The use of organic models of control in JIT ﬁrms ...directory.umm.ac.id/Data Elmu/jurnal/A/Accounting... · led to the adoption of Just-in-Time (JIT) produc-tion systems. Although

The use of organic models of control in JIT ®rms: generalisingWoodward's ®ndings to modern manufacturing practices

Suresh S. Kalagnanam, R. Murray Lindsay *University of Saskatchewan

Abstract

Prior research indicates that technology plays an important role in the determination of management control sys-

tems. A fully developed JIT system represents a radical departure from the traditional approach to organising andmanaging mass production. In probing the management control implications of JIT, this study extends some well-established concepts from organisation theory to the modern manufacturing practices literature to develop a frame-

work which suggests that mass production ®rms adopting JIT (a new technology) must abandon a mechanistic man-agement control system and adopt an organic model of control. Findings from three case studies describing the controlstructures used in JIT ®rms are also presented as part of the theoretical and hypothesis development. In addition,

survey results are reported which are highly consistent with the framework, indicating that Woodward's ®ndings(Woodward J. (1980) Industrial organization: theory and practice (2nd ed.). Oxford: Oxford University Press.) general-ise (are robust) to a new technology. Finally, a preliminary examination into whether improvements in certain keyareas are higher for those JIT ®rms utilising an organic model of management was performed and found to be con-

sistent with expectations. # 1998 Elsevier Science Ltd. All rights reserved.

The current business environment is char-acterised by intense global competition, with ®rmscompeting increasingly not only on the basis ofprice but also on quality, product ¯exibility, andresponse time. These competitive pressures haveincreasingly led organisations to focus on themanufacturing function as being of strategicimportance, providing an important source ofcompetitive advantage. For many ®rms, this hasled to the adoption of Just-in-Time (JIT) produc-tion systems.

Although several studies have examined theimpact of JIT on cost accounting and performancemeasurement systems (e.g. Kaplan, 1986; Neuman& Jaouen, 1986; Foster & Horngren, 1987;

McNair et al., 1989; Bhimani & Bromwich, 1991;Swenson & Cassidy, 1993; Mackey & Thomas,1995), little empirical attention has been placed onthe underlying form of management control that ismost appropriate. The focus of this study is toexamine the relationship between managementcontrol structure and the adoption of JIT manu-facturing technology. From an accounting per-spective, this area of inquiry is important becauseof the interdependence existing between the designof management accounting and control systemsand organisational structure (Otley, 1987).

Prior research indicates that technology playsan important role in the determination of man-agement control systems (Steers, 1988). A fullydeveloped JIT system represents a radical depar-ture from the traditional approach to organising

Accounting, Organizations and Society 24 (1998) 1±30

0361-3682/98/$Ðsee front matter # 1998 Elsevier Science Ltd. All rights reserved.

PII: S0361-3682(98)00062-7

* Corresponding author.

and managing mass production. With the con-tinuous improvement philosophy driving themanagement of the organisation, constant changethroughout the ®rm is sought rather than inhib-ited. Moreover, the increased interdependenciesbrought about by carrying reduced inventorylevels along with the focus on ¯exibility requiresincreased levels of coordination and communica-tion. These considerations would seem to suggestthat a signi®cant change from the bureaucraticmanagement control structure advocated for thetraditional mass production ®rm is necessary (seeMcNair, 1995 pp. E4±10).

In extending some well-established conceptsfrom organizational theory to the modern pro-duction methods literature (JIT, TQM), the ®rstportion of the paper makes a theoretical con-tribution by developing a framework which sug-gests that mass production ®rms adopting JITtechnology must abandon a mechanistic structureand adopt an organic model of control. Theresults of three cases studies examining JIT ®rmsare presented as part of the theoretical andhypothesis development. Following this, theempirical portion of the paper examines whetherJIT mass production companies, relative to tradi-tional mass production ®rms, are adapting theirstructural arrangements in moving towards anorganic model of control. In addition, a pre-liminary examination is conducted into whetherimprovement rates in key areas for JIT ®rms arehigher for those utilising an organic model ofmanagement.

1. Technology and management control systemdesign: theory development

1.1. Preliminary concepts and results

Management control refers to the formalisedrelationships, procedures and systems that areused to in¯uence the probability that people willbehave in ways that will lead to the attainment oforganisational objectives (Flamholtz, 1983, pp.154; Merchant, 1985). Organisational structure isa critical component of a ®rm's management con-trol system because patterns of activity, commu-

nication, behaviour and commitment are createdby it (Bell & Burnham, 1989). The purpose of thissection is to compare and contrast the manage-ment models of control for JIT and traditionalmass production ®rms using Burns and Stalker's(1961) popular bipolar continuum of standardisa-tion: mechanistic and organic. However, beforeproceeding with the comparison, a brief review ofthe key constructs is useful.

1.1.1. Mechanistic and organic models of control

Burns and Stalker (1961) posited two pureforms of management control±mechanistic andorganic±that were considered to be located atopposite ends of a continuum. While most ®rmswill be found at di�erent points between thesepolarities, they have nonetheless proved to beextremely valuable constructs in examining man-agement control systems. Mechanistic systems arecharacterised by centralisation of control andauthority and a high degree of task specialisationand standardisation. Communication is largelyformalised, utilising vertical communication andreporting paths. Decisions, rewards and punish-ments ¯ow down, while information, often in theform of exceptions, ¯ows back up. The role of linemanagement is simply to operate the establishedfacilities, systems, and personnel according tosenior management's rules, regulations and pre-determined targets. In short, mechanistic organi-sations are highly bureaucratic and rigid.

Conversely, organic systems are much ``looser''structures. They generally exhibit greater decen-tralisation of control and authority, a participativestyle of management, and more reliance on groupprocesses (team work) to achieve integration andadaptation in managing functional inter-dependencies. In particular, according to Frenchand Bell (1984, pp. 260±261), organic structurespossess two important features. First, they areadaptive and ¯exible. A continuous reassessmentof tasks and assignments occurs in order to dealwith new problems and/or opportunities facingthe organisation. Second, organic structures utilisea network pattern of authority and control,whereby the formalised, vertical communicationpatterns underlying the mechanistic organisationalstructure are largely abandoned to one that

2 S.S. Kalagnanam, R.M. Lindsay/Accounting, Organizations and Society 24 (1998) 1±30

permits and encourages widespread communica-tion throughout the ®rm. Communication ¯owsare both horizontal and diagonal, as well as ver-tical, being dependent on where the requisiteknowledge resides. Additionally, communicationis now much more in the nature of consultation,information-sharing and advice-giving.

1.1.2. Technology

Technology has been found to play an impor-tant role in the determination of organisationalstructure (Mintzberg, 1979; Fry, 1982; Steers,1988). Technology is de®ned as the knowledge,tools, techniques, and actions used to convertorganisational inputs into outputs. It includesmachinery, employee education and skills, and thework procedures used in the transformation pro-cess (Daft, 1986, p. 133).

Woodward's (1980) seminal work is of partic-ular interest to this study. She de®ned three dif-ferent types of technology: (i) unit or small batchproduction which tend to be customer orderoperations meeting the speci®c needs of customers,(ii) large batch or mass production of standardisedproducts, and (iii) continuous process productioninvolving the highly mechanised manufacture of asingle bulk product, e.g. chemicals. She found thatorganisational success was contingent on havingthe right combination of structure and technology.The best small batch and continuous processplants had more organic structures, while thesuperior mass production plants were moremechanistically structured. Woodward's ®ndingshave been supported by other researchers (Zwerman,1970; Keller et al., 1974).1

The rationale underlying the results for unit andprocess ®rms is important to understand. Thenon-standardized output of unit production ®rmsrequires that they possess ¯exible technical sys-tems. As Starbuck and Dutton (1973, p. 25)explain:

. . . the generally small lot sizes make elabo-rate setups uneconomical, and preference isgiven to equipment that sets up quickly andcheaply. . . They install simple, basic machinesthat are easily adaptable to many uses,because specialized machines are liable to bemade obsolete by changing customers' orders.However, this adaptability depends on twothings: bu�er inventories between machinesto accommodate varying machine speeds, andhighly skilled machine operators, who canunderstand the requirements of di�erent pro-ducts and understand basic machines to dif-ferent purposes.

In unit production ®rms the work is craft-like innature, with operators being responsible fordetermining the method and sequence of opera-tions, as well as the quality of the ®nished product.Coordination or integration within the productionfunction is handled by mutual adjustment amongoperators themselves or through direct supervisionby ®rst-line managers (Mintzberg, 1979). Closeinteraction among marketing, development andproduction is required until the order is complete,and this is achieved through interpersonal contacton a day-to-day basis (Woodward, 1980). In sum-mary, the product ¯exibility of unit productionsystems requires an organic structure.

Process production, on the other hand, is char-acterised by the automation of the process (Min-tzberg, 1979), resulting in the various operationsbeing tightly coupled to one another. The mainobjective is to maximise throughput in order todefray the huge capital costs involved (Woodward,1980 p. 162). Although the production process ishighly standardised, the work of the core operatorsis not. The operating personnel are highly skilledindirect workers whose main task is maintainingthe machines in order to ensure the smooth run-ning of the equipment (Mintzberg, 1979). Their

1 Some readers might question the appropriateness of exam-

ining traditional production systems and concepts which were

developed so many years ago when more and more ®rms are

adopting various aspects of new manufacturing methods and

management practices. We readily accept that such changes are

taking place; however, while the manufacturing practices (tech-

nology) may not be invariant over time, the insights from using

a ``good'' theory should be robust over time (see Lindsay, 1993,

pp. 235±236). More speci®cally, our aim is to examine the key

theoretical elements underlying the structural prescriptions

advocated for the di�erent technologies and apply them in the

context of a new technology, JIT, to develop a theoretical

position which is based, at least in part, on the use of a well-

established theoretical framework.

S.S. Kalagnanam, R.M. Lindsay/Accounting, Organizations and Society 24 (1998) 1±30 3

goal is to keep the plant from shutting down. AsHunt (1970) explains, process ®rms are problemsolvers who are concerned with exceptions. Deci-sions must be taken quickly to solve any problembecause the tight coupling of the processes meansa disruption in one operation impacts the entiresystem. The need to deal quickly with exceptionsrequires close interaction among people who pos-sess the required knowledge, which is best facili-tated through a network structure of coordinationand communication.

1.2. Management control in traditional massproduction ®rms

Traditional mass production ®rms are ``perfor-mance'' organisations (Hunt, 1970), which followa strategy based on cost and price leadership(Nemetz & Fry, 1988). They are characterised bytheir use of semi-automatic or special purposemachines which have relatively complex andlengthy set-up time requirements (Mintzberg,1979). The high level of set-up costs has led to anemphasis on pursuing economies of scale andprocess e�ciency through high product and pro-cess standardisation and the use of long produc-tion runs (Hunt, 1970; Mintzberg, 1979).

To be successful (i.e. e�cient), mass production®rms have to run smoothly without interruptionand this is only possible if variability (uncertainty)is either reduced or its impact dampened. Conse-quently, the mass production organisation hasdeveloped an obsession with control throughstandardising not only the technical process, butalso human behaviour in order to increase pre-dictability (Mintzberg, 1979; Nemetz & Fry,1988). Stabilizing forces such as rules and regula-tions, highly specialised jobs or tasks aimed atroutinising the work, and centralisation of deci-sion making have all been introduced for the pur-pose of reducing variability. In addition, carryinglarge amounts of inventory has become a centraltactic for isolating the plant and its departmentsfrom such disruptions as increases in orders, latesupplier shipments, machine breakdowns, ordefective parts, thereby reducing the need fortimely coordination both within the manufactur-ing function as well as across functions.

In summary, a mechanistic management systemor what Mintzberg (1979) calls a ``machinebureaucracy'' has evolved to control and coordi-nate traditional mass production ®rms. As long asthe organisation can shield its operating core fromhigh levels of variability, the mechanistic structureis conducive to achieving high levels of perfor-mance following a low cost strategy (Burns &Stalker, 1961; Hunt, 1970; Nemetz & Fry, 1988).However, the price paid for this bureaucracy is thedi�culty in responding or adapting to changes ina timely manner±underscoring why ®rms focus oncontrolling (stabilising) their environments (Min-tzberg, 1979, p. 326).

1.3. Management control in a JIT environment

JIT is not an inventory system, nor is it justanother production technique; it is best describedas a philosophy of management which is dedicatedtowards the continuous elimination of waste ofevery form in every area of the organisation(Schonberger, 1982; Hall, 1983; Hay, 1988;McNair et al., 1989). Moreover, consistent withSenge's (1990) logic of system dynamics, JIT isbased on a holistic conception of the organisationwhere overall ®rm optimisation requires managinginterdependent component parts in considerationof total system performance rather than simplyattempting to maximise individual components vialocal e�ciency measures (Schonberger, 1982).Finally, customer concerns are given the highestpriority under JIT (Heiko, 1991).2

Following Hay's (1988) framework, JIT hasthree main pillars or components. The ®rst is pro-duction ¯ow and this refers to the way the manu-facturing process proceeds from one operation tothe next. The most important element of produc-tion ¯ow is the demand-pull system which is basedon the concept of matching production with need.Second, JIT is based on ``doing things right thevery ®rst time'' in every activity of the organisation.In particular, achieving quality at the source,rather than through inspection, is the goal. Finally,

2 The conceptualization of JIT adopted in this paper shares

many common elements with the more general perspective of

Total Quality Management (TQM).


employee involvement through respect for work-ers is the third pillar of JIT. The active involve-ment of those individuals best acquainted with theactual work processes and day-to-day problems isa fundamental requirement in any continuousimprovement initiative (Hall, 1983; Hay, 1988).

The technology underlying JIT is radically dif-ferent from traditional mass production in thatJIT production attempts to synthesise the bene®tsof unit, mass, and continuous process technologiesthrough the focus on achieving ¯exibility, e�-ciency, and throughput. Historically, companiespursuing these goals have adopted organic struc-tures in order to facilitate the increased integrationand decentralisation of decision making that isrequired (see Nemetz & Fry, 1988 ). JIT ®rms willbe no di�erent for at least three reasons.

1.3.1. Degree of coupling

Under JIT, the degree of coupling among adja-cent processes within the manufacturing systemincreases (Duimering & Safayeni, 1991; Klein,1991). As Duimering and Safayeni (1991) explain,a key characteristic of JIT, unlike the traditionalmass production ®rm, is the idea of carrying``minimal'' inventory bu�ers due to their negativeconsequences (see Schonberger, 1982; Hall, 1983).Inventory decouples manufacturing processes fromone another by preventing the variability of oneprocess from having an impact on another,thereby allowing each process to function in relativeisolation from each other. Signi®cant inventoryreductions (below ``critical levels'') increase thedegree of coupling between linked processesbecause of the decreased time it takes for thevariability in one process to impact upon another.Below a certain level, such things as machinebreakdowns, poor quality input parts, and absentworkers will begin to rapidly disrupt the manu-facturing system. The critical level will be di�erentfrom ®rm to ®rm depending upon a company'sprogress in reducing the sources of variability (e.g.stemming from suppliers, manufacturing pro-cesses, workers and other organisational functions)and their success in adopting variability handlingmechanisms in place of inventory.

Duimering and Safayeni (1991) state that agood deal of the variability experienced within the

production system is created by the actions ofother functions within the organisation. Forexample, variability may be increased by a pur-chasing decision to change the grade of steel inorder to reduce cost or by marketing decisionswhich lead to frequent changes in productionschedule. Thus carrying reduced inventory levelscompounds the e�ects of variability (exceptions)introduced by other functional units.

In conclusion, JIT results in tightly coupledorganisational processes (and functions). Organi-sationally, increased coupling in the systemrequires a highly adaptive process of coordinationand control, i.e. mutual adjustment, in order todeal with any changes and/or problems in a timelyand e�ective manner (Thompson, 1967; Galbraith,1973; McNair, 1990; Duimering & Safayeni, 1991;Klein, 1991). This is in contrast to the traditionalsystem where coordination concernsÐboth withinmanufacturing and across functional units±aremanaged through standardised procedures andprocesses. By eliminating the bu�ers that pre-viously allowed departments to function in isola-tion of one another, JIT forces the interrelatedareas of the organisation to adopt e�ective com-munication and coordination procedures to copewith variability (unexpected occurrences).

1.3.2. Flexibility

The implications of the greater interdependenceof operations and functions is further com-pounded by the increased ¯exibility of JIT sys-tems. Flexibility re¯ects the variety and/or mix ofproducts a production system can economicallyproduce in response to market changes and custo-mer needs (Abernethy & Lillis, 1995). JIT systemsare highly ¯exible with respect to production mix(Schonberger, 1982; Hay, 1988).3 Mix ¯exibilityhas been achieved largely by obtaining drasticreductions in set-up times, thereby allowing batch

3 Unit production and JIT ®rms do not always focus on the

same notion of ¯exibility. In the main, the former emphasizes

product ¯exibility, while JIT ®rms, at least initially, concentrate

on production mix ¯exibility. However, some JIT ®rms increase

the automation of their facilities (e.g. computer integrated

manufacturing), and this allows them to increase the variety of

products o�ered in order to respond to speci®c customer's

requirements.


sizes to be decreased signi®cantly while stillremaining cost competitive (Sepehri, 1986). Other¯exibility enhancing characteristics of JIT systemsare plant simpli®cation (towards cellular manu-facturing), improvements in line balancing, andthe demand-pull orientation (Schonberger, 1982).Strategically, JIT ®rms are therefore able to bemore responsive to customer demands while stillbeing able to compete on the basis of cost (e�-ciency), superior product quality, and on-timedelivery (Schonberger, 1982; Sepehri, 1986).

Running a ¯exible production system requiresincreased integration and coordination within theproduction line, across functions and with the®rm's suppliers and customers, and this cannot beaccomplished using standardised behaviours,formalised controls and hierarchical arrange-ments that are characteristic of traditional massproduction ®rms. Meeting customers' changingrequirements dependably (i.e. on-time) usingreduced inventory levels requires that informationand communications ¯ow ¯exibly and informally,wherever and whenever they must, in order toprovide for timely coordination and control ofoperations. For example, a change in a customer'sorder (model and/or quantity) requires fast ande�ective communication between marketing, pro-duction, and purchasing to coordinate and controlthe change. As Abernethy and Lillis (1995 p. 243)explain: ``successful implementation of manufactur-ing ¯exibility requires cross-functional responsive-ness to speci®c customer-initiated demands.E�ective performance is the result of the com-bined e�orts of those functional units required tosatisfy customer demands.'' In this regard, theauthors report that ``¯atter'' organisational struc-tures are critical to achieving e�ective integration.This allows decisions to be made faster and basedon information that is more accurate (one opera-tional specialist talking to another).4 Additionally,markets are changing so quickly that operational

specialists must make many decisions themselves.There is no longer time for management to absorband assimilate information obtained from peoplelocated in various operational units if the organi-sation is to remain ¯exible. Indeed, JIT systemsaccentuate the speed at which decisions must bemade. Decision making must therefore becomeincreasingly decentralised, occurring at the pointin the organisation where the knowledge andexpertise lies. To take another example, if a com-pany produces two or more models, attainingminimal inventories and on-time delivery requiresproducing each product at the frequency requiredthrough mixed model scheduling. Line balancingbecomes an issue when di�erent models possessvaried processing times in individual operations.To cope with this problem, foremen in Japanese®rms have been given considerable line-balancingresponsibility and, as a consequence, the authorityto re-assign workers to achieve more balanced pro-duction (Schonberger, 1982, p. 139).

In summary, like ¯exible manufacturing systemsin general, running JIT requires an organic struc-ture to e�ectively coordinate and control activities(see Nemetz & Fry, 1988; Parthasarthy & Sethi,1992, 1993; Abernethy & Lillis, 1995).

1.3.3. Continuous improvement

The third reason supporting the adoption of anorganic form of control is JIT's continuousimprovement philosophy which emphasises inno-vation through the continual elimination of wastein every form. Inventory bu�er stocks are con-sidered to be the epitome of waste, serving only tomask the problems underlying the causes ofvariability. Improvement e�orts are centredaround the notion of continually reducing pro-duction lot sizes through simplifying the produc-tion process, eliminating bottlenecks (or moreproperly speaking, creating newer, less bindingones), making improvements in set-up times,quality, and increasing preventative maintenanceand supplier reliability (both in terms of qualityand timeliness). A new cycle of improvementbegins as further deliberate reductions in bu�erstocks take place to provide workers with themotivation to ferret out new sources of variability.In this way, the JIT philosophy, and its aversion

4 With respect to the issue of ¯atness of organisational

structure, Peter Drucker's (1988 p. 46) comments are insightful.

He writes ``that whole layers of [middle] management neither

make decisions or lead. Instead, their main, if not their only,

function is to serve as `relays'±human boosters for the faint,

unfocused signals that pass for communications in the tradi-

tional . . . organisation.''


to carrying inventories, provides an internalmechanism which drives cycle upon cycle ofimprovement (Schonberger, 1982).

The organic form of management control fos-ters innovation (Burns & Stalker, 1961; Linsu,1980; Steers, 1988)Ðand this is a claim which isnot a matter of dispute (Daft, 1982, p. 144). Thefollowing passage by Daft (1982, p. 135) explainssome of the reasons why:

The success of the organic organization isattributed to its ability to introduce new ideasinto the organization. . . Lateral communica-tions are a mechanism for achieving cross-fertilization of ideas from di�erent perspec-tives. Less emphasis on control and authoritylevels enables lower participants to takeinitiatives to propose changes. . . The lessonfrom the organic model is clear: decentraliza-tion, participation, . . . minimal control, lat-eral communication and other organiccharacteristics are conducive to innovation.

Innovation requires granting more responsi-bility and control to the workers or doers: thosewho know the work best and confront (or areforced to confront through deliberate withdrawalsof bu�ers) the problems and/or opportunities formaking improvements. As Schonberger (1986, p.18) indicates, employee involvement does notconsist of mere participation and the occasionalconsultation of workers; instead, it requires ``mas-sive involvement in the minute to minute problemsthat operators face on the shop ¯oor.''

In conclusion, the organic modelÐwhich focu-ses on breaking down both vertical and horizontalbarriersÐappears much more conducive towardsachieving innovation.5

1.4. Comparison of production systems

The above discussion suggests that JIT ®rmsshare commonalities with unit, mass and processproduction ®rms in varying and di�ering respects.First, both unit production and JIT ®rms focus on¯exibility. Unit production ®rms focus on product¯exibility, and this is achieved through the use ofmachines that set up quickly and are adaptable to

many uses. Conversely, JIT ®rms emphasize pro-duction mix ¯exibility, and this is achieved bydrastically reducing set-up times, using simplerplant con®gurations, and balancing productionlines through mixed-model production. Second,both JIT ®rms and traditional mass production®rms are interested in increasing the e�ciency oftheir processes through a focus on standardisa-tion. However, unlike the traditional counterpart,JIT production systems place a priority on being¯exible enough to absorb internal and externalvariation without relying on the use of high inven-tory levels (Schonberger, 1982, p.134). Third,similar to process production systems, operationsare tightly coupled in JIT ®rms, due to reducedbu�er inventory levels, the sequential (cellular)layout of the production process (as opposed tothe use of complex spaghetti paths commonlyfound in traditional mass production ®rms),and the use of the demand-pull system. Finally,and unlike the other systems, continuousimprovement is an integral part of the JIT phil-osophy. Table 1 presents a comparison of the fourproduction systems.

2. Case studies

There is a paucity of material in the academicliterature which describes the control structuresused in JIT ®rms along with an explanation ofwhy they use them and how they work (i.e. interms of organisational processes). As a con-sequence, three case studies were conducted forthe purposes of determining the soundness of the

5 Following Klein (1991), it should be noted that the JIT

view on employee involvement is quite di�erent from the view-

point put forward in the socio-technical literature espousing

individual and group autonomy over the pace of work and

work methods. Under JIT, employee contributions relate to

participating in making changes to task design rather than to

task execution. There is no ¯exibility with respect to task

execution because JIT production systems can only operate

with reduced inventory bu�ers by reducing unexpected con-

sequences (variability), and this is achieved by utilising a high

degree of standardisation throughout the manufacturing pro-

cess (a basic feature of quality organisations in general). Other

avenues of involvement include control over production ¯ow

and management of quality and resources (Klein, 1989).


above theoretical framework, to add to the theoryunderlying the structuring and control of JIT ®rms,and to provide a ®rmer basis for constructing thestudy's hypothesis. Each case study was comprisedof a four day site visit by the researchers.6

2.1. Plant 1

2.1.1. BackgroundPlant 1 is a branch plant of a Canadian sub-

sidiary of a large American corporation. The plantmanufactures over 1500 di�erent parts, consistingof bearings, sleeves, axles, stub shafts, journalsand ball yokes for Ford and General Motors.While the main activities are machining andassembly operations, the plant is also involved inforging. Its production is sold in both NorthAmerica and Europe. The plant employs 225 peo-ple and has sales of $34 million. JIT has beenpursued for approximately six years.


Continuous improvement has become a buzzword in the automotive industry and this is one areathat both corporate management and the plant'scustomers stress. In this plant, continuousimprovement is emphasized every day because, asthe quality manager put it, `ìf you stay where youare, you are going backwards.'' Consequently, each

year's improvement targets become increasinglyrigorous. In order to overcome the ``plateaue�ect'' that can occur upon reaching intermediaryobjectives, the plant's long run objectives arecommunicated in the form of ideals: zero inven-tories, zero scrap, zero setup time, etc.

Many of the ideas for continuous improvementare presented and discussed at four di�erent kindsof meetings that are held in the plant: daily, safety,project team, and, in particular, JIT improvementmeetings. Several JIT groups have been formedwithin the plant, one for each production line.Typically, each group consists of the foreman, thesetup man and process engineer in charge of theline, and two or three operators; however, the cir-cumstances of each project determine the speci®cgroup members and the timing of their participa-tion. The group selects the chairperson (whocannot be a foreman) to direct the meeting. Theynormally meet once a month, although they canmeet more often if necessary. Through the use ofthese meetings, management is consciouslyattempting to get factory workers to accept own-ership of their problems and the solutions pro-posed to counter them.

The meetings are typically focused around sol-ving one problem at a time. These problems areeither suggested by group members or are re¯ectedin the organization's measurements, e.g. scrap.Once a speci®c problem is identi®ed to be pursued,the group attempts to trace the problem to itssource. Often this requires data from a variety ofsourcesÐdata which must be speci®cally tracked.

Table 1

Comparison of producton systems

6 Portions of the ®rst two case studies presented are excerpts

from a much wider discussion found in Lindsay and

Kalagnanam (1993).


For example, in tracing the major cause of scrapon the journal line, the group members proceededin the following manner (as reported in the min-utes of the meeting).

We decided to start at the trunnion grindersas our ®rst objective at reducing scrap. As weinvestigated the problem we found that over aperiod of 1 year, 4000 pieces were scrappedout at the grinders. We have 10 grinders, sowhen you break it down daily it is less than 2pieces a day per grinder. Then we looked atthe Kingsbury drill scrap. It was 2600 piecesfor the year and we only have 2 machines. . .Less than half our production goes throughthis operation, so we are double the percentagescrap o� the Kingsbury drill. We will nowwork on reducing scrap at the Kingsbury.

Following the identi®cation of the troublesomeoperation, attention would then be placed ondetermining the speci®c causes of the problem,perhaps through running `èxperiments.'' Oncedetermined, the group decides on alternativecourses of action. Determining the solution mightrequire further tests and trial runs.

The plant operates in an environment wheremistakes in implementing improvement ideas arenot only tolerated, they are viewed as a positiveoutcome providing for organizational learning. Tocite the plant manager:

There's nothing wrong in making mistakes.Because if you want to change, you have to try,and in trying, one will make mistakes. Yes, wewill do things in small steps so that the mistakeswe make are not severe and a�ect our custo-mers badly. . . Unless you try and make mis-takes you don't learn. If a thing doesn't work,you learn more from that decision because youcan try to get at the root cause of the mistake.

The plant manager argues that this attitudemust prevail (to the extent possible) even whenyou know a person's idea is poor. Such mistakesmust be tolerated in order to encourage people'sactive involvement and participation at con-tinuous improvement.

Even if we know that it's a bum idea we'vegot to let them do it. They will realize themistake themselves. The next time he willevaluate his idea more carefully. If you nip[their idea] in the bud, they'll never come backwith another idea. If you do not encouragethem, then it is [hypocritical] if you say thatpeople are your most important asset. Wedon't want his muscle, we want his brains too.

2.1.3. Management style

A conscious e�ort has been made by the currentplant manager to push decision making, wherepossible, down to the lowest level possible in theorganisation and to encourage participation. Asone foreman explains:

[Y]ears ago it was based on position [hier-archy]. Now, although the person with posi-tion has the ®nal say, it is done witheverybody's input. If there is any change inthe situation, it then goes back again to thepeople that will be a�ected to get their input,and goes back upwards. . .

The plant manager focuses on strategic planningand liaises with corporate and parent companypersonnel. Most day to day operating decisionshave been delegated to the management team andforemen; however, the plant manager does exertauthority with respect to large capital expendi-tures. He believes in giving considerable freedomto his ®ve member management team comprisedof the production manager, plant accountant,production control manager, quality manager,and facilities manager. As he sees it, his role isto provide the plant with its direction andvision, and guidance towards it. Once thedirection is established, it is the managementteam's responsibility to implement it and derivespeci®c solutions.

Each of the ®ve managers are fully respon-sible for what's underneath them. I don'tstick my nose into what they do. They have tounderstand where they are going; once theyunderstand, I have to leave them to do theirjobs the way they like.


The plant manager has made a conscious e�orttowards instilling a team spirit within the organi-sation±centred around serving the customer bet-ter±by shedding the functional boundaries whichguided people's thinking in the past. The focus ison getting involved with each others' jobs andaccepting new responsibilities. This is evident atthe management level, where everyone possesses agood understanding of the details of the entireoperation and the developments which are occur-ring in each area. Comments by the facilitiesmanager are illustrative of this orientation.

I think we have a common goal to make thisplant better and we all strive to just con-tinually improve. Our attitude is that ifsomething needs to be done, we go and do itwithout thinking [whether] it's so and so'sresponsibility or mine. We talk about itthough. We don't have the luxury of sayingthat's not my job. . . We are all here for onereason, that is to make sure we satisfy thecustomer. We don't want a customer to be[mad] at us.

Notwithstanding this boundary overlap,department heads do not direct the subordinatesof their management colleagues.

The plant manager explained that his organisa-tion is moving from the traditional organisationalstructure, with its hierarchical approach to controland decision making, to a transitional model,where functional boundaries overlap, peoplerelinquish their ownership of a particular job in thecourse of becoming more ¯exible, and decisionmaking and control is increasingly pushed tolower levels. In the end, his vision is to attain a``participative model'' of organisational structureand control.

The participative model (in its extreme version)represents a highly integrative organisation whichis characterised by the absence of functionalboundaries. Under this model, there is no need forsymbols of power (e.g. plant manager and depart-mental heads) because decision making authorityand control resides at the shop ¯oor. Members ofthe organisation are highly ¯exible and orientedtowards pursuing a common goal±serving the

customer. The thinking underlying this model isthat the removal of functional boundaries willfacilitate a holistic approach to decision making;speci®cally, making people realise that the out-comes of decisions are not isolated, but impactupon the entire organisation. In other words, thereis recognition that maximising local criteria maynot necessarily lead to a global optimum.

2.1.4. Decentralisation

The implementation of the transitional modelhas resulted in a number of structural changes.According to the plant manager, the speed ofdecision making increases under JIT. His responsehas been to reduce the number of administrationpersonnel and to combine responsibilities. Theresult is improved coordination; moreover, every-body's job is closer together and they begin toappreciate how their decisions a�ect people loca-ted in other functional units. In the long run, theplant manager wants his direct subordinates(members of the management team) to eliminatetheir jobs by getting the system to run smoothly andto decentralise tasks on the shop ¯oor. His thinkingis based on the viewpoint that the most importantperson in the organisation is the individual ``dril-ling the holes.'' Everyone else, according to him, isburden. In this connection, it is the corporation'slong run target to operate each plant with a max-imum of two organisational levels and fewer thantwo hundred people.

Although decision making is considerablywidespread within the plant, the plant managerbelieves that the involvement of foremen andoperators is still lacking. To this end, managementis striving to empower production ¯oor peoplewith more responsibility. The long run plan is toimplement the ``participative model'' by treatingeach production line as a separate business withthe foreman becoming its ``manager.'' Foremenwill be eventually responsible for their own rawmaterial, work in process, and ®nished goodsinventory levels. This idea is currently being pilo-ted in one production line. The foreman isresponsible for his own buying and meeting thecost budget as well as achieving assigned produc-tion targets. Although the budgeting seems to bemore ``top-down'' than negotiated, which has led


to some perceptions of unfairness, management'sintent is to get foremen to appreciate the ®nancialaspects of their decisions.

Under the new arrangement, both the foremanand the process engineer share the responsibilityfor ordering tools and other supplies. Previouslythis was done by a central buyer for the entireplant. Several bene®ts are expected to be obtainedunder the new system. Only the foreman andprocess engineer's signatures are now required toauthorise a purchase, whereas ®ve signatures werepreviously required (the latter three being man-agement personnel). As a consequence, theauthorisation process led to unnecessary and oftenlengthy delays (e.g. a manager might be out oftown for a week)±hindering the plant's JIT objec-tive of purchasing in small quantities as and whenrequired. In addition, shop ¯oor personnel have abetter knowledge of the quality of tools and partsthey purchase as compared to the buyer who is farremoved from the production line. Analysis ofcost bene®t trade-o�s is therefore facilitated.Moreover, the foreman can deal directly with thesupplier on quality problems, rather than havingto go through a third party (i.e. the buyer).Finally, one department is responsible for bothon-time delivery and purchasing (and eventuallyinventory levels), thus facilitating both JIT pur-chasing and maintaining small inventory levels.Visual stock control becomes possible, therebyreducing problematic and costly administrativecontrol systems and procedures.7

2.1.5. Decision making

The early morning production meetings providethe key coordinating mechanism for running theplant. The participants in the meeting typicallyconsist of the foremen of the various productionlines, the quality supervisor, production manager,production control manager, facilities manager,metallurgist and the plant accountant (the latterdoes not always attend). Among other things,these meetings discuss such issues as progressappraisal, customer updates, changes in schedul-ing, troubleshooting, expediting, purchasing ofraw materials, critical repairs, the use of premiumfreight, and updating personnel on externaldevelopments.

The production manager plays the critical roleof directing the meeting. Much of the discussioncentres around questioning each foreman of anyimpending problems and the corrective action tobe implemented or, if the department did not meetthe previous day's schedule, the problems experi-enced, the remedial steps being undertaken, andwhether the missed production can be recoupedwithin the time remaining in order to meet thecustomer's delivery date. Based on such informa-tion sharing and updating, the production man-ager, in consultation with the production controlmanager (scheduler), makes production schedulechanges and may authorise overtime in order toreduce the backlog. He also provides advice to indi-viduals. In addition, other items such as the prioritieslist for the month, plant and machine maintenance,scheduling problems and options, customer feed-back and concerns are discussed, with the relevantperson updating or informing the rest of the group.

Although the production manager makes manyof the decisions at these meetings, and to someextent plays an authoritarian role, they are basedon input received from most people in attendance.The meetings are short (typically 45 min) andprovide day to day feedback and control of theplant's operations. The multi-functional repre-sentation allows everyone to be informed aboutdevelopments occurring throughout the plant.Prior to the adoption of JIT, these meetings wereheld once or twice a week rather than daily.

Nonoperating decisions (e.g. strategic, longterm, capital expenditures) are typically made on aconsensual basis in either management meetingsor project team meetings consisting of memberspossessing the required knowledge or information.Typically, at least some members of the manage-ment team are present in these meetings. On thebasis of being involved in both operating andnonoperating meetings, management is able toexert strong direction and receive constant feed-back on the activities of the organisation.

7 As this discussion suggests, the bene®ts of decentralising

purchasing decisions appear to be great; however, one needs to

be cognisant of the potential internal control problems that

may arise from such practices and incorporate appropriate

measures [see the teaching note to the Stanadyne Diesel Sys-

tems [A] case in Robinson (1989)].


2.1.6. Operator involvementThe major form of operator involvement

(including the setup person assigned to a parti-cular line) occurs through attending JIT improve-ment and safety committee meetings; however, noteveryone belongs to these groups. Another sourceof involvement stems from ad hoc meetings ofshop ¯oor personnel. They discuss scrap reductionand quality improvements. In addition, the foremenand process engineers seek operator input (eitherin improvement meetings or on the shop ¯oor)with respect to changing production line lay outsand their concerns are addressed (typically ergo-nomic and ease of operation). However, once theprocess and layout are decided, the operatorsfollow a set procedure. Thus, they have inputinto task design but relatively little in taskexecution. The pace of work is largely under thecontrol of the operator (inventory is not yetminimal).

The operators have considerable control overthe quality of the parts they produce. Each shiftthey check the reliability of their measurementgauges. They also prepare their own statisticalprocess control (SPC) charts and adjust themachine's speci®cations if the results lie outsidethe SPC control limits. They stop their machines ifthe SPC chart indicates that the part speci®cationslie beyond the tolerance limits set by the customer.In this situation, the entire line does not usuallystop because the plant does not operate hand tomouth as considerable WIP still exists. Operatorsdo their own cleaning of machines and many per-form routine maintenance on their machines (e.g.changing belts, sharpening drills). They typicallyreplace their own tools based on a pre-determinedstandard established for each tool, as well as per-forming tool adjustments. The extensive experi-ence of the operators has lead them to becomingvery knowledgeable about their equipment andmany are able to run a number of machines(although they do not practice work rotationmethods). On one line, they assist the foreman withscheduling decisions.

Plantmanagement's long term objective is to haveoperators performing the bulk of their own setups.Currently, operators only perform minor setups,with the more complex ones being undertaken by

the production line's setup mechanic (typicallywith the assistance of the operator).

2.2. Plant 2

2.2.1. BackgroundPlant 2 is a Canadian branch plant of a division

of a large U.S. company. It operates a 100,000square foot plant which stamps automotive partsprimarily for the Big Three auto makers, withmost shipments going to the United States. Someassembly work is also performed. Examples ofsome of the products include bumpers, car doortrim mouldings, and side door beams. The plantemploys 142 people and has sales of $20 million.JIT has been in operation for approximately twoyears in an attempt to achieve World Class Man-ufacturing (WCM) status.

The switch to JIT was mandated by the needto survive in an increasingly di�cult industry,where the customer is truly king. Survivalrequires continual improvement through innova-tion and waste reduction. The days of maintain-ing the status quo are gone. As the productionmanager put it:

The only guys that are going to win the battleare the guys that are innovating to allow themto lower their real costs or at least maintaintheir costs [in times of in¯ation]. To win thegame, you must show that you are trying tomeet the customer's whims and needs, andare improving as you go along. If that custo-mer knows that you are proactive and knowsyou are trying to do everything in your powerto meet his needs, you aren't going to losethat customer's respect, their trust, and theirbusiness.


Both the plant manager and the productionmanager associate WCM with the possession of anattitude±that perspective of simply wanting to bethe best. This attitude is manifested in continuallyquestioning current practices, and never being totallysatis®ed with past accomplishments. Although thisattitude appears to be straightforward, organiza-tional tendencies tend to counter it particularly


when things are going well. The productionmanager elaborated with the following comments.

The revelation here two years ago was we hadone product line where we were making abundle of money. It was told to corporatesta� that as soon as this job was over [theplant] would lose money. We were abruptlytold ``That's none of your concern'' andwithin six weeks of [®nishing the job] we werelosing money. But it's the old theory of just-in-time manufacturing. When the water is upto here [pointing high], nobody sees the rocks.Then all of a sudden the water was gone. Allthe rocks started showing up. For years andyears those problems were there, but every-body had the attitude, ``Hey, we're makingmoney, we don't have a problem.'' But theattitude here now is whether we're makingmoney or not, let's go and see what else wecan do to make more money.

Without this mindset, people will simply con-tinue perpetuating the old system or becomesatis®ed with small gains, while large ones con-tinue to elude them. For example, for many yearsthe plant was scrapping approximately 28% of thebumpers they made. This scrap level became nor-mal: the plant would order extra material andproduce approximately ®ve bumpers for every fourthey needed.

Continuous improvement requires that youlook at processes from a di�erent perspective byasking ``Do we have to do it that way?'' Anexample provided by the production managerillustrates the mindset required.

I remember the ®rst WCM meeting [at ParentCompany] I attended. A fellow got up andstated how they took so many people out bycombining operations and doing this and this.They took it from 8 operators to 4 operatorsand they cut down on material handling. I gotup and said to the guy, ``Yeah, you reallyimproved the process, but could you changethe whole process?'' Everybody just lookedand said ``What do you mean?'' Who saysyou can't do that in a totally di�erent way?

We had one [job] where we changed the wholeprocess. It came in like this [from corporate]and we said, ``No, we're going to change thewhole process.'' So we did. And we cut $2[per unit] out of the manufacturing costs.Without that attitude we would've scratchedour heads for months trying to improve onthe process they had given us instead ofthinking, hey, do we have to do it that way?Who said you can't do it this way.

In instilling the continuous improvement atti-tude within the organization, it is essential for topmanagement to set the example and be open tochange. As the tra�c coordinator put it:`Àlthough we talked about WCM four years ago,the di�erence is that the plant manager and theproduction manager are doing it rather thanjust talking about it.'' In addition, top manage-ment must have the attitude that mistakes willoccur and that they represent a useful learningexperience. This point is critical because redu-cing inventory levels often results in the organi-zation walking on a tightrope. However, thesolution is not to return to high levels ofinventory; instead, it is to learn why the pro-blem occurred, and to keep trying to ®nd a wayto make it work.

2.2.3. Management style/decentralisation.

The management style and philosophy of theplant manager is an important component of thecontrol system, and is characterised by severalaspects. He believes strongly in the necessity ofteamwork and has endeavoured to harmonise hissta� into a well blended team. This team orienta-tion was visibly apparent to the researchers. Vir-tually all non-operator personnel talked in termsof ``we are a team'' and the important role thateveryone contributes towards achieving the jointobjectiveÐmeeting the customer's requirements.As the coordinator put it:

We're all a team. We all have to work toge-ther. I couldn't do it without [the productionmanager], and the other supervisors [foremen]and the people on the machines. . . I have tohave that feedback from them in order to get


back to the customer and [the customer] is themost important person in the organisation.

The plant manager has orchestrated his teamtowards ``the customer is ®rst'' attitude. Theparochial orientation that is so often witnessed intraditional pyramid structures is virtually non-existent. The activities of the plant are focusedtotally around meeting the requirements of thecustomer. The importance of the customer isapparent to everyone, from the plant managerdown to the operator level.

The plant manager operates on the basis of hir-ing good people and letting them make decisions.He is adamant about the wealth of knowledgepossessed by employees. He keeps informed andwants to know the bene®ts of proposed changes,but allows people to make decisions. A recentexample typi®es his management style. The plantrecently made its ®rst quote on a job (previouslyall quotes were performed at corporate). Thequote was worth $1.5 million per year andrequired tooling of $1 million (signi®cant amountsfor this plant). The plant manager assigned theentire responsibility for the job (pricing, toolingand designing the process) to an ad hoc teamconsisting of the production manager, qualityassurance manager, general foreman, controller,plant engineer, tool room supervisor, and pur-chaser. The quality assurance manager summarisedthe outcome of the ®nal meeting as follows:

[The plant manager] was part of the meeting,but he didn't direct us to do anything. Welooked at tool shops, ®xture shops, this typeof stu� and decided on where the quoteshould fall. He was there, he de®nitely wantedto know what the discussion was and how the[previous] meetings went. But it became aconsensus decision. I think that was prettygutsy!

2.2.4. Decision making

The plant's decisions consist of two types. Forthe day to day operations, the production manageris the nerve centre for the plant, where he directsthe operations of the organisation based on con-sultation and information sharing with people. But

to stop at this description would inaccuratelydepict the information ¯ows in the plant. As Fig. 1illustrates, to a considerable extent the plant uti-lises a network structure of control, with two waycommunication ¯owing throughout the organisa-tion. People possessing the required informationrather than hierarchical position dictate commu-nication paths. The only break in the network (asindicated by the dashed line in Fig. 1) appears tobe between the ``tra�c coordinator'' (marketing/customer liaison) and purchasing, where in mostcases coordination is achieved via the productionmanager because of the need to incorporate pro-duction information. Communications take placein the form of informal meetings as the needarises, sometimes several times a day (e.g. betweenproduction line supervisors and the productionmanager). Constant communication among theproduction manager, tra�c coordinator, generalforeman, and purchaser is, according to theproduction manager, the key requirement in run-ning operations JIT. When inventories are low,any deviation causes potential problems andmutual adjustment is required. The bulk of thecommunication is consultation and informationsharing.

For all other kinds of decisions (e.g. futureplanning of day to day operations, new orders,problems, improvement projects) the plant reliesalmost exclusively on consensus decision makingwithin teams. For improvement projects, self-directed, ad hoc teams are created consisting ofpersonnel possessing the required knowledge. Asthe project progresses, people are pulled o� theteam and others are added according to informa-tion requirements. Any person can chair themeeting. Everybody provides input into the meet-ings, and decisions are arrived at on the basis ofconsensus. For all other decisions, the managementteam consisting of the plant manager, productionmanager, quality assurance manager, general fore-man, tool room foreman, and the plant super-intendent (engineer) meet bi-weekly.

The researchers attended a management teammeeting and were left with the two followingimpressions. The ®rst is the extent of democracyexisting in the meetings. Every person contributedto the discussion, with decisions being clearly


arrived at on the basis of consensus. The purposeof the meeting was to provide an exchange ofinformation en route to making a consensus deci-sion. The meeting provided an opportunity for theplant manager to keep abreast of developmentsoccurring within the factory. His role was tobring his external information bearing on thediscussion (e.g. corporate directives, informationreceived from customers), to direct the discus-sion, and to ensure that all relevant personnel'sinput was obtained. The second impression wasthe shortness of the meeting and the rapidity ofthe decisions taken. Information was shared,pointed questions were asked, and decisionswere taken, or, if insu�cient information existedto properly make a decision, people wereassigned the task of acquiring the necessaryinformation.

2.2.5. Operator responsibility, motivation, andinvolvement.

The JIT/WCM implementation strategy ofmanagement is to concentrate on improving theprocess ®rst before focusing on worker involve-ment. Nonetheless, operators have been consulted

for determining problem areas and their ideas areactively encouraged and responded to.

A considerable amount of responsibility hasbeen placed on operators for assuring that qualityparts are produced. They do their own inspection,utilizing statistical process control. Operators arealso empowered to stop a line if necessary. Theforeman then has the ultimate responsibility toeither continue producing or to shut the line downand repair the problem. In general, the operatorshave welcomed this increased responsibility. It hasled to them becoming more quality conscious andmore involved in the quality process.

Presently, operators make few machine adjust-ments. Setups are usually performed by tool roompersonnel and repairs by maintenance sta�. Themanufacturing process consists primarily of pressesand welding machines. With respect to presses, thesetup of the die is critical and requires experiencedand trained personnel. It is not realistic to expect thatmost operators could perform this task.With respectto the welders, the equipment is quite sophisticatedand requires engineers to adjust the machines. How-ever, management is considering allowing operatorsto change welding tips in the future.

Fig. 1. Communication ¯ows in plant 2.


Each foreman schedules the production for eachday and the work procedure is set, with little or novariation possible. Workers are given a standardrate that they must meet each day (they do haveinput for overriding an unfair standard); however,within this rate, they have considerable leeway inpace during the day (work in process is not yetminimal). In the two cells at the plant, the opera-tors discuss among themselves when they will taketheir breaks in order to keep everyone's work asevenly balanced as possible.

The use of work cells provides operators withan important controlling in¯uence that the pro-duction manager attempts to capitalize upon. Incells, poor workers are either brought up to speedby peer pressure or they tend to leave the plant.Through this use of group dynamics, managementcan rely more on the operators to control them-selves rather than resorting to disciplinary actionfor poor workers.

2.3. Plant 3

2.3.1. BackgroundPlant 3 is a branch plant of a wholly owned

Canadian subsidiary of a large US based multi-national corporation. The plant manufacturesthousands of controls that regulate the tempera-ture, humidity, cleanliness, ventilation and secur-ity of homes and buildings. Although sales of thisplant's products are primarily in Canada, thereare products produced for sale in numerous for-eign markets. This plant employs 640 people andhas sales of over $425 million (including inter-divisional sales). The plant has been following theJIT philosophy for six years in the course of pur-suing its objective of achieving manufacturingexcellence through continuous improvement. Animportant change on the production shop ¯oor hasbeen the implementation of ¯ex lines that canaccommodate model changes within a productfamily thereby increasing production mix ¯exibility.For example, on one line, as many as thirty-twomodels within a product family can be produced.


Continuous improvement (CI) is becoming away of life. For example, the receiving manager

mentioned that ``the only thing to rememberabout JIT is that you don't do it once and leaveit. . . There's always a way to improve what youhave done.'' Similarly, the design managerindicated that employees are motivated to alwaysthink about how to do better rather than regretpast mistakes. As the production managerexplains:

I guess there is always an element of dis-satisfaction, there's something you can dobetter and you are never quite satis®ed. Youhave to reward or acknowledge the improve-ment that people have made but in the nextbreath you are always saying ``well what'snext, how do we get one step better?''. . . Inevery case, we have started conservatively, wehave started with a pilot and then, once wehave proven it out, we have gone and donemore. It just never ends . . . that is why . . .[acquiring] the MRP-II system may have beenthe right investment to make ®ve years ago.Would we make the same investment now?Probably not. But that is something you haveto accept in a JIT mode, in a CI mode, that adecision you made even a year ago mighthave been the right decision . . . at that timebut a year down the road you are doingsomething di�erent.

2.3.3. Decentralisation

The plant manager is attempting to push theresponsibility for decision making down into thefactory ¯oor as much as possible, where he isbeginning to get small groups of people to runtheir areas like small businesses. For example, onegroup leader on the shop ¯oor has been assignedthe responsibility for buying his own cartonsfor packaging. He faxes his daily carton require-ments directly to the supplier. All requests faxedbefore 10:00 a.m. are delivered by noon the follow-ing day. This system is called FAXCAN (similar toKANBAN) and is being extended to all packagingmaterials. The production manager's ultimateobjective is to make the group leaders on the lineresponsible for buying all their raw materials.

Production line supervisors, set-up personnel,and operators all have more responsibilities than


they did a few years ago. Production line super-visors are more involved in production scheduling.Operators have been give more responsibility withrespect to quality. They now do their own statis-tical process control and are responsible for thecalibrations of the switches, a task which was for-merly done by inspectors. According to one qual-ity control manager, ``there is no longer a need forinspectors, they are an extra burden, an extra cost.''In some areas, operators have the authority to stopthe line due to quality problems. For knowledgereasons, they do not have the authority to adjustmachine set-ups±which is the task of set-up person-nel. However, if a problem exists, the operator takesthe tool to the tool room where it gets ®xed.

The transition from a traditional assembly linelayout to ¯ex lines has resulted in a substantialamount of training which will facilitate increaseddecentralisation. A corporate newsletter outlinedthe training program as follows:

Last year, hourly employees began training toprepare for their coming responsibilities. Thetraining involves facilitation and problem-solving skills, communications and personaldevelopment. Once the factory is running full-tilt with autonomous ¯ex lines, hourlyemployees will be involved in the decisionmaking process. The team will be an autono-mous unit and will receive continuoustraining in areas as safety, MRP and Just-In-Time.

2.3.4. Cross-functional communication patterns

Given the rather large size of this plant, formalcommunications are still prevalent. For example,the plant manager was informed about the imple-mentation date for KANBAN by a memo fromthe production individual responsible for itsimplementation. Nonetheless, the level of informal(verbal) communications, particularly acrossfunctional areas, has increased considerably,allowing the plant to respond more quickly toproblems.

Several structural changes have facilitated thechange in cross-functional communication pat-terns. Production engineering, which used to be

under engineering, is now under production.Moreover, production engineers have been relo-cated and are physically closer to the manufactur-ing area so that they can work more closely withsta�. This relocation is intended to contribute toquality and delivery improvement through theircontinuous interaction with shop¯oor personnel.Similarly, purchasing and production control havemerged in order to increase coordination betweenthe two areas.

In addition, the formidable barriers which usedto exist between the marketing, production engi-neering, process engineering, design and qualitydepartments have been largely removed. Forexample, production engineering personnel nowtalk directly to the quality control group, ratherthan following the chain of command, in responseto solving a problem. Similarly, sales people com-municate directly with production people ifrequired. The concept of removing cross-func-tional barriers has also been extended to externalparties, where partnerships are being formed withcustomers and suppliers. Supplier and customerrepresentatives now visit the plant and interactdirectly with technical employees. They are muchmore involved in the new product developmentprocess as well as in modifying existing products.These partnership e�orts, resulting in what Ross(1990) calls the ``meta-®rm'', produce considerablebene®ts. For example, working in consultationwith the plant, a supplier changed the design of athree-way zone valve and made it into a two-wayvalve with no loss in functionality. This allowedthe supplier to eliminate ®ve forging operations,reduce the set-up time for this product by 80%,and increase yield by 15%. The plant was able toenjoy signi®cant cost savings from this partnershipin the form of obtaining lower prices from thesupplier. Moreover, the modi®ed zone valve wasmuch lighter than before and easier to handle.

Finally, the increased use of teams within theplant also facilitates communication within andacross functional areas. There are several ad hocteams which have been created to work on specialprojects. These teams consist of individuals whocan contribute to the issue under discussion. Forexample, a product development team may consistof the marketing person, the sales person, product


specialist, and the customer. In addition to usingad hoc teams, the plant also uses program teams,which are more or less permanent, to deal withongoing concerns and issues or large projects.Program teams are comprised of individuals fromdi�erent specialist functions and are under thedirection of a program manager. Unlike the spe-cial project teams, which typically follow a matrixapproach resulting in dual reporting responsi-bilities for members, program team membersreport to their program managers for everythingÐeven vacation requests. They are also evaluated bytheir program manager.

2.4. Discussion of case ®ndings

The case studies describe the approaches thatthe three plants have adopted in order to increasetheir ¯exibility and to deal with the increasedinterdependencies created by running JIT. The®rst two plants, and to a lesser extent the thirdplant, have eschewed the traditional hierarchicalstructure of communication, authority and controland are moving towards a network structure.8 Toan increasing extent, decision making authorityrests with the people possessing the relevantknowledge (at least above the operator level).Information and decision processes ¯ow ¯exiblyand informally wherever they must in order toprovide the necessary coordination and controlover operations (particularly in plants 1 and 2).Consensual decision making for nonroutine deci-sions was also prevalent (plants 1 and 2). Teammeetings of multi-disciplinary specialists are usedextensively for coordination and control ofoperators (particularly in plant 1), as well as tode®ne problems and to develop solutions forthem (all three plants). Moreover, there is aconscious e�ort to remove, or at least reduce,functional boundaries between people. This wasaccomplished through the extensive use of teammeetings (in all three plants), by assigning peopleincreased responsibilities that overlap traditional

boundaries (plant 1), by restructuring and mer-ging functional areas (plant 3), and by utilisingself-managing product teams (plant 1 and, to alesser extent, plant 3). In general, more peopleare being involved in decision making and theirresponsibilities are being broadened (all threeplants).

The necessity of moving towards the organicmodel of control was clearly indicated by threefactors in all three organisations. First, customerconcerns are receiving the highest priority and thisis dramatically facilitated by introducing mechan-isms for reducing cross-functional barriers.Indeed, the pursuit of customer satisfaction isreaching the point where customer concerns havebecome the ultimate arbitrator in resolving jur-isdictional disputes. Second, an implication ofcarrying reduced inventory levels is that the timeavailable to react to problems and/or changes isconsiderably compressed, thereby requiring aquick means of adaptation and integration. Third,senior management is increasingly recognisingthat people at lower levels play a vital role in pro-ducing quality products and assisting in the inno-vation process.

In conclusion, the case study descriptions arelargely consistent with the theoretical frameworkdeveloped on the basis of the literature review:organisations that pursue JIT are moving towardsan organic form of management control. Whileeach organisation is ®nding its own particular wayin becoming more `òrganic'', the common threadis the increasing use of informal and lateral (cross-functional) communications, the extensive use ofteams composed of individuals from di�erentfunctional areas, and decentralisation of decisionmaking to lower levels. In addition, self-managingteams (mini-factories within the factory) which arecharged with being highly responsive to their cus-tomers' evolving needs and requirements arebeginning to be used.

3. Hypotheses

The theoretical discussion along with the casestudies examining JIT ®rms suggest that while themechanistic form of control is appropriate for

8 The fact that plant 3 is considerably larger than the other

two plants should be considered when assessing their greater

reliance±relative to the other two plants±on facets of the tradi-

tional hierarchical structure (see Mintzberg, 1979).


traditional mass production ®rms, it is not appro-priate for those mass production ®rms pursuingthe JIT philosophy, resulting in the followinghypothesis:

H1: JIT mass production ®rms will utilise anorganic model of control to a greater extentthan traditional mass production ®rms.

Following from contingency theory, perfor-mance should be expected to increase for ®rmswhose structure more appropriately matches theirtechnology. This was a main ®nding inWoodward's(1980) study. In addition, Abernethy and Lillis(1995) observed a positive association (r � 0:38)between the use of integrative liaison devices and®rm performance for those ®rms committed to pur-suing ¯exibility. Finally, Duimering and Safayeni(1991) reported substantial negative correlationsbetween total system inventory (as measured byplant lead time) and cross-functional coordination(r � ÿ0:704) and cooperation (r � ÿ0:935). Thisleads to the following hypothesis:

H2: With respect to JIT mass production®rms, the utilisation of an organic model ofcontrol will be associated with higher rates ofimprovement in key manufacturing perfor-mance areas.

This hypothesis will be assessed by examiningseven key areas of performance improvement:quality, inventory reduction, production cycletime reduction, on-time delivery, setup timereduction, production lot-size reduction and over-all cost reduction. A review of studies examiningJIT implementation reveals that the ®rst four wereranked among the top ®ve reasons reported by®rms for implementing JIT (see Horngren et al.1994, p. 848).

The logic underlying the importance of theseareas is as follows. The basic idea of JIT is toproduce and deliver the right quantity at the righttime without relying on the use of bu�er inven-tories (Schonberger, 1982; Hay, 1988). This is notpossible without making changes within the man-ufacturing system. First, it is essential to improvequality at the source, without which there is no

way to eliminate inventories. Second, setup timesmust be reduced to facilitate production in smalllot sizes, resulting in reduced inventories andincreased production mix ¯exibility. Third, shorterproduction cycle time results in less inventorybeing required to meet customer requests. Takentogether, improvements in quality and reductionsin production lot size and cycle time allow ®rms torespond more quickly to customer demands. On-time delivery represents an external assessment ofthe ®rm's performance which is based on the cus-tomer's viewpoint. ``The ability to deliver a pro-duct precisely when that product is required tocomplete a sale is a pivotal objective of JIT''(Dodd, 1995, p. A3±7). Finally, Schonberger (1982)argues that the overall e�ect of better qualityand lower inventory translates into cost savingsthrough less material waste (scrap), fewer reworklabour hours, lower carrying costs for inventoryand plant space, less administrative tracking ofinventory, and reduced obsolescence.

4. The survey

4.1. Sample and data collection

A total of 1580 questionnaires were mailed toplant managers (addressed by name in most cases)of Canadian manufacturing plants in 1991. These®rms operated in eleven major industry groupings,located in the Atlantic, Ontario and Westernregions. Process type industries (e.g. petroleumre®ning, textiles) were not included in the popula-tion because the literature indicates that JIT isused primarily in unit and mass production ®rms.Follow-up letters were mailed four weeks after theinitial mailing; simultaneously, a number of ®rmswere contacted by telephone in order to increasethe response rate.

A total of 169 questionnaires were returned tothe researchers, providing a response rate of10.7%. Of these, eleven missed the cut-o� dateand three were unusable, resulting in a useabletotal of 155. In assessing the response rate, it isimportant to consider the results of the telephonecalls, presented in Table 2, which were undertakenin the hope of increasing it.


The telephone calls resulted in mailing another142 questionnaires because the ®rst copy was notreceived in 80% of the cases for such reasons asthe addressee was no longer with the company, themailing address had changed, or the survey hadnot been forwarded to the correct person. Thisresult indicates that many people had no chance ofparticipating simply because they never receivedthe original questionnaire. The remaining 20% ofthe 142 cases reported that the original ques-tionnaire had been misplaced. In addition, itemnumbers 5, 7, 8, 9 and 11 in Table 2 suggest thatanother signi®cant percentage (17.6%) had a verylow or no probability of responding. If these tworesults are extrapolated to the population, morethan a third of the sample had no chance ofresponding. The conclusion to be drawn from thisanalysis is that, although the response rate remainslow, it is signi®cantly higher than 10%. This con-clusion receives further support by the result that 42of the 142 additional questionnaires sent to respon-dents were returned to the researchers (these 42responses are included in the summary total of 155),providing for a 29.6% response rate on this subset.

An analysis was performed to examine thethreat of nonresponse bias (see Wallace & Mellor,1988; Wallace & Cooke, 1990). A comparison ofthis study's respondent ®rms with those in thepopulation suggests that the sample is reasonablyrepresentative of the population on at least twokey criteria±type of industry and ®rm size (asmeasured by total employees). Despite the resultsof this analysis, the researchers cannot rule out thepossibility of nonresponse bias. The respondentsmight be di�erent than non-respondents on manyother variables. Indeed, evidence was obtainedsuggesting that the proportion of JIT ®rms in thesample is not representative of those existing in thepopulation in that the respondents appeared to bebiased towards being employed in JIT ®rms (seebelow). Readers are therefore cautioned againstmaking sweeping generalisations.

4.2. Measures

4.2.1. Classi®cation of ®rmsAs previously discussed, JIT is considered to

consist of three primary components: production

¯ow, total quality control, and employee involve-ment (Hay, 1988). Although all components areimportant, the extant literature consistently indi-cates that it is the demand-pull concept that is thechief characteristic of JIT production (Hall, 1983,1987; Sepehri, 1986; Manoochehri, 1988; McNairet al., 1989; Wilkinson & Oliver, 1989). This isbecause the pull system facilitates producing todemand, which is the basic idea behind JIT(Schonberger, 1982). Firms were therefore classi-®ed as JIT or non-JIT on the basis of whether theyhad implemented a demand pull system in at leastsome of their production operations.

Of the 155 usable responses, 106 ®rms (68.4%)implemented JIT to some extent. Given the factthat JIT production was relatively new to NorthAmerica at the time the survey was conducted, thispercentage is higher than one might expect, per-haps indicating that the study was of more interestto individuals employed in JIT ®rms, therebyresulting in a biased estimate. Random telephonecalls to ®fty non-respondents indicated that thissuspicion was warranted and that a more realisticpopulation estimate of JIT ®rms was 42%. It isalso important to stress that the JIT classi®cation

Table 2

Results of telephone calls made to the initial group of

nonrespondents

N %

1. Concerned individual/assistant

requested another copy.

142 27.2

2. Addressee not available 123 23.5

3. Addressee not available at the time

of calling, nobody available or willing

to take a message

88 16.8

4. Agreed to participate 30 5.7

5. Not a manufacturing facility 29 5.5

6. Declined to participate 26 5.0

7. No one answered on repeated callings 21 4.0

8. Addressee no longer with the plant,

no replacement

16 3.1

9. Plant closed down or closing 15 2.9

10. Given to another individual to complete 11 2.1

11. Addressee on vacation, nobody in charge 11 2.1

12. Other 11 2.1

13. Total 522 100.0%


method used in this study does not imply that a®rm has adopted a fully developed JIT system.The median proportion of production capacityrunning JIT is only 47%.

A ®rm was classi®ed as being predominantlymass production if the total active productioncapacity devoted to large batch production is 40%or more and the remaining two types (e.g. job orprocess) are less than 35% each. For JIT ®rms,this classi®cation criterion was based on the situa-tion prior to adopting JIT. Although it wasrecognised at the outset that such a stringent clas-si®cation procedure would likely cause the samplesize of mass production ®rms to decrease, it wasfelt important that the classi®cation method re¯ectthe dominant technology. This is because a ®rm'sstructure and other control mechanisms will likelyreceive the greatest in¯uence by the predominanttechnology (Woodward, 1980). Given this proce-dure, a total of 12 plants were classi®ed as tra-ditional mass production ®rms and 40 ®rms wereclassi®ed as JIT mass production. JIT mass pro-duction ®rms are larger than their traditionalcounterparts based on the median (mean) numberof employees (170 (435) and 140 (107), respec-tively) and plant sales ($37.5 ($48) and $17.5 ($15)million, respectively).

Finally, various tests on the validity of the JITclassi®cation procedure indicate that it was highlysatisfactory. These tests are displayed in Table 3.Spearman rank correlation coe�cients betweenproduction classi®cation (JIT scored 1, non-JITscored 0) and measures on the extent of totalquality control and extent of employee involve-ment were 0.57 and 0.48 respectively. Moreover,these correlations increase substantially when only``high JIT'' ®rms are examined (®rms with 60% ormore production capacity devoted to JIT produc-tion). Reducing setup time is a key component inmoving towards a JIT production system (Schon-berger, 1982). A substantial di�erence exists insetup times between JIT and non-JIT ®rms, parti-cularly for ``high JIT'' ®rms. Similarly, cellularmanufacturing is utilised much more extensively inJIT ®rms. Finally, and although they are veryweak, the correlations associated with machineavailability and supplier lead times are also in thepredicted direction.

4.2.2. Dimensions of organisation structure

Within the organic conceptualisation of manage-ment control, participation, decentralisation, andnetwork structure of control were a priori dimen-sions of primary interest. Thirty statements (referredto as variables) representing these various dimen-sions were included in the questionnaire. Of these,eleven statements were developed by the researchersin order to capture the speci®c area of interest to thestudy (production and its relations within itself andwith other departments). The balance were chosenfrom the work of Burns and Stalker (1961), Keller etal. (1974) and Robbins (1983).

An initial examination of the correlation matrixindicated that ®ve variables did not have sub-stantial correlations with at least one other vari-able and were therefore eliminated from theanalysis. The remaining variables were factor ana-lysed using principal components analysis withoblique rotation. Oblique rotation, rather than themore customary varimax method, was used becausethe researchers expected the underlying dimensionsto be somewhat correlated (see Kim & Mueller,1978, p. 59; Kass & Tinsley, 1979, p. 134). Six factorsaccounting for 60.2% of the variance were extracted.Table 3 shows the questionnaire items (variables),along with the factor loadings, percentage of vari-ance explained and eigenvalues for each item. Onlyvariables with unambiguous loadings on factorswere included in the construction of the scales. Fac-tor 4 could not be meaningfully interpreted and wastherefore deleted from further analysis. All otherfactors could be easily interpreted and were callednetwork structure of control, authority and commu-nication (factor 1), task ¯exibility (factor 2), jobcodi®cation (factor 3), hierarchy of authority (factor5), and participation/decentralisation (factor 6).

The internal consistency of the items comprisingthe ®ve scales (factors) was then tested. Factors 1(eight items), 2 (two items), 3 (two items) havecoe�cient alpha reliabilities of 0.84, 0.85, and 0.82,respectively, all highly satisfactory. The reliabilityof factor 5 is 0.51. Nunnally (1967) argues thatreliability coe�cients of 0.70 or more are con-sidered adequate. On this basis factor 5 wasdeleted from further consideration. Finally, theanalysis indicated that the reliability coe�cient of


factor 6 increased from 0.64 to 0.84 if item number22 is deleted. Thus only ®ve items were included inthe ®nal version of this scale.

In conclusion, the factor analysis was successfulin developing the scales (variables) of interest:network structure of control (factor 1) and parti-cipation/decentralisation (factor 6). These scaleswill be used in examining hypothesis 1 and 2. On ana priori basis, decentralisation and participationwere considered to be separate, although closelyrelated, variables (see Schermerhorn et al., 1985,p.402). The factor analytic results suggest thatthey can be meaningfully combined into a singlescale. In addition, a correlation of 0.63 wasobtained between the network structure of controland participation/decentralisation variables. Givensuch a high correlation and in the interest ofparsimony, the decision was taken to sum the twovariables into a single variable, representing themechanistic/organic form of management control.9

The internal reliability for the combined thirteenitem variable is 0.88.

4.2.3. An examination of the validity of themechanistic/organic measure of control

Attempting to measure quantitatively the degreeto which a ®rm is mechanistic or organic using anarms-length questionnaire is a di�cult under-taking (Mintzberg, 1979, p. 225; Parthasarthy &Sethi, 1993; Abernethy & Lillis, 1995). Inacknowledging this concern, the validity of thecontrol measure was examined from a number ofperspectives. First, the case study results supportboth the conceptualisation of the mechanistic/organic form of management control as well as theoperationalisation of the construct. In particular,the ®ndings depicting the increasing use of infor-mal and lateral (cross-functional) communica-tions, the extensive use of teams composed ofindividuals from di�erent functional areas, anddecentralisation of decision making to lower levelsare all represented by the measure used (see items1 to 8, 20, 21, 23, 24, 25 in Table 4). The only sub-dimension that appears to be missing from thisstudy's operationalisation is the use of self-mana-ging production teams (i.e. mini-factories) and thecase study results indicate that they were justbeginning to be used.

Second, it is possible to examine the measure'sconstruct (theoretical) validity by examining the

Table 3

Tests for the adequacy of the JIT classi®cation

rs

Item Non-JIT JIT High JITd All JITe High JITde

1. Cellular manufacturinga 16.3 57.5 50.0 0.39 0.36

2. Total quality control (TQC)b 2.6 4.9 5.8 0.57 0.76

3. Employee involvement (EI)b 2.6 4.5 5.3 0.48 0.65

4. Level of machine availabilityb 5.7 5.9 6.0 0.05 0.16

5. Longest set-up time (mean hours) 6.4 3.9 3.0 ÿ0.11 ÿ0.236 Shortest set-up time (mean hours) 0.9 0.4 0.3 ÿ0.14 ÿ0.227. Longest supplier lead time (mean days) 91.0 96.0 81.7 ÿ0.02 ÿ0.148. Shortest supplier lead time (mean days) 13.0 9.0 12.0 ÿ0.11 ÿ0.08Total number of ®rmsc 49 106 36 106 36

a Figures in columns 2 to 4 represent the percentage of ®rms that have adopted cellular manufacturing.b Figures in columns 2 to 4 represent mean values determined from a seven point Likert scale.c Not all ®rms completed every question concerning rows 1 to 8 above.d ``High JIT'' represents those ®rms with 60% or more production capacity running JIT.e ``Spearman rank correlations between non-JIT(=0)/JIT(=1) classi®cation and the corresponding item in column 1.

9 Running the hypothesis tests separately for the two vari-

ables produced very similar results to those using the combined

variable.


Table 4

Results of principal component analysis (oblique rotation)

Statement

(variable)

F1 F2 F3 F4 F5 F6

1. People from di�erent departments and/or functions often work

in groups to solve important problems.

0.77* 0.04 ÿ0.16 0.20 0.09 0.04

2. Work teams are used by production workers to solve problems. 0.73 ÿ0.02 ÿ0.21 0.10 0.10 0.003

3. Hierarchical position, not knowledge, determines the stature of

an individual.

0.64 0.09 0.01 ÿ0.07 0.10 0.05

4. The Production Department often works very closely with

the Design and Engineering Department in designing products

for ease of manufacturing.

0.64 ÿ0.10 0.20 0.08 ÿ0.13 ÿ0.09

5. Marketing and Production Departments work together very closely

in achieving customer satisfaction.

0.57 ÿ0.12 ÿ0.03 ÿ0.08 ÿ0.32 ÿ0.14

6. Communication is both vertical and horizontal, depending

upon where needed information resides.

0.51 0.12 0.06 ÿ0.12 ÿ0.06 ÿ0.33

7. Subordinates are normally consulted before a decision is made. 0.49 ÿ.11 0.01 ÿ0.09 0.35 ÿ0.208. The management control system in your plant provides ¯exibility

for all managers to respond to problems as they occur.

0.42 ÿ0.20 ÿ0.08 ÿ0.25 ÿ0.12 ÿ0.30

9. Managerial employees are allowed to deviate from standard

procedures.

ÿ0.16 0.90 0.02 0.02 ÿ0.04 ÿ0.05

10. Non-managerial employees are allowed to deviate from standard

procedures.

0.11 0.87 ÿ0.03 0.04 0.06 0.03

11. Written job descriptions are available for all non-managerial

employees.

0.04 ÿ0.05 0.89 ÿ0.004 ÿ0.002 0.05

12. Written job descriptions are available for all managerial employees. ÿ0.18 0.02 0.87 ÿ0.04 ÿ0.02 ÿ0.0213. For many decisions, the rules and regulations are developed as

the decision making progresses.

0.17 ÿ0.07 0.06 0.71 0.02 ÿ0.10

14. Inter-departmental interactions are normally done through

integrators or liaison people.

ÿ0.06 0.19 ÿ0.13 0.60 ÿ0.18 ÿ0.08

15. The Marketing Department works closely with the Design and

Engineering Department in designing new products.

0.41 0.20 ÿ0.003 ÿ0.43 ÿ0.39 ÿ0.18

16. People have to check with their superiors before doing almost

anything on important matters.

0.21 0.08 ÿ0.10 ÿ0.05 0.69 ÿ0.07

17. People here always get instructions from their superiors in

dealing with important matters.

ÿ0.18 0.07 0.02 ÿ0.09 0.60 0.17

18. Tasks tend to remain rigidly de®ned unless altered formally by

top management.

0.04 0.17 0.21 ÿ0.02 0.50 ÿ0.32

19. Superiors often seek advice from their subordinates before decisions

are made.

0.35 ÿ0.08 0.01 ÿ0.06 0.36 ÿ0.28

20. Production workers are encouraged to suggest product design

improvements which are then followed-up.

0.09 0.03 0.12 0.25 ÿ0.18 ÿ0.74

21. Middle managers are granted latitude to make decisions. 0.03 0.26 ÿ0.08 ÿ0.20 0.13 ÿ0.6922. In this plant rules and regulations exist in writing. 0.32 0.22 0.26 0.09 0.11 0.67

23. All employees are encouraged to make suggestions when decisions are

made.

0.19 ÿ0.02 ÿ0.02 0.09 0.18 ÿ0.66

24. Production workers are encouraged to suggest process improvements

which are then followed-up.

0.25 ÿ0.11 ÿ0.35 0.25 ÿ0.01 ÿ0.63

25. First-line managers are granted latitude to make decisions. 0.22 0.24 ÿ0.01 0.03 0.25 ÿ0.57Eigen values 6.70 2.58 1.78 1.47 1.32 1.19

Percentage of variance (cumulative) 26.8 37.1 44.3 50.1 55.4 60.2

Statements 3, 11, 12, 16, 17, 18 and 22 are reverse scored.

*Variables comprising factors based on unambiguous high factor loadings.


network of relationships existing between themechanistic/organic construct and other variables(see Neale & Liebert, 1986, p. 47). It is well estab-lished in the organisational literature that, allthings being equal, the larger the organisation, themore mechanistic the control structure is expectedto be (Mintzberg, 1979, pp. 230±235), and this isthe case for the ®ndings of this study. The rankcorrelation between size and CONTROL (lowvalues=mechanistic) is ÿ0.27 for traditional smallbatch production ®rms (n � 16), ÿ0.42 for tradi-tional mass production ®rms (n � 11) and ÿ0.12for JIT ®rms (n � 105). In addition, Abernethyand Lillis' (1995) ®eld research indicates that``¯atness'' in the managerial authority structure iscritical to promoting e�ective integration amongfunctions, i.e. becoming more organic. The Spear-man correlation coe�cient between CONTROLand the number of hierarchical levels in the cur-rent study is ÿ0.20 (n � 123) and this increases toÿ0.29 (n � 123) when SIZE is controlled.10 Inother words, ®rms which possessed more organicstructures also tended to be ¯atter. Finally,Woodward's results indicate that traditional smallbatch and process ®rms should have more organicstructures than traditional mass production ®rms.A correlation was therefore computed betweenCONTROL and a dummy (0/1) technology vari-able for non-JIT ®rms (0=mass; 1=small batchor process) with size being held constant. TheSpearman semi-partial correlation obtained was0.25 (n � 36)Ða result which is consistent withWoodward's ®ndings. Taken together, theseresults indicate that the mechanistic/organic con-trol variable has a reasonable level of supportattesting to its validity. Accordingly, the broadscale statistical tests of the hypotheses can proceedwith some degree of con®dence.

4.2.4. Improvement variablesThe improvement rate variables were measured

by simply asking respondents to indicate theapproximate percentage improvement (or dimin-ishment) experienced by their plants since intro-ducing JIT. Improvement rates rather thanabsolute levels are the focus in this study becausethe researchers were interested in examining per-formance changes after the implementation of JIT.Readers should recognise that the measurement ofimprovement rate variables in this study is ratherweak in terms of reliability. Not all respondentscan be expected to have used the same oper-ationalisations. Moreover, a number of the ®rmscontacted by telephone indicated that their infor-mation system did not measure some improve-ment items. This resulted in respondentsestimating the level of improvement or, more fre-quently, leaving the item blank. Finally, bothParthasarthy and Sethi (1992) and Abernethy andLillis (1995) provide persuasive arguments insuggesting how the contingency ``®t'' hypothesisbetween performance and structure is impactedby various internal and external factors whichtypically manifest themselves in making it di�-cult to obtain results which support priorexpectations.

4.3. Results

4.3.1. Tests of hypothesis 1The ®rst hypothesis states that relative to tradi-

tional mass production ®rms, the structural (con-trol) arrangements for JIT mass production ®rmswill be more organic.11 This hypothesis will betested at two di�erent levels of JIT adoption: (1)all JIT ®rms regardless of the level of JIT adoptionand (2) ®rms with 60% or more production capa-city devoted to JIT manufacturing (``High JIT'').This two level approach to testing is based on theunderlying premise that the results should providegreater support for the hypothesisÐa higher semi-partial correlation coe�cientÐas more of a ®rm'sproduction capacity is converted to JIT. The e�ect

10 The reporting of p-values obtained from running sig-

ni®cance tests is not considered to be appropriate in this study.

This is because no random sampling mechanism was intro-

duced into the study's design (a questionnaire was sent to every

®rm in the population, from which a percentage of responses

was obtained). As a result, it is virtually impossible to de®ne

what chance means and, in turn, what constitutes a valid

probability distribution (see Lindsay, 1995). Instead, the

emphasis in this study is on the size and sign of the obtained

correlations and observing the network of relations among the

various variables.

11 Hereinafter, all references to JIT ®rms should be read as

JIT ®rms that were predominantly mass production ®rms prior

to the adoption of JIT.


of this approach is to test the hypothesis twice,resulting in a more severe test, and thereforegreater corroboration for the hypothesis (seeLindsay, 1995, pp. 40±42).

A multiple regression analysis was used in orderto statistically control for organisational size (totalemployees). Although there is some controversywith respect to the primacy of either technology orsize on structure, the literature indicates that bothhave an a�ect (Hickson et al., 1969; Mintzberg,1979, pp. 262±263;Steers, 1988, p.66). This leadsto the use of the following regression model:

Y � b0 � b1X1 � b2X2

where, Y is the form of management control: lowvalues=mechanistic; high values=organic; X1 issize (based on the surrogate of total employees);and X2 is a binary variable (mass=0/JIT=1)representing technology.

An examination of the residuals revealed thatthe normality assumption of ordinary leastsquares regression was not being met. The datawere therefore transformed into ranks before run-ning the normal regression procedure. This trans-formation results in a ``robust regression methodthat is not sensitive to . . . non-normal distribu-tions to the extent that the regular regressionmethods on the data are a�ected'' (Conover, 1980,p. 338). The testing of hypothesis 1 is based onexamining the semi-partial (rank) correlationcoe�cient associated with X2, denoted as sr2. Thesemi-partial correlation represents the correlationbetween that portion of X2 (technology) that isuncorrelated with the remaining independentvariable (in this case X1 or size) and Y (Cohen &Cohen, 1983, p. 101).

The results are consistent with the hypothesis, asillustrated in Table 5. The semi-partial correlationfor the technology variable is 0.50, with the coef-®cient (b2) being in the predicted direction (positive).The adjusted R2 for the model is 0.23Ðparticularlyhigh by behavioural research norms. Moreover, aspredicted, sr2 increases to 0.65 when only ``high JIT''®rms are examined. The adjusted R2 for this modelincreases substantially to 0.38. The regression resultsfor these ®rms are presented in Table 6.

The results also indicate that SIZE in¯uencesthe form of control structure in the correct direc-tion (i.e. the coe�cient is negative) but that thisin¯uence is clearly secondary to the technologyvariable. This result is not necessarily inconsistentwith Hickson et al.'s (1969) since the ®rms con-sidered in this sample are much smaller. The meansize of ®rms in this study was 351 employees ascompared to 3370 in theirs. As they state, ``thesmaller the organisation the more its structure willbe pervaded by such technological e�ects'' (Hick-son et al., 1969, p. 394).

In conclusion, the results are highly supportiveof the ®rst hypothesis. Obtaining such high corre-lations is rare in organisational and behaviouralresearch. Moreover, the fact that the results werestronger for the ``High JIT'' group (as predicted)and that the SIZE variable had the correct signprovides further evidence for the validity of thestudy's classi®cation of JIT ®rms and the con-struct validity of the CONTROL variable.

4.3.2. Tests of hypothesis 2

The second hypothesis states that JIT massproduction ®rms possessing organic forms of con-trol will achieve higher rates of improvement. Thishypothesis will be examined using several impor-tant areas of improvement for JIT ®rms. Given thelow reliabilities of the performance improvementvariables, the small sample sizes, as well as the``third-variable'' problems associated with examin-ing the contingency ``®t'' hypothesis betweenperformance and structure, the examination

Table 5

Regression (based on ranks) of the form of management

control on size and technology (for `àll JIT'' mass ®rms)

CONTROL=b0+b1SIZE+b2TECH

Variable Coefficient Estimate Standard

error

Semi-partial

correlation

coefficient

Constant b0 14.01 5.03 ±

SIZE b1 ÿ0.23 0.12 ÿ0.23TECH b2 0.66 0.16 0.50

n=52; adjusted R2=0.23; F2.49=8.75.

CONTROL=The form of management control: low values=

mechanistic, high values=organic.

SIZE=Number of employees.

TECH=Technology (Mass=0/JIT=1).


of hypothesis 2 is best considered preliminary, andas a consequence, only simple Spearman rankcorrelations are used to identify whether theexpected relationships emerge.

The results are presented in Table 7. All corre-lations have the expected (positive) sign; more-over, the correlations obtained for qualityimprovement, set-up time reduction, inventoryreduction, lot-size reduction and overall costreduction are of considerable magnitude, particu-larly given the concern for the low reliabilities ofthe performance improvement variables which actto attenuate the observed correlations (O'Grady,1982). The correlations for production cycle timereduction and on-time delivery improvement aremore modest but nonetheless o�er support for thehypothesis.

In summary, the results support the hypothesisthat JIT ®rms adapting their structural arrange-ments towards an organic model of control willhave higher improvement rates than ®rms whosestructures remain bureaucratic or mechanistic.

5. Further discussion and conclusion

This study sought to determine whether Wood-ward's theoryÐdeveloped for di�erent technolo-giesÐcould be extended and generalised in thecontext of a new technology, JIT. In ®nding thatJIT mass production ®rms are adapting theirstructural arrangements in moving towards an

organic structure, the results of this study andothers suggest that the theroretical framework inthis paper appears to be fairly robust; it is de®-nitely not a ¯ash in the pan.

In elaborating, while the survey's low responserate precludes the statistical generalisability ofthese ®ndings to speci®c populations, the resultsare already displaying a high degree of analyticalor theoretical generalisation over di�erent popu-lations possessing variations in the conditions ofobservation (see Yin, 1989; Scapens, 1990, 1992;Lindsay and Ehrenburg, 1993; Lindsay, 1995). Inanalytical generalisation, the researcher is strivingto generalise a particular set of results to somebroader theory (attempting to explain some pro-cess) rather than to some speci®c population.Analytical generalisation for the main ®nding isshown in several ways. First, as the theory pre-dicted, the size of the expected relationshipincreased as a greater amount of productioncapacity was converted to JIT. Second, the resultsare consistent with the Canadian study of Dui-mering and Safyeni (1991) reporting a correspon-dence between functional changes in support ofJIT and increased levels of cross-functional com-munication (r=0.845) and cooperation (r=0.758).Similarly, they complement Abernathy and Lillis'(1995) ®ndings in Australia of a substantial corre-lation (r=0.46) between ®rms pursuing ¯exibilityand the use of integrative liason devices. The latteris particularly signi®cant in that there is pre-liminary evidence to suggest that the results may

Table 6

Regression (based on ranks) of the form of management

control on size and technology (for ``high JIT''; ®rms)

CONTROLS=b0+b1SIZE+b2TECH

Variable Coefficient Estimate Standard

error

Semi-partial

correlation

coefficient

Constant b0 5.01 3.00 ±

SIZE b1 ÿ0.10 0.14 ÿ0.11TECH b2 0.69 0.16 0.65

n=28; adjusted R2=0.38; F2.25=9.10.

CONTROL=The form of management control: low values=

mechanistic, high values=organic.

SIZE=Number of employees.

TECH=Technology (mass=0/High JIT=1).

Table 7

Spearman rank correlations between performance improvement

variables and CONTROL (mechanistic/organic structure) for

JIT mass production ®rms

Performance

improvement variable

Spearman rank

correlation

coefficient

Quality 0.72 (n=25)

Set-up time reduction 0.39 (n=24)

Inventory reduction 0.45 (n=33)

Production cycle time reduction 0.22 (n=27)

On-time delivery 0.26 (n=24)

Lot size reduction 0.53 (n=21)

Overall cost reduction 0.56 (n=20)


generalise across countries (at least Western, capi-talist countries). Third, the robustness of the main®nding is all the more remarkable given that threedi�erent approaches (i.e., methods, instruments)were used to study the phenomenon in the threedi�erent papers, i.e., the results appear to beinvariant to method. Fourth, the results of thisstudy are a�ected by the size of the ®rmÐcon-sistent with a highly relicated ®nding in the orga-nisational literatureÐalthough the technologyvariable dominates, suggesting that the ®nding issomewhat invariant to the size of the plant, atleast in its application to the maufacturing plantscomprising this sample. Finally, and although thesample size was too small to check for this directlyby stratifying the data, the results may generaliseto most industries in Canada as a check on theoverall sample found it to be reasonably repre-sentative of the study's population with respect toindustry. Such industry invariance would beexpected based on our current state of knowledge(although much more work needs to be done inthis area). In summary, a considerable amount ofanalytical generalisation for the theory has alreadyemerged, although further studies exmining theseand other conditons are still necessary.

This study also examined whether the adoptionof the organic form of control in JIT ®rms leads togreater rates of improvement being observed inseven key performance areas for JIT ®rms. In allseven improvement areas, the survey data wereconsistent with this hypothesis, with the bulk ofthe correlations being of considerable magnitude.Again it needs to be repeated that the data under-lying the improvement rate variables are weak andthat the study's research design did not attempt tocontrol for the various ``third-variables'' thatcould contaminate the results; consequently, theresults should only be considered preliminary withfurther research in this area being necessary.Nonetheless, these results are encouraging andprovide an important starting point for follow-upactivities. Increasingly, ®rms are changing theirperformance metrics and obtaining reliable dataon a wide scale should now be feasible.

Finally, this study makes a methodologicalcontribution in providing a method for classifyingJIT ®rms and measuring the mechanistic/organic

construct. In both cases, the extensive checksundertaken to examine their validity indicate thatthe measures are highly satisfactory for survey-type research; moreover, the large correlationsobtained (by behavioural research standards) inthe examination of the ®rst hypothesis along withobserving the expected network of relationshipsprovide further evidence attesting to their validity.Nonetheless, improvements to the mechanistic/organic measure are possible. One possible mod-i®cation would be to include information pertain-ing to the use of self-managing teams. Anotherwould be to determine whether and how to incor-porate the number of hierarchical levels (¯atness)into the measure given Abernethy and Lillis'(1995) results indicating its importance in achiev-ing e�ective integration.12

The fact that JIT ®rms are adopting organicstructures has several implications for futuremanagement accounting research. First, theincreased emphasis on self-management (i.e.empowerment and self-guidance) creates a needfor lower level employees to have information sothat they can monitor their own performance,determine the cause of their problems, learn aboutthe e�cacy of their changes, and receive the satis-faction that comes with the recognition of makingimprovements. Empowerment alone is notenough; frequent feedback of relevant and reliableinformation is also necessary (Schonberger, 1982;Drucker, 1990). Statistical process control, paretocharting, cause and e�ect diagrams, and the use ofnon-®nancial performance measures are allimportant tools in this endeavour. The use ofmicro-pro®t centres also has possibilities (seeCooper, 1995), but much more research needsto be performed on understanding theirdynamics and behavioural consequences. For

12 As a start in this direction, a factor analysis was per-

formed on the set of variables comprising (a) factor 1, (b) fac-

tor 6, and (c) the number of hierarchical levels (¯atness) to

determine how ¯atness relates to the other two factors. The

results indicated that ¯atness loaded on a separate (third) fac-

tor. In addition, running the test of hypothesis 1 with ¯atness

rather than CONTROL as the dependent variable did not

produce any signi®cant patterns. Thus it is not clear how or

even whether ¯atness should be incorporated within a measure

of the mechanistic/organic construct. Progress on this issue will

likely require theoretical development and guidance.


example, Stata (1989) reports that the adversebehavioural consequences attributed to Analog'stransfer pricing system led to its discontinua-tion for internal purposes. Finally, the balancedscorecard measurement conceptÐwhich putsstrategy and vision, and not control, at thecentreÐappears valuable in focusing individuals'behaviour strategically (Kaplan & Norton, 1996).Thus, the information bricks supporting self-management appear to be in place; however, whatis now required are more implementation stories(cases) and studies examining their consequenceson job performance and job satisfaction.

Second, the focus on horizontal integrationposes new informational demands. Drucker (1988,p. 49) made the observation that everyone needsto ask: ``Who in this organisation depends on mefor what information? And on whom do Idepend?'' In addition, we must consider: ``Who isa�ected by my actions and decisions?'' This listwill not only include superiors and subordinates,but also colleagues in other departments withwhom coordination must occur. This requirementwill become increasingly necessary as more andmore middle management jobs are withdrawn. Inthis connection, the role of the managementaccountant is to create the kind of informationthat enables individuals or teams to see morequickly the impact of their actions on overall per-formance (Vowles, 1993). The basic point is thatwithout such information it is di�cult to deter-mine whether a proposed modi®cation is truly animprovement, or whether it simply makes thingslook better in one corner only to makes thingsworse overall. Activity-based responsibilityaccounting is one approach that has been advo-cated for managing interdependent activities (pro-cesses) and measuring group performance (seeMcNair, 1990). The use of strategically-driven,integrated performance measurements which inte-grate actions across functional boundaries isanother (see Nanni et al., 1992). Again, moreresearch is needed on how organisations are actu-ally grappling with the information requirementsposed by horizontal integration.

Finally, the concepts of horizontal integrationand teamwork run counter to the prevailing socialand organisational culture of North America

emphasising individuality, competition and indi-vidual achievement. In particular, the case studiesperformed by the researchers indicate that com-panies are struggling with how to modify incentivesystems to better promote teamwork. At a prac-tical level, cultural and union impediments hinderattempts at changing the reward structure. Atanother level, management is struggling with theissue of what performance level should be the basisfor o�ering rewards: individual vs team vs com-pany performance. Lastly, once the answer to thelatter issue is known, there still exists the furtherissue of determining the appropriate performancemeasure(s) and, if there is more than one measure,the manner in which they should be combined.Armitage and Atkinson's (1990) study is animportant starting point in this regard but muchmore research on the innovative reward systemsbeing adopted by leading companies is necessary.

Acknowledgements

We thank Mark Covaleski, Bob Miller, MarilynSagrillo, K. R. Subramanyam, Urban Wemmer-loÈ v, participants at the 1995 AAA Mid-Westregional meetings in Dearborn, MI and the 1996Strategic Management Accounting conference inEdmonton, and the Strategic ManagementAccounting conference's anonymous reviewerfor their helpful comments and suggestions.Research support from the Society of Manage-ment Accountants of Canada is gratefullyacknowledged.

References

Abernethy, M. A., & Lillis, A. M. (1995). The impact of man-

ufacturing ¯exibility on management control system design.

Accounting, Organizations and Society, 20, 241±258.

Armitage, H. M., & Atkinson, A. A. (1990). The choice of pro-

ductivity measures, Hamilton, Ontario: The Society of Man-

agement Accountants of Canada.

Bell, R. R., & Burnham, J. M. (1989, September±October) The

paradox of manufacturing productivity and innovation.

Business Horizons, 58±64.

Bhimani, A., & Bromwich, M. (1991). Accounting for just-in-

time manufacturing systems. CMA Magazine, 65 (1), 31±34.

Burns, T., & Stalker, G. M. (1961) The management of innova-

tion. London: Tavistock Publications.


Cohen, J., & Cohen, P. (1983). Applied multiple regression/corre-

lation analysis for the behavioral sciences (2nd ed.). Hillsdale,

NJ: Lawrence Erlbaum Associates.

Conover, W. J. (1980). Practical nonparametric statistics (2nd

ed.), New York: John Wiley & Sons.

Cooper, R. (1995). When lean enterprises collide: competing

through confrontation. Boston, MA: Harvard Business

School Press.

Daft, R. L. (1982). Bureaucratic versus nonbureaucratic struc-

ture and the process of innovation and change. In S. B.

Bacharach (Ed.), Research in the sociology of organizations,

Vol. 1 (pp. 129±166). Greenwich, CT: JAI Press.

Daft, R.L. (1986). Organization theory and design (2nd ed.). St.

Paul, MA: West Publishing.

Dodd, A.J. (1995). The just-in-time environment. in B.J. Brin-

ker (Ed.), Handbook of cost management. Boston, MA:

Warren, Gorham & Lamont.

Drucker, P., (1988, January±February) The coming of the

organization. Harvard Business Review, 45±53.

Drucker, P. (1990, May±June). The emerging theory of manu-

facturing Harvard Business Review, 94±102.

Duimering, P. R., & Safayeni, F. (1991). A study of the orga-

nisational impact of the just-in-time production system, In.

A. Satir (Ed.), Just-in-time manufacturing systems: opera-

tional planning and control issues, (pp. 19±32). New York:

Elsevier.

Flamholtz, E. G. (1983). Accounting, budgeting and control

systems in their organisational context: theoretical and

empirical perspectives. Accounting, Organizations and

Society, 8, 153±169.

Foster, G., & Horngren, C. T. (1987). JIT: Cost accounting and

cost management issues. Management Accounting, 68 (12),

19±25.

French, W. L., & Bell, C. H. Jr. (1984). Organisational devel-

opment: behavioral science interventions for organization

improvement (3rd ed.). Englewood Cli�s, NJ: Prentice-Hall.

Fry, L. W. (1982). Technology-structure research: three critical

issues. Academy of Management Journal, 25 (3), 532±552.

Galbraith, J. R. (1973). Designing complex organizations.

Reading, U.K: Addison±Wesley.

Hall, R. W. (1983). Zero inventories. Homewood, IL: Dow-

Jones Irwin.

Hall, R. W. (1987). Attaining manufacturing excellence. Home-

wood, IL: Dow-Jones Irwin.

Hay, E. J. (1988). The just-in-time breakthrough: implementing

the new manufacturing basics. New York: John Wiley &

Sons.

Heiko, L. (1991). The conceptual foundations of just-in-time.

In A. Satir (Ed.), Just-in-time manufacturing systems: opera-

tional planning and control issues (pp. 3±18). New York:

Elsevier.

Hickson, D. J., Pugh, D. S., & Pheysey, D. C. (1969). Opera-

tions technology and organization structure: an empirical

reappraisal. Administrative Science Quarterly, 14, 378±397.

Horngren, C.T., Foster, G., & Datar, S. (1994). Cost account-

ing: a managerial emphasis, 8th (ed.), Englewood Cli�s:

Prentice-Hall.

Hunt, R. G. (1970). Technology and organization. Academy of

Management Journal, 13 (3), 235±252.

Kaplan, R.S. (1986). Accounting lag: the obsolescence of cost

accounting systems. California Management Review, 28 (2),

174±199.

Kaplan, R. S., & Norton, D. (1996). Translating strategy into

action: the balanced scorecard, Boston. MA: Harvard Busi-

ness School Press.

Kass, R. A., & Tinsley, H. E. A. (1979). Factor analysis. Jour-

nal of Leisure Research, 11, 120±138.

Keller, R.T., Slocum Jr., J. W., & Susman, G.L. (1974).

Uncertainty and type of management system in continuous

process organizations. Academy of Management Journal, 17,

56±68.

Kim, J. & Mueller, C. W. (1978) Introduction to factor analysis.

Beverly Hills, CA: Sage Publications.

Klein, J. A., (1989, March±April). The human costs of manu-

facturing reform. Harvard Business Review, 60±66.

Klein, J. A. (1991). A reexamination of autonomy in light of

new manufacturing practices. Human Relations, 44 (1), 21±

38.

Lindsay, R. M. (1993). Achieving scienti®c knowledge: the

rationality of scienti®c method. In M. J. Mumford & K. V.

Peasnell (Eds.), Philosophical perspectives on accounting:

essays in honour of Edward Stamp (pp. 221±254). London:

Routledge.

Lindsay, R. M., (1995). Reconsidering the status of tests of

signi®cance. an alternative criterion of adequacy. Accounting,

Organizations and Society, 20, 35±53.

Lindsay, R. M. and Ehrenberg, A. S. C., (1993, August). The

design of replicated studies. American Statistician. 217±

228.

Lindsay, R. M., & Kalagnanam, S. (1993). The adoption of just-

in-time production systems in Canada and their association

with management control practices. Hamilton, Ontario: The

Society of Management Accountants of Canada.

Linsu, K. (1980). Organisational innovation and structure.

Journal of Business Research, 18, 225±245.

Mackey, J., & Thomas, M. (1995). Costing and the new

operations management. In D. Ashton, T. Scapens, & T.

Hopper (Eds.), Issues in management accounting. London:

Prentice Hall, pp. 87±112.

Manoochehri, G. H. (1988). JIT for small manufacturers.

Journal of Small Business Management, 26 (4), 22±30.

McNair, C. J. (1990). Interdependence and control: traditional

vs. activity-based responsibility accounting. Journal of Cost

Management (Summer), 15±24.

McNair, C. J. (1995). Responsibility accounting and controll-

ability networks. In B. J. Brinker (Ed.), Handbook of cost

management (pp. E4±1±E4±33). Boston, MA: Warren, Gor-

ham & Lamont.

McNair, C. J., Mosconi, W., & Norris, T. F. (1989). Beyond the

bottomline. Homewood, IL: Dow-Jones Irwin.

Merchant, K.A. (1985).Control in business organizations, Boston,

MA: Pitman.

Mintzberg, H., (1979). The structuring of organizations. Engle-

wood Cli�s, NJ: Prentice Hall.


Nanni Jr., A. J., Dixon, J. R., & Vollman, T. E. (1992). Inte-

grated performance measurement: management accounting

to support the new manufacturing realities. Journal of Man-

agement Accounting Research, 4, 1±19.

Neale J. M., & Liebert, R. M., (1986). Science and behavior: an

introduction to methods of research (3rd ed.), Englewood

Cli�s, NJ: Prentice Hall.

Nemetz, P. L., & Fry, L. W. (1988). Flexible manufacturing

organizations: implications for strategy formulation and orga-

nization design. Academy of Management Review, 13, 627±638.

Neumann, B. R., & Jaouen, P. R. (1986). Kanban zips and cost

accounting: a case study. Journal of Accountancy, 162 (2),

132±141.

Nunnally, J. C., (1967). Psychometric theory. New York:

McGraw±Hill.

O'Grady, K. E. (1982). Measures of explained variance: cau-

tions and limitations. Psychological Bulletin, 92, 766±767.

Otley, D. T., (1987). Accounting for divisional planning and

control. In K. R. Ferris and J. L. Livingstone (Eds.), Man-

agement planning and control: the behavioural foundations,

(pp. 83±120). Revised ed., Beavercreek, OH: Century VII

Publishing.

Parthasarthy, R., & Sethi, S. P. (1992). The impact of ¯exible

automation on business strategy and organisational struc-

ture. Academy of Management Review, 17, 86±111.

Parthasarthy, R., & Sethi, S. P. (1993). Relating strategy

structure to ¯exible automation: a test of ®t performance

implications. Strategic Management Journal, 14, 529±549.

Robbins, S. P. (1983). Organization theory: the structure and

design of organizations. Englewood Cli�s, NJ: Prentice Hall.

Robinson, M. A. (Ed.) (1989). Stanadyne diesel systems (A). In

Cases from management accounting practice (Vol. 5, pp. 19±

26). Montvale, NJ: National Association of Accountants.

Ross, G. H. B., (1990, November). Revolution in management

control.Management Accounting, 23±27.

Scapens, R. W., (1990, September). Researching management

accounting practice: the role of case study methods. British

Accounting Review, 259±282.

Scapens, R. W. (1992, December). The role of case study

methods in management accounting research: a personal

re¯ection and reply. British Accounting Review, 369±384.

Schermerhorn Jr., J. R., Hunt, J. G., & Osborn, R. N. (1985).

Managing organisational behavior, (2nd ed.), New York:

John Wiley & Sons.

Schonberger, R. J. (1982). Japanese manufacturing techni-

ques: nine hidden lessons in simplicity. New York: The

Free Press.

Schonberger, R. J. (1986). World class manufacturing: the les-

sons of simplicity applied. New York: The Free Press.

Senge, P. (1990). The ®fth discipline. New York: Doubleday

Press.

Sepehri, M. (1986). Just-in-time, not just in Japan. Falls

Church, VA: American Production and Inventory Control

Society.

Starbuck, W. H., & Dutton, J. M. (1973). Designing

adaptive organizations. Journal of Business Policy, Summer,

21±28.

Stata, R. (1989). Organisational learningÐthe key to management

innovation. Sloan Management Review, Spring, 63±74.

Steers, R. M. (1988). Introduction to organisational behavior

(3rd ed.). Glenview, IL: Scott, Foresman and Company.

Swenson, D. W., & Cassidy, J. (1993). The e�ect of JIT on

management accounting. Journal of Cost Management,

Spring, 39±47.

Thompson, J. (1967). Organizations in action. New York:

McGraw±Hill.

Vowles, A. (1993, April). Gaining competitive advantage

through organisational learning. CMA Magazine, 12±14.

Wallace, R. S. O., & Cooke, T. E. (1990). Nonresponse bias in

mail accounting surveys: a pedagogical extension. British

Accounting Review, 22, 283±288.

Wallace, R. S. O., & Mellor, C. J. (1988). Nonresponse bias in

mail accounting surveys: a pedagogical note. British

Accounting Review, 20, 131±139.

Wilkinson, B., & Oliver, N. (1989). Power, control and the

Kanban. Journal of Management Studies, 26 (1), 47±58.

Woodward, J. (1980). Industrial organization: theory and

practice. 2nd Edn. Oxford: Oxford University Press.

Yin, R. K., (1989). Case study research: design and methods (2nd

ed.). Beverly Hills, CA: Sage Publications.

Zwerman, W. L. (1970). New perspectives of organisational

theory. Westport, CT: Greenwood Press.


Documents

The use of organic models of control in JIT ﬁrms ...directory.umm.ac.id/Data Elmu/jurnal/A/Accounting... · led to the adoption of Just-in-Time (JIT) produc-tion systems. Although