic-choice leveland, to a lesser extent, at a capabilities level First,

a number of studies have examined strategy at astrategic-choice level: (i) strategic priorities (e.g.Chenhall, 2005; Marginson, 2002), and (ii) strate-gic change (e.g Abernethy & Brownell, 1999;Chenhall & LangWeld-Smith, 2003) Other studiesrefer indirectly to strategy at a capabilities level interms of innovation or organizational learning (e.g.Bisbe & Otley, 2004; Davila, 2000; Kloot, 1997).These conceptualizations consider strategy asbeing inXuenced by MCS, consider it from a pro-cess perspective (HuV & Reger, 1987), and expandits scope to the notion of emergent strategy(Mintzberg & Waters, 1985) In these studies, therole of MCS in the formulation of strategy is rec-ognized as well as their continuous implicationduring the strategic-management process Thisconceptualization of MCS follows a processualapproach whereby the perspective is dynamic andthe focus is on such issues as the dialogue andinteraction surrounding the use of MCS (Chap-man, 1997, 1998; Dent, 1987)

Numerous authors have pointed out that theWndings provided by the MCS-strategy stream ofresearch remain ambiguous and sometimes contra-dictory (e.g Abernethy & Brownell, 1999; Chap-man, 1997; Chenhall, 2003; Ittner et al., 2003;LangWeld-Smith, 1997) These ambiguous resultscan be attributed in part to the various deWnitions,conceptualizations and operationalizations ofstrategy and MCS (Kald, Nilsson, & Rapp, 2000;LangWeld-Smith, 1997; Simons, 1990) They canalso be explained by two elements: (i) the absence

of a theoretical framework founded on theresource-based view, and (ii) the limited attentiondevoted to the dynamic tension resulting from

diVerent uses or roles of MCS

First, the relationship between MCS and egy may not have been studied at the right level ofanalysis As suggested by Ittner and Larcker(2001), one key element in studying strategy andMCS is to identify the speciWc factors that do infact lead to strategic success Following the RBV,the link between strategy and MCS may occur atthe capabilities level rather than the strategic-

strat-1 Based on the work of Mintzberg and Waters (1985)

“intend-ed” strategies are distinguished from “emergent” strategies The

former are associated with precise intentions by the

organiza-tion and occur before acorganiza-tion, while the latter re Xect the absence

of intentions and occur during action Both types can lead to

the notion of “realized strategies”.

choice level The RBV rests on the principle that

competitiveness is a function of the strength,

expert exploitation, and leveraging of speciWc

internal resources and capabilities controlled by a

Wrm (Lengnick-Hall & WolV, 1999) These

resources and capabilities are distinctive, valuable,

and must be protected from imitation, adoption, or

substitution by competitors to create a sustainable

competitive advantage (Barney, 1991; Wernerfelt,

1984) They support strategic choices by providing

the competitive advantage necessary to materialize

these choices MCS must be aligned with

capabili-ties to be eVective and consistent with strategic

choices Hence, the notion of strategic choice itself

may not be directly traceable to MCS Instead, the

relationship should be examined between

capabili-ties and MCS, rather than between strategic choice

and MCS

Second, the traditional role of MCS in the

implementation of strategy is commonly

recog-nized (e.g Andrews, 1971; AnsoV, 1965; Anthony,

1965) Following the work of Simons (Simons,

1990, 1991, 1994, 1995), several studies have

exam-ined a more active role of MCS in the formulation

of strategy and the implementation of strategic

change (e.g Abernethy & Brownell, 1999; Bisbe &

Otley, 2004; Chenhall & LangWeld-Smith, 2003)

Another line of research describes how the

organi-zations balance the traditional and more active

roles of MCS (e.g Ahrens & Chapman, 2004;

Chapman, 1998; Dent, 1987; Haas & Kleingeld,

1999) However, less attention has been devoted to

the eVects of dynamic tension resulting from the

balance use of MCS in various ways Notable

exceptions are the work of Chenhall and Morris

(1995) and Marginson (2002) While the former

has examined the joint eVect of organic processes

and formal MCS on performance, the latter has

used the model of Simons to report some trade-oVs

resulting from the various uses of MCS A more

complete understanding of the relationships

between MCS and strategy requires the integration

in the theoretical and empirical analyses of both

traditional and more active roles of MCS, as well

as the tension resulting from those uses

Building on the work of Simons, this study aims

to examine, from a resource-based perspective,

how the use of MCS by top management team can

act as an antecedent to organizational capabilitiesleading to strategic choices SpeciWcally, this studyfocuses on the traditional feedback role of MCS tosupport the implementation of strategy (‘diagnos-tic use’) and the more active role of MCS associ-ated with the signals sent throughout the Wrm tofocus organizational attention, stimulate dialogueand support the emergence of new strategies(‘interactive use’) These two types of use worksimultaneously but for diVerent purposes Collec-tively, their power lies in the tension generated bytheir balanced use which simultaneously reXects anotion of competition and complementarity.Hence, three speciWc research questions areinvestigated in this study: (i) To what extent do thediagnostic and interactive uses of MCS contribute

speciWcally to the creation and maintenance of

capabilities leading to strategic choices? (ii) Towhat extent do the diagnostic and interactive uses

of MCS act in combination to produce dynamic

tension which contributes to the creation andmaintenance of these capabilities? (iii) To whatextent does the use of MCS contribute to organiza-tional performance? A theoretical model is devel-oped and tested with empirical data gathered from

a survey

The remainder of this paper is organized as lows The next section brieXy examines theresource-based view and the use of MCS followingthe model of Simons Thereafter, a theoreticalmodel is developed and a set of hypotheses is pre-sented The next two sections include a description

fol-of the survey design, the analysis fol-of the data usingstructural equation modelling and a discussion ofthe results The Wnal section presents the theoreti-cal contributions, practical implications, limita-tions and insights for future research

Theoretical framework

DeWnition of constructs

Resource-based view and capabilities

The RBV conceptualizes Wrms as bundles ofresources heterogeneously distributed across Wrms,and that resource diVerences persist over time(Amit & Schoemaker, 1993; Wernerfelt, 1984)

Resources that are valuable, rare, inimitable and

non-substitutable lead to the achievement of

sus-tainable competitive advantage that cannot be

eas-ily duplicated by competitors (Barney, 1991)

Resources include various elements that can be

used to implement value-creating strategies:

spe-ciWc physical assets (e.g., specialized production

facilities, geographic location), human resources

(e.g., engineering experience, expertise in

chemis-try), organizational assets (e.g., management skills,

superior sales force), and competencies (e.g.,

minia-turization, imaging) (Barney, 1991; Eisenhardt &

Martin, 2000; Teece, Pisano, & Shuen, 1997).2

Capabilities forge a link between resources and

permit their deployment (Day, 1994) They are the

organizational processes by which Wrms synthesize

and acquire knowledge resources, and generate

new application from those resources (Kogut &

Zander, 1992) Formally stated: “The Wrm’s

pro-cesses that use resources—speciWcally the processes

to integrate, reconWgure, gain and release

resources—to match and even create market

change Dynamic capabilities thus are the

organi-zational and strategic routines by which Wrms

achieve new resource conWgurations as market

emerge, collide, split, evolve, and die.” (Eisenhardt

& Martin, 2000, p 1107)

Innovation, organizational learning, market

ori-entation and entrepreneurship are recognized as

primary capabilities to reach competitive

advan-tage, to match and create market change Past

research suggests that each of these four

capabili-ties is adequate to oVer strengths, but is not

suY-cient to develop sustained advantages Only

collectively can they help a Wrm to be uniquely

competitive (Bhuian, Menguc, & Bell, 2005; Hult &

Ketchen, 2001; Hurley & Hult, 1998; Ireland et al.,

2001) Hence, this paper investigates the inXuence

of MCS on each of these four capabilities

First, innovativeness refers to the notion of the

organization’s openness to new ideas, products

and processes, and its orientation toward

innova-tion (Hurley & Hult, 1998) Innovation is

consid-ered by many scholars and managers to be criticalfor Wrms to compete eVectively in domestic andglobal markets, and one of the most importantcomponents of a Wrm’s strategy (Hitt, Ireland,Camp, & Sexton, 2001) Firms that have a greatercapacity to innovate are able to develop a competi-tive advantage, achieve corporate renewal andachieve higher levels of performance (Danneels,2002; Hurley & Hult, 1998)

Second, organizational learning refers to thedevelopment of insights, knowledge and associa-tions among past actions, the eVectiveness of theseactions, and future actions (Fiol & Lyles, 1985) Anorganization’s ability to survive and grow is based

on advantages that stem from capabilities that resent collective learning (Nevis, Dibella, & Gould,

rep-1995) Learning is considered to be an importantfacilitator of competitive advantage by improving

a Wrm’s information processing activities at a fasterrate than rivals do (Baker & Sinkula, 1999).Third, market orientation refers to the organi-zational emphasis on customers’ expressed needsand on the development of long-term thinkingbased on customers’ latent needs (Slater & Narver,1998; Slater & Narver, 1999) It speciWcally relates

to three components, namely customer orientation,competitor orientation and inter-functional coor-dination Market orientation eVectively and

eYciently creates the necessary behaviors for thecreation of superior value for customers, and thus,continuous performance for the business (Kohli &Jaworski, 1990; Narver & Slater, 1990)

Fourth, entrepreneurship refers to the ability ofthe Wrm to continually renew, innovate, and con-structively take risks in its markets and areas ofoperation (Miller, 1983; Naman & Slevin, 1993).Entrepreneurial actions entail creating newresources or combining existing resources in newways to develop and commercialize new products,move into new markets, and/or service new custom-ers (Hitt et al., 2001) Entrepreneurship is identiWed

as a critical organizational process that contributes

to Wrm survival and performance (e.g., Barringer &Bluedorn, 1999; Hitt et al., 2001; Miller, 1983)

Use of management control systems

MCS are deWned as formalized procedures andsystems that use information to maintain or alter

2 The resources must be distinguished from factors of

produc-tion which are undi Verentiated inputs available in disaggregate

form in factor markets, such as land, unskilled labour and

capi-tal.

patterns in an organizational activity (Simons,

1987) This deWnition includes planning systems,

reporting systems, and monitoring procedures that

are based on information use In this study, one

component of MCS is examined, namely the

per-formance measurement systems (PMS) The latter

represent a set of metrics used to quantify actions

(Neely, Gregory, & Platts, 1995) These metrics can

be Wnancial or non-Wnancial, internal or external,

short or long term as well as ex post or ex ante

Simons’ framework on the levers of control

(Simons, 1990, 1991, 1994, 1995) relies on the

con-cept of tension The essence of MCS is to manage

the inherent organizational tension between creative

innovation and predictable goal achievement

More speciWcally, three kinds of inherent tension

must be reconciled and balanced to allow the

eVec-tive control of business strategy: (i) unlimited

opportunity versus limited attention, (ii) intended

versus emergent strategy, and (iii) self-interest and

desire to contribute Managers use MCS as

posi-tive and negaposi-tive forces to create dynamic tension

that contributes to manage inherent organizational

tension

The diagnostic use of MCS represents the

tradi-tional feedback role as MCS are used on an

excep-tion basis to monitor and reward the achievement

of pre-established goals Following a traditional

mechanistic notion of control, a diagnostic use

provides motivation and direction to achieve goals

by focusing on and correcting deviations from

pre-set standards of performance The diagnostic use

comprises the review of critical performance

vari-ables (i.e., factors enabling the achievement of

intended strategy) to monitor and coordinate the

implementation of intended strategies It

repre-sents a negative force for two reasons On the one

hand, diagnostic use focuses on mistakes and

nega-tive variances On the other hand, the sign of the

deviation that is derived when outputs and goals

are compared is reversed in the feedback signal to

adjust the process

The interactive use of MCS represents a positive

force as MCS are used to expand

opportunity-seeking and learning throughout the organization

The interactive use focuses attention and forces

dialogue throughout the organization by reXecting

signals sent by top managers It stimulates the

development of new ideas and initiatives andguides the bottom–up emergence of strategies byfocusing on strategic uncertainties (i.e., contingen-cies threatening or invalidating underlyingassumptions of current strategies) When MCS areused interactively, (i) the information generated is

a recurrent and important agenda for top ers; (ii) frequent and regular attention is fosteredthroughout the organization; (iii) data are dis-cussed and interpreted among organizationalmembers of diVerent hierarchical levels; and (iv)continual challenge and debate occur concerningdata, assumptions and action plans

manag-Diagnostic and interactive uses of MCS, ing PMS, represent two complementary and nesteduses They work simultaneously but for diVerentpurposes While diagnostic use represents a mecha-nistic control used to track, review and support theachievement of predictable goals, interactive use is

includ-an orginclud-anic control system supporting the gence of communication processes and the mutualadjustment of organizational actors SpeciWcally, adiagnostic use limits the role of PMS to a measure-ment tool, while an interactive use expands its role

emer-to a strategic management emer-tool (Kaplan & Norton,

2001) According to Simons, diagnostic and active uses of MCS represent countervailing forcesused to balance the inherent organizational ten-sion Haas and Kleingeld (1999) point out thatdiagnostic use of PMS may not be an end in itselfbut a means necessary to initiate strategic dialogueand interactive use of PMS Referring to Argyrisand Schön (1978b), diagnostic use represents sin-gle-loop learning and acts as a prerequisite forinteractive use and double-loop process Thus, theuse of MCS (and PMS) ranges from mostly diag-nostic to a combination of diagnostic and interac-tive

The joint use of MCS in a diagnostic and active fashion to manage inherent organizationaltensions creates dynamic tension Dynamic tensiondenotes contradictory but interrelated elements(Lewis, 2000) Formally stated, tension can be deW-ned as two phenomena in a dynamic relationshipthat involve both competition and complementar-ity (English, 2001) The joint use of PMS in a diag-nostic and interactive manner creates dynamictension reXecting competition (positive versus

inter-negative feedback) and complementarity (focus on

intended and emergent strategies) The notion of

dynamic tension is not necessarily new in the

aca-demic literature, and is related to other terms such

as conXict, paradox, dilemma, and contrast

(English, 2001) For instance, some authors have

examined the paradox related to the propensity to

seek risk and innovation while simultaneously

exe-cuting a safe and incremental implementation (e.g

Bourgeois & Eisenhardt, 1988; Cameron, 1986)

Other studies have examined conXicts in the use

and implementation of control and cost systems

(e.g Barrett & Fraser, 1977; Chenhall, 2004;

Shank, Niblock, & Sandalls, 1973) As suggested

by the conXict literature, tension is not necessarily

negative but instead may be beneWcial to

organiza-tions (DeDreu, 1991; Nicotera, 1995) This study

investigates the inXuence of the dynamic tension

resulting from the joint use of PMS in a diagnostic

and interactive fashion on capabilities leading to

strategic choices

Theoretical model and hypotheses

Fig 1 presents a summary of the theoretical

model that reXects the relationships among two

PMS use (diagnostic and interactive), four

capabil-ities (innovativeness, organizational learning,

mar-ket orientation and entrepreneurship), and

organizational performance As previously

men-tioned, the aim of this paper is to understand the

speciWc and joint contributions of two tary uses of PMS on capabilities and performance.Consequently, the theoretical model considers theindividual eVect of diagnostic and interactive usesseparately, as well as their collective eVects Whenexamined speciWcally, a diagnostic use is expected

complemen-to have a negative inXuence on the four ties, while interactive use is expected to have a pos-itive impact on these capabilities Furthermore, thebalanced use of PMS in a diagnostic and interac-tive fashion results in dynamic tension This ten-sion is expected to contribute positively to the fourcapabilities by ensuring that positive eVects ofinteractive use will be achieved and by expandingthese positive eVects Lastly, PMS use is expected

capabili-to have an indirect eVect on organizational mance through the four capabilities These rela-tionships are discussed speciWcally below

perfor-Relationships between diagnostic use and capabilities

In the management of inherent organizationaltension between creative innovation and predict-able goal achievement, diagnostic use of PMS sup-ports the attainment of pre-established goals.Indeed, diagnostic use is described as a negativeforce that creates constraints and ensures compli-ance with orders: “[Diagnostic systems] constraininnovation and opportunity-seeking to ensure pre-dictable goal achievement needed for intendedstrategies” (Simons, 1995, p 91) Traditional PMS

Fig 1 Theoretical model.

Market orientation

ship

Entrepreneur- ness

Innovative-Organizational learning

CAPABILITIES PMS interactive use

PMS diagnostic use H1

(-)

Diagnostic * Interactive

Organizational performance

H2 (+)

H4 (+)

Dynamic tension

H3 (+)

encourage conservatism and a “playing it safe”

attitude: “Managers need to be encouraged to

identify deWned areas within which a degree of

experimentation and risk-taking might be bene

W-cial Too often we stiXe creativity and learning by

insisting upon good performance from all

activi-ties” (Otley, 1994, p 297)

Relying on cybernetic logic and reXecting

tradi-tional control systems, diagnostic use of PMS may

not represent an adequate means to foster

capabil-ities of market orientation, entrepreneurship,

inno-vativeness and organizational learning Diagnostic

use reXects two important features associated with

mechanistic controls: (i) tight control of operations

and strategies, and (ii) highly structured channels

of communication and restricted Xows of

informa-tion (Burns & Stalker, 1961) Globally, there is a

mismatch between the requirements of the four

capabilities and mechanistic use of control systems

(Chenhall & Morris, 1995; Galbraith, 1982)

First, diagnostic use is associated with tight

con-trol of operations and strategies through

sophisti-cated control systems These systems include

action plans derived from strategies, detailed

Wnan-cial targets, comparison of actual outcomes with

targets, and explanation of variances This formal

use of PMS provides a mechanistic approach to

decision making resulting in organizational

inat-tention to shifting circumstances and the need for

innovation (Van de Ven, 1986) Furthermore, the

concept of organizational learning encompasses

the notion of single- and double-loop learning

(Argyris & Schön, 1978a) Diagnostic use

repre-sents single-loop learning but not the higher level

learning (double-loop), which is necessary for

innovative behaviors (Haas & Kleingeld, 1999)

Also, the four capabilities may create an

organiza-tional momentum leading to innovative excess,

overzealous experimentation and diminished

returns Diagnostic use of PMS is used to signal

when productivity and eYciency have fallen, and

when innovation needs to be curbed (Miller &

Friesen, 1982) Hence, PMS is used diagnostically

to limit the deployment of the four capabilities by

providing boundaries and restrict risk-taking

Lastly, as a mechanistic control, diagnostic use has

been associated with several dysfunctional

behav-iors based on distortion of information: gaming,

smoothing, biasing, focusing, Wltering, and illegalacts (Birnberg, Turopolec, & Young, 1983; Hofst-ede, 1978; Simons, 1995) These distortions consti-tute defensive routines that aim to reduce potentialembarrassment or threat, or to improve personalinterest They consequently impede the potentialfor learning and innovation (Argyris, 1990).Second, diagnostic use of PMS is associatedwith highly structured channels of communicationand restricted Xow of information However,notions of communication and dialogue gravitatetowards the four capabilities They rely on cross-functional processes, and thus require the free Xow

of information and open channels of tion (Kohli & Jaworski, 1990) Diagnostic useundercuts the commitment of organizationalactors to these cross-functional processes by rein-forcing the existing lines of authority and responsi-bility (Abernethy & Brownell, 1999) AsVandenbosch (1999) argued, the discussion trig-gered by the diagnostic use leads to correctiveaction at best At worst, it causes discussion togravitate towards unproductive topics, such as thebelievability of the numbers or why things are notbetter, and ultimately does not trigger any action.Corrective actions are not suYcient to sustain suchcapabilities; new ideas must be developed Thesearguments lead to the following hypothesis:

communica-Hypothesis 1 A diagnostic use of PMS tends tonegatively inXuence capabilities of market orienta-tion, entrepreneurship, innovativeness and organi-zational learning

Relationships between interactive use and capabilities

In the management of inherent organizationaltension between creative innovation and predict-able goal achievement, interactive use of PMS sup-ports the development of ideas and creativity.Indeed, interactive use has the power to represent apositive trigger that fosters creative and inspira-tional forces: “ƒsenior managers use interactivecontrol systems to build internal pressure to breakout of narrow search routines, stimulate opportu-nity-seeking, and encourage the emergence of newstrategic initiatives” (Simons, 1995, p 93) Accord-ing to Dent (1990), curiosity and experimentation

can be fostered by control systems Planning and

control systems could create new images of the

organization for employees as the organization

interacts with its environment Thus, obsolete

par-adigms and organizational attempts can be

uncou-pled (unlearning) and recouuncou-pled in diVerent ways

(learning)

Relying on organizational dialogue and

signal-ing, interactive use of PMS represents an adequate

means to foster capabilities of market orientation,

entrepreneurship, innovativeness and

organiza-tional learning Interactive use reXects two

impor-tant features associated with organic controls: (i)

loose and informal control reXecting norms of

cooperation, communication and emphasis on

get-ting things done, and (ii) open channels of

commu-nication and free Xow of information throughout

the organization (Burns & Stalker, 1961) Globally,

there is a natural Wt between the requirements of

the four capabilities and organic use of control

sys-tems (Chenhall & Morris, 1995; Van de Ven, 1986)

Capabilities of innovativeness, organizational

learning, entrepreneurship and market orientation

lead to complexity and changes in product design

This context requires the employment of experts in

the process of creation and implementation of new

product design (Burns & Stalker, 1961; Mintzberg,

1979) The collaboration of experts and managers

from diVerent functional areas is needed to foster

innovation and new product development (Miller,

1988) Reciprocal interdependencies are then

expected from the people who need to be in close

contact (Galbraith, 1973) Also, this context of

complexity and change brings uncertainty and

ambiguity for the sub-ordinates as top

manage-ment is often redeWning goals and objectives

(Abernethy & Brownell, 1999) In those

circum-stances, additional pressure is imposed on the

organization’s information processing capacity

and more interaction between top management

and sub-ordinates is required to increase the Xow

of information (Galbraith, 1973)

The interactional needs and the information

processing capacity necessary for the capabilities

are likely to be fostered by an interactive use of

PMS Indeed, in providing an agenda and a forum

for the regular face-to-face debate and dialogue, an

interactive use allows top management to send

sig-nals that stimulate and concentrate organizationalattention toward top management preferences,strategic uncertainties and organizational goalsand objectives (Simons, 1995) Considering thecharacteristics of integrativeness within PMS, topmanagement can provide an understanding ofcause–eVect relationships between operations,strategy and goal, as well as between variousaspects of the value chain (Chenhall, 2005) Also,with a focus on dialogue and communicationbetween organizational actors of diVerent or iden-tical hierarchical levels, the interactive use of PMSacts as an integrative liaison device that breaksdown the functional and hierarchical barriers thatrestrict the Xow of information (Abernethy &Brownell, 1999; Abernethy & Lillis, 1995) Lastly,

by focusing regular attention on strategic tainties, interactive use of PMS provide a lever toWne-tune analyses and actions, and alter strategy ascompetitive markets change (Bisbe & Otley, 2004)

uncer-In terms of information processing activities,Kohli and Jaworski (1990) identify three basiccomponents, namely intelligence generation, intel-ligence dissemination, and responsiveness Simi-larly, Huber (1991) speciWes four processes:knowledge acquisition, knowledge distribution,information interpretation and organizationalmemory.3 Therefore, internal mechanisms must be

in place: (i) to ensure knowledge generationthroughout the organization, (ii) to communicate,disseminate and sell this knowledge throughoutthe organization, and (iii) to plan actions and coor-dinate their implementation (Kohli & Jaworski,

1990) An interactive use of PMS has the power tofocus organizational attention on the speciWc stra-tegic uncertainties for which knowledge must begenerated and cause–eVect relationships under-stood PMS is an important formal mechanismused to collect information to develop capabilities

3 Intelligence generation, knowledge acquisition, and mation interpretation refer to the collection and assessment of information Intelligence dissemination and knowledge distri- bution refer to the process by which information is shared throughout the organization Responsiveness is the action tak-

infor-en in response to the knowledge gained and shared, while nizational memory refers to the means by which knowledge becomes institutionally available and stored for future use ( Huber, 1991; Kohli & Jaworski, 1990 ).

orga-(Chenhall, 2005) Moreover, by fostering

organiza-tional dialogue and debate, and encouraging

infor-mation exchange, interactive use contributes to

knowledge dissemination, information distribution

and communication, and the emergence of

strate-gic actions (Haas & Kleingeld, 1999; Malina &

Selto, 2001; Simons, 1995) Hence, an interactive

use of PMS contributes to expanding the

organiza-tion’s information processing capacity and

foster-ing interaction among organizational actors

Consequently, an interactive use fosters the

deployment of the four capabilities Formally

stated:

Hypothesis 2 An interactive use of PMS tends to

positively inXuence capabilities of market

orienta-tion, entrepreneurship, innovativeness and

organi-zational learning

Relationships between joint use of PMS and

capabilities

As illustrated by the previous two hypotheses,

interactive use of PMS stimulates

opportunity-seeking and fosters dialogue, while diagnostic use

creates constraints and ensures compliance with

orders Together, diagnostic and interactive uses

create a dynamic tension which has two eVects: (i)

ensuring that positive eVects of interactive use on

capabilities will be achieved; and (ii) expanding

these positive eVects of interactive use

First, a diagnostic use of PMS ensures that the

positive eVects of interactive use on capabilities

will be achieved In some circumstances, the

poten-tial beneWts of interactive use may vanish due to

insuYcient diagnostic use to set boundaries and to

highlight eVectiveness issues This can produce a

loss of direction, wasted energy and a disruption of

continuity (Cameron, 1986; Chenhall & Morris,

1995) Similarly, the potential beneWts of

interac-tive use can be lost due to excessive diagnostic use

which constrains innovation and risk taking This

can produce stagnation, loss of energy and

declin-ing morale (Cameron, 1986; Chenhall & Morris,

1995)

More importantly, a diagnostic use of PMS

helps to increase the positive eVects of an

interac-tive use on capabilities Indeed, beyond underlying

assumptions that conXict and tension are negative

and destructive, growing evidence from the conXictliterature suggests that they may be beneWcial toindividual and organizational performance, andthat avoiding and suppressing conXict reduces cre-ativity, decision quality, product development, andcommunication (DeDreu, 1991; Nicotera, 1995).ConXict and tension foster organizational dia-logue, stimulate creativity, and focus organiza-tional attention (Amason, 1996; Tjosvold, 1991;DeDreu, 1991; English, 2001; VanSlyke, 1999).These three elements, which have been presented

as positive eVects of interactive use on capabilities(see Hypothesis 2), are ampliWed by the combina-tion of diagnostic and interactive use They are dis-cussed more speciWcally below

Dynamic tension between diagnostic and active use of PMS stimulates organizational dia-logue (Dent, 1987) It provides the opportunity fordialectically styled interactions by providing ameans to debate vigorously opposing positions(Chenhall, 2004) More speciWcally, dynamic ten-sion provides valuable information that increasesXexibility, innovation, and improvement It stimu-lates continual communication concerning strategicissues and promotes mutual understanding Ten-sion also encourages open and lively discussions,and helps employees group their ideas and actions(Amason, 1996; DeDreu, 1991; Tjosvold, 1991).Moreover, creativity is enhanced by dynamictension, which leads organizational members tointegrate seemingly opposed elements (VanSlyke,

inter-1999) Tension triggers the identiWcation of native ways of doing things by supporting theidentiWcation and synthesis of a variety of view-points (Chenhall, 2004) Finally, dynamic tensionresulting from the balanced use of PMS in a diag-nostic and interactive fashion contributes to focus-ing organizational attention Indeed, tensionmakes underlying issues explicit and helps groups

alter-to deWne their boundaries Thus, it provides themotivation and strength to deal with tough prob-lems Tension also fosters involvement andempowerment by providing incentives for diVerentgroups to pull together toward a common goal(Amason, 1996; DeDreu, 1991; Tjosvold, 1991)

To summarize, the joint eVect of a balanced use

of PMS diagnostically and interactively constitutescountervailing forces that create dynamic tension

This tension ensures the achievement of the

posi-tive eVects of interactive use on capabilities

Dynamic tension also increases these positive

eVects by fostering organizational dialogue,

stimu-lating creativity, and focusing organizational

attention Formally stated:

Hypothesis 3 The dynamic tension resulting from

a balanced use of PMS in a diagnostic and

interac-tive fashion tends to posiinterac-tively inXuence the

capa-bilities of market orientation, entrepreneurship,

innovativeness and organizational learning

Relationships between PMS, capabilities and

organizational performance

Following the resource-based view of the Wrm,

unique resources and capabilities lead to a

sus-tained competitive advantage, which in turn

con-tributes to performance diVerences among Wrms

Market orientation, organizational learning,

inno-vativeness, and entrepreneurship constitute four

capabilities that are valuable, hard to duplicate,

and non-substitutable They are considered to be

key drivers of organizational transformation and

strategic renewal by manipulating resources into

new value-creating strategies (e.g., Bhuian et al.,

2005; Danneels, 2002; Eisenhardt & Martin, 2000;

Hitt et al., 2001; Ireland et al., 2001) Empirically,

previous studies provide evidence showing that

these four capabilities contribute positively to

per-formance (e.g., Hult & Ketchen, 2001; Lee, Lee, &

Pennings, 2001; Naman & Slevin, 1993; Narver &

Slater, 1990; Spanos & Lioukas, 2001)

Diagnostic and interactive use of PMS, as well as

the dynamic tension resulting from their balanced

use, have been linked to capabilities of market

ori-entation, organizational learning, innovativeness,

and entrepreneurship (Hypotheses 1–3) These

capabilities are expected to lead to organizational

performance Thus, the use of PMS can be expected

to have indirect implications for performance by

inXuencing the deployment of capabilities which

are considered to be valuable, hard to duplicate,

and non-substitutable Hence, diagnostic and

inter-active use of PMS and the dynamic tension

result-ing from their balanced use inXuence the four

capabilities, which in turn increase performance

Therefore, the following hypotheses are proposed:

Hypothesis 4a The diagnostic and interactive use

of PMS have an indirect eVect on organizationalperformance through their contribution to capa-bilities of market orientation, entrepreneurship,innovativeness and organizational learning

Hypothesis 4b The dynamic tension resulting from

a balanced use of PMS in a diagnostic and tive fashion has an indirect eVect on organizationalperformance through its contribution to capabili-ties of market orientation, entrepreneurship, inno-vativeness and organizational learning

interac-No speciWc hypotheses supporting a direct tionship between PMS use and performance havebeen formulated Despite the fact that priorresearch has examined the relationship betweenMCS and performance using a notion of Wt to thecontext of the organization (e.g., Govindarajan,1988; Govindarajan & Fisher, 1990; Perera, Harri-son, & Poole, 1997; Sim & Killough, 1998), anddespite the fact that another line of research hassupported a positive relationship between thedesign of PMS (increased reliance on non-Wnancialinformation) and performance (e.g Baines & Lang-Weld-Smith, 2003; Davila, 2000; Said, Elnaby, &Wier, 2003; Scott & Tiesen, 1999), the exact nature

rela-of the relationship between the use rela-of PMS and formance remains ambiguous Theoretical supportand prior empirical evidence in the literature areinsuYcient to justify a direct relationship betweenPMS use and performance at an organizationallevel (Bisbe & Otley, 2004) Also, recent studiesusing the Simons’ framework did not Wnd empiricalevidence supporting a direct relationship betweenthe interactive use of MCS and performance (Aber-nethy & Brownell, 1999; Bisbe & Otley, 2004).Furthermore, according to the resource-basedview, information and control systems are gener-ally not a source of competitive advantage for tworeasons: (i) they lead Wrms to fully realize the ben-

per-eWts of the resources they control but do not erate sustainable rents, and (ii) they can be readilytransferred (Barney et al., 2001) Hence, followingthis line of reasoning, PMS use may not contrib-ute directly to performance, but instead contributeindirectly through capabilities On the other hand,the accounting literature has demonstrated theways in which the use of MCS aVects their role

gen-and functioning as well their impact within the

organizations (e.g., Abernethy & Brownell, 1999;

Ahrens & Chapman, 2004; Bisbe & Otley, 2004;

Chapman, 1998; Chenhall & Morris, 1995;

Simons, 1995) From a theoretical standpoint and

following resource-based logic, it could be argued

that the use of PMS in a joint diagnostic and

inter-active fashion has a positive impact on

perfor-mance Indeed, the balance between diagnostic

and interactive use may be considered as a

capa-bility In this regard, the ability to reach a balance

between two opposing uses of PMS which,

simul-taneously, try to stimulate innovation while

searching for predictable achievements represents

a capability that is valuable, distinctive, and

imperfectly imitable The ability to manage the

combination of diagnostic and interactive use

depending on various internal and external factors

is complex and may not be readily transferred

These opposing views from the RBV and MCS

lit-erature combined with the absence of substantive

empirical evidence preclude the formulation of

any hypotheses Nevertheless, in order to

contrib-ute to this debate and to expand current literature,

the links between diagnostic and interactive use of

PMS and dynamic tension versus performance

will be tested Since a large proportion of the

rela-tionships between PMS use and performance is

expected to come indirectly through the four

capabilities (Hypothesis 4), the direct eVects (if

any) are expected to be relatively small

Validation of the model

Various internal and external contextual factors

interact together to cause uncertainty As the level

of uncertainty varies, diVerent forms of

communi-cation are necessary to mobilize and integrate

information (Chapman, 1997) Environmental

context, organizational size and organizational

culture4 are important contextual factors which

inXuence the role of PMS (e.g Bhimani, 2003;

Chenhall, 2003; Henri, in press; Hoque & James,

2000) These variables also suggest conXicting

implications and potential tension (Dent, 1987;

Quinn & Cameron, 1983).5 While size favorsbureaucratic formalization, the complex andchanging environmental context calls for Xuidityand Xexibility in the management practice (Dent,

1987) On the other hand, following the values model, there are likely to be simultaneouspressures for control values fostering order andformality and Xexibility values reXecting adapt-ability and responsiveness (Quinn & Cameron,1983; Quinn & Kimberly, 1984) As the diagnosticand interactive use of PMS fulWll diVerent roles,their use and the dynamic tension emerging fromtheir balanced use might vary depending on thelevel of uncertainty In order to validate therobustness of the theoretical model, sub-groupanalyses are used to assess cross-sample validationand to reinforce the hypothesis tests Splitting thesample at the median for each contextual variable,two sub-samples will be created and compared

competing-Methododology

Research design

Data were collected through a structured tionnaire sent to one member of top managementteams (CEO, COO, CFO, or senior vice-presidents).The survey implementation followed four steps: pre-notiWcation, initial mailing, Wrst follow up, and sec-ond follow up To generate early interest, the Wrststep was to notify respondents in the form of a let-ter, phone call or e-mail A mail-out package includ-ing a cover letter, the questionnaire and a businessreply envelope was then sent to every contact name

ques-In a few cases, the questionnaire was sent by fax ore-mail The Wrst follow up consisted of a postcardreminder which was sent to every respondent, whilethe second was a phone call or replacement ques-tionnaire sent only to those who had not answered.The target population consisted of 2175 Cana-dian manufacturing Wrms listed in Scott’s 2002database with primary and secondary SIC codes in

4 Organizational culture is de Wned here as the shared values

that interact with an organization’s structures and control

sys-tems to produce behavioural norms Uttal and Fierman (1983)

5 Dent (1987) also proposed task unpredictability as an important contextual factor in a context of tension on the de- sign of MCS Since the current study examines phenomena at

an organizational level and task unpredictability is an ual-level construct, the latter is not included in the analyses.

individ-the range of 21–39 These Wrms were either

inde-pendent companies or SBUs However, the lack of

a contact name for the top management teams in

several cases reduced the number of usable Wrms in

the target population to 1692 The Wrms were large

enough to ensure that organizational and strategy

variables applied (Miller, 1987), and to ensure that

a formal PMS was in place (Bouwens &

Abern-ethy, 2000) Thus, the Wrms selected in the sample

respected the following two criteria: (i) sales were

at least $20 million Canadian; and, (ii) each had at

least 150 employees

Current literature reports that a sample size

vary-ing between 100 and 200 cases, or between 5 and 10

subjects per estimated parameter, is adequate for

small-to-medium size structural equation models

(Anderson & Gerbing, 1988; Bentler & Chou, 1987)

In the present case, a total of 383 Wrms participated

to the study giving a response rate of 24%, which is

similar to the 15–25% range reported in similar

recent studies (e.g., Baines & LangWeld-Smith, 2003;

Lee et al., 2001; Spanos & Lioukas, 2001).6

More-over, a ratio of 6.17 subjects per parameter was

obtained, which is adequate to test the proposed

model Appendix 1 presents the statistics of the

respondents in terms of position, experience, size

(number of employees) and industry classiWcation

To test whether the respondents diVered from

the non-respondents, a two-step analysis was

con-ducted Respondents were Wrst compared with

non-respondents in terms of sample characteristics

(size, location, industry) Next, early and late

respondents were compared to detect any

diVer-ence in the mean score of each variable.7 Using

chisquare statistics, no signiWcant diVerences

(p > 0.01) were found between the size, location

and industry of respondent Wrms and dent Wrms A comparison of the means of the vari-ables found little diVerence between early and late

non-respon-respondents The t-value for only one variable is

signiWcant (organizational learning, t D 2.27,

p < 0.05), but this is not believed to be a serious

problem considering its isolated eVect While there

is unlikely to be a systemic bias due to diVerencesbetween respondents and non-respondents, thegeneralization of results related to organizationallearning should be made with caution

Measurement of constructs

All measures are drawn from existing ments Descriptive statistics of the constructs andcorrelation matrix are presented in Table 1.Appendix 2 shows the questionnaire items, Cron-bach Alpha for each construct, and statistics from

instru-a conWrmatory factor analysis (Wrst- and order loadings, and goodness-of-Wt indices8).Interactive and diagnostic uses of PMS are mea-sured using an adapted version of the Van-denbosch’s (1999) instrument Developed originally

second-to measure the use of executive support systems(EES), this instrument is based on several dimen-sions, notably score keeping (diagnostic) and atten-tion-focusing (interactive) The choice of thisinstrument is justiWed by its development based ontheories of accounting control, including Simons(1990), before its adaptation to a management-information context Furthermore, EES is used as asurrogate for accounting and management infor-mation and is restricted to the accounting, manage-ment and control information provided Thus,PMS and EES have common base that allow theadaptation of the instrument to our speciWc con-

6 The response rate was calculated as the percentage of usable

returned questionnaires in relation to the number of

question-naires sent, after adjusting for the Wrms which had closed, ended

manufacturing activities or moved, or for which the contact

person had left the organization As discussed in Appendix 3 , to

assess interrater reliability for survey items, duplicate surveys

were sent to a second member of the top management team in

the Wrms that originally returned the questionnaire Twenty-one

questionnaires were returned For those 21 Wrms having two

di Verent respondents, a mean score has been computed.

7 Early respondents correspond to the managers that have

Wlled out the questionnaire before the Wrst follow-up Late

respondents correspond to the managers that have Wlled out the

questionnaire after the second follow-up.

8 The indices used to assess the model are among the most frequently reported, namely NNFI (non-normed Wt index), CFI (comparative Wt index), and RMSEA (root mean square error

of approximation) The threshold values recommended are (i) NNFI > 0.90 Tabachnick & Fidell (2001) , (ii) CFI > 0.95 Hu & Bentler (1995) , and (iii) RMSEA < 0.l0 ( Browne & Cudeck (1993)

text Two items were added to the interactive

dimension to better reXect its use in a context of

MCS All Wrst- and second-order loadings are

sig-niWcant (p < 0.01), the Cronbach Alpha coeYcients

exceeded common cut-oV level of 0.70 (Nunnally,

1967), and the goodness-of-Wt indices respected the

recommended threshold values

Dynamic tension is operationalized as a product

term between diagnostic and interactive use A

product term can be treated as a variable without

any theoretical meaning (to test an interaction) or as

a construct based on a theoretical justiWcation (

Cor-tina, Chen, & Dunlap, 2001) In this study, the

prod-uct term is treated as a constrprod-uct having its own

theoretical meaning Several procedures are

avail-able to create and test multiplicative terms within

structural equation models The seminal work of

Kenny and Judd (1984) provides the foundation for

these procedures Among various approaches,

Jac-card and Wan (1995) propose a procedure having

the same logic as Kenny and Judd but simpler to

implement and adapted for the latest versions of

LISREL Essentially, all possible cross products of

the existing indicators are used as indicators of the

latent product.9 Although this procedure is one ofthe most technically robust, it is also one of the mostcomplicated ones (Cortina et al., 2001).10

Four diVerent validated scales are used to sure internal capabilities The well-establishedMKTOR instrument developed by Narver andSlater (1990) is used to measure market orientation.The instruments proposed by Naman and Slevin(1993) and Hult (1998) are respectively used tomeasure entrepreneurship and organizationallearning Finally, following Hurley and Hult (1998),the instrument developed by Burke (1989) is used

mea-Table 1

Descriptive statistics and correlation matrix

Notes: The scores of diagnostic and interactive uses have been centered Before centering, the mean scores (standard deviations) for

diagnostic and interactive were, respectively, 5.63 (0.98) and 5.07 (1.05).

¤ SigniWcant at the 0.05 level; ¤¤ SigniWcant at the 0.01 level.

Market orientation

neurship

Entrepre- tiveness

Innova-Organizational learning

Diagnostic use1

Interactive use1

Dynamic tension

ci Wcally, the four indicators are: (i) du1 ¤ iu1, (ii) du1 ¤ iu2, (iii) du2 ¤ iu1, and (iv) du2 ¤ iu2 Moreover, with this approach, sev- eral parameters are constrained to equal values determined by various equations, namely the variance of the latent product, the paths from the latent product to its indicators, and the error variances of these indicators.

10 To validate our results and provide a sensibility analysis, the single-indicator approach suggested by Ping (1995) has also been used It has provided similar results and conclusions.

to measure innovativeness All Cronbach Alpha

coeYcients exceeded the common cut-oV level of

0.70, all Wrst- and second-order loadings are

signiW-cant (p < 0.01), and the goodness-of-Wt indices

respect the recommended threshold values

Organizational performance is measured with a

subjective instrument using three indicators: (i)

sales volume; (ii) return on investment; and, (iii)

proWts As several authors argue (e.g., Dess &

Rob-inson, 1984; Venkatraman & Ramanujam, 1987),

in terms of consistently providing valid and

reli-able performance assessment, neither objective nor

subjective measures are superior The three path

loadings are signiWcant (p < 0.01), the Cronbach

Alpha coeYcient is 0.81, and the goodness-of-Wt

indices respect the recommended threshold values

Lastly, the validation variables used to test

robustness of the model are measured as follows

Organizational culture follows a competing-values

model (Quinn & Rohrbaugh, 1983) and is

mea-sured using the instrument designed by (Krakower

& Niwa, 1985).11 Govindarajan’s (1984)

instru-ment is used to assess environinstru-mental uncertainty,

while size is measured using the natural log of the

number of employees

Validation of constructs

Besides the conWrmatory factor analysis (CFA)

discussed above to establish convergent validity,

several other procedures and tests were conducted

to establish the reliability and validity of

con-structs: pre-test of the questionnaire in three steps,

tests of convergence and discriminant validity, and

assessment of interrater reliability Appendix 3

describes these elements and presents the main

results Overall, based on the CFA and other tests,

all constructs reXect strong validity and reliability

Structural equation model

The theoretical model discussed in this study

reXects two features that must be considered whenchoosing a statistical tool: (i) presence of multipleand interrelated dependence relationships, and (ii)presence of latent variables that cannot beobserved directly Structural equation modeling(SEM) represents a set of multivariate techniquesthat allow the simultaneous study of several causalrelationships between endogenous and exogenousvariables (Mueller, 1996) Data collected from thesurvey were analyzed with LISREL 8.52 Consider-ing multivariate non-normality of the data and thepresence of a product term, maximum likelihoodestimates were used (Bentler & Chou, 1987; Cor-tina et al., 2001) Furthermore, composite indicesand a partial disaggregation approach were used

to represent latent construct (Bagozzi & ton, 1994).12 As Landis, Beal, & Tesluk (2000) sug-gest, three indicators were used per latent constructexcept for PMS use Indeed, to minimize conver-gence problems associated with the use of multipli-cative terms in SEM, and to reduce the number ofparameters associated with the product term, twoindicators were used for PMS diagnostic and inter-active use

Heather-Considering the presence of a product term(dynamic tension) in the model, it is usually recom-mended that variables involved in the creation ofthe product term be centered prior to their forma-tion (Cortina et al., 2001; Hartmann & Moers,

1999) Two main reasons justify the use of tion scores First, they minimize identiWcation

devia-11 Respondents were asked to distribute 100 points among the

four ideal cultural types (rational, hierarchical, developmental,

and group) along four dimensions A score was compiled for

each cultural type by averaging the ratings obtained on the four

dimensions An overall measure of culture was developed by

subtracting the mean score of the developmental and group

cul-ture (focus on Xexibility) from the mean score of the rational

and hierarchical culture (focus on control) By representing the

“net value” of the control/ Xexibility dimension, the results

reX-ect the importance of control values for each organization.

12 Composite indices represent aggregates of items which are used as manifest indicators of a latent construct As suggested

by Hall, Snell, and Foust (1999) , items parcelling presents eral advantages First, it tends to provide results that are more reliable and normally distributed, and to have values that are more continuously distributed Furthermore, by reducing sources of contamination, composite indices contribute to the overall Wt of the model Finally, these indices are useful to re- duce the number of parameters in the model and thus contrib- ute to model identi Wcation Using a partial disaggregation approach, each dimension is represented as a separate latent variable indicated by composites of sub-scales Several compos- ites are formed for each dimension in which each composite is a mean of items ( Bagozzi & Heatherton, 1994 ).

sev-problems caused by the correlation between the

variables and the products created from them

Sec-ond, they allow the interpretation of the coe

Y-cients obtained for the lower-order eVects (main

eVects).13 Hence, PMS diagnostic and interactive

use were centered prior to the formation of the

product term

Results

Structural equation models

Table 2 presents the results of two structural

equation models Model A tests the speciWc

rela-tionships between diagnostic and interactive uses

of PMS, capabilities and performance (main

eVects) This model is used to speciWcally test

Hypothesis 1, 2 and 4a Model B introduces the

dynamic tension resulting from a balanced use of

PMS in a diagnostic and interactive fashion

(inter-action term) This model is used to test Hypotheses

3 and 4b speciWcally, and to provide a

complemen-tary testing of Hypotheses 1, 2 and 4a For both

models, goodness-of-Wt indices respect the

recom-mended threshold mentioned previously (see

Foot-note 7) and thus, reXect a good Wt of the data to the

model Table 3 presents the results of six sub-group

analyses performed to validate the robustness of

the theoretical model using environmental

uncer-tainty, size and organizational culture as splitting

variables Every model respects the recommended

threshold mentioned previously

Hypothesis tests

Diagnostic and interactive use, and capabilities

As reXected by models A and B of Table 2,

diag-nostic use of PMS is signiWcantly and negatively

related to capabilities of market orientation,

entre-preneurship, innovativeness and organizational

learning (p < 0.05) The same negative and

signiW-cant relationships are also suggested by the six

sub-group analyses (Table 3), providing strong supportfor Hypothesis 1

Furthermore, in both model A and B, there arestatistically signiWcant positive relationshipsbetween interactive use of PMS and capabilities ofmarket orientation, entrepreneurship, innovative-

ness and organizational learning (p < 0.05) The

same positive and signiWcant relationships are also

reXected by Wve out of six sub-group analyses tive but non-signiWcant relationships being associ-ated to a context of low environmental uncertainty).Hence, Hypothesis 2 also receives strong support

(posi-Dynamic tension and capabilities

Based on model B, Hypothesis 3 does not receivemuch support Indeed, despite positive coeYcients,

a statistically signiWcant relationship betweendynamic tension and the four capabilities does notexist However, the results of the sub-group analy-ses suggest that for Wrms facing high environmentaluncertainty, dynamic tension has a positive and sig-

niWcant eVect on organizational learning (0.129,

p < 0.05) and entrepreneurship (0.09, p < 0.10)

How-ever, in a context of low uncertainty, dynamic sion has a negative eVect on organizational learning

ten-(¡0.290, p < 0.05) Also, Wrms favoring Xexibilityvalues and those favoring control values reXectopposite results In fact, the four capabilities anddynamic tension exhibit positive and signiWcantrelationships for the Xexible Wrms Conversely, sig-

niWcant and negative relationships are foundbetween capabilities and dynamic tension whenWrms reXect control values Globally, no signiWcantrelationship is observed for small or large Wrms.Hence, Hypothesis 3 receives support only for Wrmshaving Xexibility values, and partial support forWrms facing high environmental uncertainty.Hypothesis 3 is reversed for Wrms having controlvalues and to some extent, when Wrms face lowenvironmental uncertainty

Relationships with organizational performance

Despite statistically signiWcant links betweendiagnostic and interactive use of PMS and capabil-ities of market orientation, entrepreneurship, inno-vativeness and organizational learning, hypothesis4a does not receive empirical support Indeed, theresults provided by model A and B show positive

13 Linear transformations do not change the coe Ycient of the

product term, nor its t-statistic and level of signiWcance

( Hartmann & Moers, 1999 ).