Lean manufacturing: context, practice bundles, and …files.ferraz8.webnode.com/200000065-e27a6e46de/Artig… · · 2013-10-04Journal of Operations Management 21 (2003) 129–149

Journal of Operations Management 21 (2003) 129–149

Lean manufacturing: context, practice bundles, and performance

Rachna Shaha, Peter T. Wardb,∗a Carlson School of Management, University of Minnesota, Minneapolis, MN 55455, USA

b Fisher College of Business, The Ohio State University, Columbus, OH 43221, USA

Received 12 December 2000; accepted 24 June 2002

Abstract

Management literature has suggested that contextual factors may present strong inertial forces within organizations thatinhibit implementations that appear technically rational [R.R. Nelson, S.G. Winter, An Evolutionary Theory of EconomicChange, Harvard University Press, Cambridge, MA, 1982]. This paper examines the effects of three contextual factors, plantsize, plant age and unionization status, on the likelihood of implementing 22 manufacturing practices that are key facets oflean production systems. Further, we postulate four “bundles” of inter-related and internally consistent practices; these arejust-in-time (JIT), total quality management (TQM), total preventive maintenance (TPM), and human resource management(HRM). We empirically validate our bundles and investigate their effects on operational performance. The study sample usesdata fromIndustryWeek’s Census of Manufacturers. The evidence provides strong support for the influence of plant size onlean implementation, whereas the influence of unionization and plant age is less pervasive than conventional wisdom suggests.The results also indicate that lean bundles contribute substantially to the operating performance of plants, and explain about23% of the variation in operational performance after accounting for the effects of industry and contextual factors.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Contextual factors; Lean manufacturing practices; Lean bundles; Operational performance

1. Introduction

Heightened challenges from global competitorsduring the past 2 decades have prompted many USmanufacturing firms to adopt new manufacturing ap-proaches (Hall, 1987; Meredith and McTavish, 1992).Particularly salient among these is the concept oflean production (Womack and Jones, 1996; Womacket al., 1990). Lean production is a multi-dimensionalapproach that encompasses a wide variety of man-agement practices, including just-in-time, qualitysystems, work teams, cellular manufacturing, sup-

∗ Corresponding author. Tel.:+1-614-292-5294.E-mail addresses: [email protected] (R. Shah),[email protected] (P.T. Ward).

plier management, etc. in an integrated system. Thecore thrust of lean production is that these practicescan work synergistically to create a streamlined, highquality system that produces finished products at thepace of customer demand with little or no waste.Anecdotal evidence suggests that several organiza-tional factors may enable or inhibit the implementa-tion of lean practices among manufacturing plants.With the notable exception ofWhite et al. (1999),there is relatively little published empirical evidenceabout the implementation of lean practices and thefactors that may influence implementation.

A majority of articles on the topic of lean pro-duction system focus on the relationship between im-plementation of lean and performance. While mostof these studies have focused on a single aspect of

0272-6963/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0272-6963(02)00108-0

130 R. Shah, P.T. Ward / Journal of Operations Management 21 (2003) 129–149

lean and its performance implications (e.g.Hackmanand Wageman, 1995; Samson and Terziovski, 1999;McKone et al., 2001), a few studies have exploredthe implementation and performance relationship withtwo aspects of lean (e.g.Flynn et al., 1995; McKoneet al., 2001). Even fewer studies have investigatedthe simultaneous synergistic effects of multiple as-pects of lean implementation and performance impli-cation. A noteworthy exception isCua et al.’s (2001)investigation of implementation of practices related tojust-in-time (JIT), total quality management (TQM),and total preventive maintenance (TPM) programs andtheir impact on operational performance. However,conceptual research continues to stress the importanceof empirically examining the effect of multiple dimen-sions of lean production programs simultaneously.

We examine the relationship between contextualfactors and extent of implementation of a number ofmanufacturing practices that are key facets of leansystems. These contextual factors have been sug-gested as possible impediments to implementing leanproduction systems. Specifically, we focus on threecontextual factors, plant size, plant age, and extent ofunionization. Further, we extendOsterman (1994)andMacDuffie’s (1995)notion of “bundles” from humanresource practices to a larger set of manufacturingpractices. Specifically, we postulate four “bundles”of inter-related and internally consistent practices;these are just-in-time (JIT), total quality management(TQM), total preventive maintenance (TPM), and hu-man resource management (HRM). We empiricallyvalidate our bundles and further investigate their si-multaneous and synergistic effects on operationalperformance.

2. Literature and propositions

2.1. Lean practices and lean bundles

In the last several years, scholarly journals havepublished a number of articles that focus on the con-tent of lean production or comprise of case studiesthat concentrate on individual firm experiences. A re-view of this literature reveals a number of manufac-turing practices that are commonly associated withlean production.Table 1summarizes our review bycross-listing key practices identified with references.

Table 1 links substantive literature on high per-formance, lean manufacturing with each of the leanpractices addressed in our research. As can be seen,there is varying degree of frequency that each of theitems selected is considered in the studies reviewed.JIT/continuous flow production and quick changeovermethods are included most frequently, while safetyimprovement methods are referenced least frequentlyin the literature. The practices shown inTable 1emerge from a fairly extensive literature review andprovide a representative view of the componentscomprising lean production. Our discussion and mea-surement of lean production is necessarily related tothe manufacturing practices that are commonly ob-served in the literature describing high performance,lean manufacturers. This literature is also rich withdescriptions of values and philosophy associatedwith lean production that are less readily measuredthan practices. For example, lean production philos-ophy focuses on avoiding seven cardinal wastes andon respecting customers, employees and suppliers(Schonberger, 1986). Although we do not directly ad-dress such philosophical positions, we recognize thatthey are important and believe that they are reflectedin the implementation of the lean practices that we doaddress.

There are multiple ways to combine the individualpractices to represent the multi-dimensional nature oflean manufacturing. In combining these practices, theresearcher has to contend with the method used tocombine and the actual content of the combinations.The dominant method in operations management lit-erature has been to use exploratory or confirmatoryfactor analysis to combine individual practices in amultiplicative function to form orthogonal and uni-dimensional factors (Flynn et al., 1995; Cua et al.,2001). A review of research from organization theory,and labor and human resource management showsless reliance on factor analysis and offers multipleways for combining individual practices and creat-ing an index. One such method is the additive indexused byOsterman (1994)and MacDuffie (1995)indeveloping “bundles” of inter-related human resourcemanagement practices. We used the additive indexmethod for developing our lean bundles.

There is general agreement within operations man-agement literature that just-in-time (JIT), total preven-tive maintenance (TPM), total quality management

R. Shah, P.T. Ward / Journal of Operations Management 21 (2003) 129–149 131

Table 1Lean practices and their appearance in key references (adapted fromMcLachlin, 1997)

Lean practice Sources

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Bottleneck removal (production smoothing)Cellular manufacturing ∗ ∗ ∗ ∗ ∗ ∗Competitive benchmarkingContinuous improvement programs ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Cross-functional work force ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Cycle time reductions ∗ ∗ ∗ ∗ ∗Focused factory production ∗ ∗ ∗ ∗ ∗ ∗ ∗JIT/continuous flow production ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Lot size reductions ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Maintenance optimizationNew process equipment/technologies ∗ ∗ ∗Planning and scheduling strategiesPreventive maintenance ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Process capability measurements ∗ ∗ ∗ ∗ ∗Pull system/kanban ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Quality management programs ∗Quick changeover techniques ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Reengineered production processSafety improvement programs ∗ ∗ ∗Self-directed work teams ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗Total quality management ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗(1) Sugimori et al. (1977); Monden (1981); Pegels (1984); (2) Wantuck (1983); (3) Lee and Ebrahimpour (1984); (4) Suzaki (1985);(5) Finch and Cox (1986); (6) Voss and Robinson (1987); (7) Hay (1988); (8) Bicheno (1989); (9) Chan et al. (1990); (10) Piper andMcLachlin (1990); (11) White (1993); (12) Shingo Prize Guidelines (1996); (13) Sakakibara et al. (1997); (14) Koufteros et al. (1998);(15) Flynn et al. (1999); (16) White et al. (1999).

(TQM), and human resource management (HRM) areconceptually, theoretically, and empirically well es-tablished (Ahire et al., 1996; Samson and Terziovski,1999; Davy et al., 1992; Cua, 2000; Flynn et al.,1994, 1995; McKone and Weiss, 1999; McKoneet al., 1999; McLachlin, 1997; Sakakibara et al.,1997; Schroeder and Flynn, 2001; Osterman, 1994;Pil and MacDuffie, 1996; Ichniowski et al., 1994). Weselected inter-related practices to combine into fourpractice bundles associated with JIT, TPM, TQM andHRM.

We use individual practices to investigate the asso-ciation between contextual factors and the pattern ofimplementation because we are interested in the pat-tern of implementation among the 22 manufacturingpractices. We use the four bundles in examining therelationship between implementation and operationalperformance because we are interested in evaluatingthe synergistic effects of implementation of all com-plementary facets of lean.

2.2. Implementation and contextual variables

In general, the success of implementation of anyparticular management practice frequently dependsupon organizational characteristics, and not all orga-nizations can or should implement the same set ofpractices (Galbraith, 1977). Consideration of organi-zational contexts has been noticeably lacking in re-search on implementation of JIT and TQM programsor other lean manufacturing practices. Perhaps be-cause of the failure to consider organizational context,evidence on the impact of JIT and TQM programs onorganizational performance has been mixed (Adam,1994; Powell, 1995; Samson and Terziovski, 1999).

In this study, we examine three organizational con-text characteristics—unionization, plant age and plantsize—that may influence the implementation of man-ufacturing practices. A limited number of empiricalstudies suggest that implementation or adoption ofa manufacturing practice is contingent upon specific


organizational characteristics (White et al., 1999;McKone et al., 1999). For example,White et al.(1999)found significant evidence that large US man-ufacturers adopted JIT practices more frequently thansmall manufacturers. While the role of plant size inimplementation process has been studied previously,plant age and unionization status remains compara-tively unexplored. There is not only a lack of empiri-cal attention given to contextual factors’ relationshipwith lean practices, but there is also a paucity of the-ory to guide our expectations about the direction ofpossible effects. Because of the exploratory nature ofthis research, we develop a set of propositions ratherthan formal hypotheses.

2.2.1. UnionizationIt is often assumed that because implementation

of most manufacturing practices requires negotiatingchanges in work organization, unionized facilitieswill resist adopting lean practices and thus lag behindnon-unionized facilities.Drucker (1987) discussedthe problems of existing union work rules and jobclassifications in the implementation of JIT systems.In a similar vein, the business press has often as-serted that unionization prevents the adoption of some“Japanese” manufacturing practices in US manu-facturers. Further, there are also instances in whichunions have been cooperative and helpful in the imple-mentation process (Katz, 1985; Cappelli and Scherer,1989).

While there is some evidence that unionization isnegatively associated with organizational performance(Machin, 1995; Meador and Walters, 1994; Bronarset al., 1994), empirical evidence linking unionizationwith adoption and implementation of manufacturingpractices is scant. For instance,Ahmed et al. (1991)found that in a sample of 177 manufacturing firms,union’s attitude towards flexibility was not signifi-cantly related to JIT usage. In the investigation ofinfluence of unions on human resource managementpractices,Ng and Maki (1994)concluded that theimpact of union varies, depending upon the natureof the practice. In manufacturing firms,Osterman(1994) found no empirical support for associationbetween unionization and four innovative work prac-tices. Unionization seems to be an important factor forimplementation although the direction is not exactlyclear. In addition, it hasn’t been empirically investi-

gated in association with a wider set of lean manufac-turing practices. Thus, the net effect is an empiricalquestion, which we propose to test as follows:

Proposition 1: Unionized plants are less likely toimplement lean manufacturing practices than non-unionized or partially unionized plants.

2.2.2. Age of the plantPlant age may imply either a tendency toward

resistance to change or a liability of newness. The“resistance to change” view is supported by the or-ganizational sociology literature which suggests thatthe age of an establishment should inversely influencethe rate of adoption of innovations, because organiza-tional forms tend to be “frozen” at birth (Stinchcombe,1965). Evolutionary economics provides additionalsupport for this view (Aldrich, 1979; Nelson andWinter, 1982). The evolutionary perspective suggeststhat organizations develop a set of organizationalroutines (manufacturing practices) over a period oftime and these practices change infrequently. Further,the longer an organization has experience with thepractices, even if the results are inferior relative tothe new practices, the harder it is for the organiza-tion to replace the older, inferior practices (Pil andMacDuffie, 1996). So, an older organization findsimplementing new practice(s) more challenging. Or-ganizational ecologists espouse the same view as“liability of newness.” This perspective suggests thatolder organizations have an advantage over youngerones because it is easier to continue existing routinesthan to create or borrow new ones, even if the newroutines are inherently superior (Nelson and Winter,1982; Hannan and Freeman, 1984).

Thus, irrespective of the theoretical perspective,plant age is found to impede adoption and implemen-tation of new, innovative work practices. However,empirical evidence from industrial and labor relationliterature indicates that age of an establishment is nota significant determinant of adoption of work prac-tices. Specifically,Osterman (1994)using data from694 US manufacturing establishments found that agewas not associated with the adoption of innovativework practices such as teams, job rotation, quality cir-cles, and total quality management.Osterman (1994)study was limited to innovative work practices relatedto human resource management.


We study a wider set of manufacturing practicesrelated not only to human resources but also work-flow in production, preventive maintenance and qual-ity management programs. In this regard, anecdotalevidence suggests that newer plants have a natural ad-vantage in implementing new lean practices becauseof a younger, arguably less cynical workforce and alsobecause of fewer physical barriers to lean practicessuch as set up time reduction. This implies that plantage has a negative impact on the likelihood of im-plementation of lean manufacturing practices. Thus,we propose the following with regards to age of theplant:

Proposition 2: Older plants are less likely to imple-ment lean manufacturing practices than newer plants.

2.2.3. Size of the plantSeveral authors (e.g.Chandler, 1962; Child, 1972)

have noted that since any administrative task tends tobe more complicated in large firms, managers maynot even attempt to change, instead they may allowexisting systems to linger. This is equally true ofimplementation of new operational practices. Thatis, large organizations suffer from structural inertialforces (Hannan and Freeman, 1984) that negativelyeffect the implementation of lean manufacturing prac-tices. Further, inertial effects of size are more preva-lent in manufacturing industry than in service industry(Gopalakrishnan and Damanpour, 1997). However,large size also implies the availability of both capi-tal and human resources that facilitate adoption andimplementation of lean practices as well as returns toscale for investments associated with lean practices.The influence of size is pervasive and has been iden-tified in relation to technology practices (Germainet al., 1994)and manufacturing practices (White et al.,1999), across industries as diverse as hospitals (Moch,1976), education (Baldrige and Burnham, 1975), aswell as manufacturing (Ahmed et al., 1991). Thus,while theoretical arguments can be made both in sup-port of a positive and a negative association betweenlarge size and implementation of lean practices, em-pirical evidence overwhelmingly supports a positiverelationship. That is, despite the inertial effects, largefirms are more likely to implement lean practices thantheir smaller counterparts. So, we propose the follow-ing with regard to firm size and implementation:

Proposition 3: Large manufacturers are more likely toimplement lean practices than small manufacturers.

Each of these issues has the same underlyingtheme—strong inertial forces within organizationsthat prevent even implementations that appear techni-cally rational from manifesting themselves in practice(Nelson and Winter, 1982) balanced against a com-petitive imperative of continuous improvement. Thus,organizational context may significantly impact theimplementation of lean manufacturing practices.

2.3. Bundles of lean manufacturing practices andoperational performance

Lean practices are generally shown to be associ-ated with high performance in a number of studiesof world-class manufacturing (e.g.Sakakibara et al.,1997; Giffi et al., 1990). Overall, review of related re-search indicates that implementation of lean practicesis frequently associated with improvements in oper-ational performance measures. The most commonlycited benefits related to lean practices are improvementin labor productivity and quality, along with reductionin customer lead time, cycle time, and manufacturingcosts (Schonberger, 1982; White et al., 1999).

Most of the empirical studies focusing on the im-pact of lean implementation on operational perfor-mance are constrained to one or two facets of lean,often JIT or TQM. Improved operational performanceassociated with JIT practices (Koufteros et al., 1998;Sakakibara et al., 1997; Im and Lee, 1989; White et al.,1999) has been shown to outweigh the results asso-ciated with TQM practices (Samson and Terziovski,1999; Powell, 1995). In a study of JIT and TQM,Flynnet al. (1995)found incremental performance effectsattributed to JIT and infrastructural practices commonto both JIT and TQM, but not specific TQM practices.In case of TQM practices by themselves, whileChoiand Eboch (1998), andSamson and Terziovski (1999)found a significant direct impact of TQM practiceson operational performance,Adam (1994)andPowell(1995) found little impact of TQM practices on op-erational and other measures of performance.Powell(1995) found positive support for only 3 out of 12TQM practices that he studied.

Despite variable evidence regarding performanceimprovements related to these lean practices, relatively


little empirical research exists to gauge the extent ofperformance improvements. Further, an empirical in-vestigation of simultaneous use of several manufac-turing programs representing multiple facets of leanis also lacking in the literature. We seek to addressthis apparent gap in literature by examining the per-formance implications of implementing lean practicesafter controlling for the effects assigned to industrydifferences and the contextual factors.

Many researchers argue that a lean production sys-tem is an integrated manufacturing system requiringimplementation of a diverse set of manufacturingpractices (e.g.Womack and Jones, 1996). Further,they also suggest that concurrent application of thesevarious practices should result in higher operationalperformance because the practices, although diverse,are complementary and inter-related to each other. Forinstance, reduction in work-in-process (WIP) inven-tory goes hand in hand with development of humanresource capability in the form of empowered/self-directed work teams capable of solving problems.Thus, the problem solving capabilities that arise asa result of empowered work teams can help boostperformance by identifying root causes of qualityproblems, by helping to improve workflow, and byimproving equipment efficiency. Thus, we posit thatsimultaneous application of multiple aspects of leanmanufacturing will have a significant positive impacton operational performance:

Proposition 4: Implementation of lean bundles, eachrepresenting groups of related lean practices, will havea positive impact on operational performance.

3. Methods

3.1. Instrument development and data collection

Penton Media Inc., publishers ofIndustryWeek andother manufacturing-related publications, conduct anannual survey of manufacturing managers and madetheir 1999 data available to the authors.IndustryWeek(IW) is an industrial magazine targeted at executivesand managers of US manufacturing firms. PentonMedia Inc. and PricewaterhouseCoopers jointly de-veloped a mail survey with input from a number ofmanufacturing experts external to both firms. The

survey included four pages of questions related to hu-man resource management, quality issues, customerand supplier relations, and technology.

The study sample consists of approximately 28,000subscribers to Penton Media Inc.’s manufacturing-related publications, and includes managers of plantsbelonging to manufacturing firms. Survey recipientshold titles such as plant manager, plant leader, andmanufacturing manager. The unit of analysis in thepresent study is the manufacturing plant. Each in-dividual in the sample was sent a letter describingthe survey, a survey form, a business reply envelope,and a separate participation card to ensure anonymityof responses. The questionnaires were mailed inmid-April 1999 and completed questionnaires wereaccepted through early June 1999. A total of 1757completed surveys were received corresponding toa response rate of approximately 6.7%. While theresponse rate is smaller than many empirical stud-ies, it compares favorably with large sample oper-ations surveys (e.g.Roth and van der Velde, 1991;Stock et al., 2000). In light of the low response rate,non-response bias was assessed by comparing theproportion of respondent firms to proportion of totalmailed surveys for each SIC code. The data indi-cate no significant difference (χ2 = 12.3, d.f . = 19,P = 0.83).

3.1.1. Sample characteristicsThe sample resembles the population profile re-

ported byUS Census of Manufacturers (1997)andis not unlike those for similar studies of US manu-facturers (e.g.Im and Lee, 1989; White et al., 1999).A comparison of the number of establishments inthe IW sample with the population profile reportedby US Census of Manufacturers (1997)is shown inTable 2. The comparison indicates that compared tothe population as a whole, the IW sample is some-what biased in favor of paper, chemicals, primarymetal, electric and electronic equipment and trans-portation equipment. Despite the differences from theUS population’s industry composition, the sampleprovides diverse and fairly representative industrialcoverage.

The manufacturers included in the sample representa wide variety of industries. Although implementinglean practices is usually associated primarily with dis-crete part manufacturers, pervasiveness of practices


Table 2Comparison of the study sample withUS Census of Manufacturers (1997)by SIC code

Standard industrial classification Study sample US Census (1997)

Number % %

(20) Food and products 66 3.8 5.5(21) Tobacco products 3 0.2 0.04(22) Textile mill products 41 2.3 1.6(23) Apparel and other textile 22 1.3 6.2(24) Lumber and wood 39 2.2 9.7(25) Furniture and fixtures 58 3.3 3.2(26) Paper 86 4.9 1.7(27) Printing and publishing 27 1.5 16.4(28) Chemicals 130 7.4 3.3(29) Petroleum and coal 10 0.6 0.6(30) Rubber and miscellaneous plastic 107 6.1 4.4(31) Leather 6 0.3 0.5(32) Stone, clay, and glass 53 3.0 4.3(33) Primary metal 90 5.1 1.7(34) Fabricated metal 202 11.5 10.1(35) Machinery, except elect. 307 17.5 14.9(36) Electric and electronic equipment 219 12.5 4.5(37) Transportation equipment 141 8.1 3.3(38) Instruments and related products 112 6.4 3.1(39) Miscellaneous manufacturing 29 1.7 4.8

Totala 1748 100 100

a SIC code was not provided by nine respondents.

across the industrial spectrum is unknown. For thisreason, we include respondents from all manufactur-ing industries (SIC 20–39) in the sample. We reportthe analysis related to industry effects in our findings.A majority of plants included in the sample have anon-union production environment (65%), are morethan 20 years old (72%), and are relatively small insize, having fewer than 250 employees (52%). The65% of respondent plants that are not unionized cor-responds to an estimated 59% of all manufacturingplants that are not unionized as reported in the mostrecent economic census (US Census of Manufactur-ers, 1997). All of the manufacturers included in thesample have implemented at least one lean practice.

3.1.2. Contextual variablesEach of the contextual variables is re-coded into

three categories. When a union represents none ofthe plant’s production workforce, we classify it as“non-unionized” plant, instances of 100% unioniza-tion are classified as “unionized” plants, and plantswithin these two extremes are classified as “partlyunionized” plants. Guidelines for classifying plants on

the basis of age are scant.IndustryWeek suggests cate-gorizing plants less than 10 years old as “new” plantsand more than 20 years old as “old” plants. We usethese criteria for our study. In addition, we classifyplants between 10 and 20 years of age as “adolescent”plants. Size is re-coded into small, medium and largeplants. We designate plants with fewer than 250 em-ployees as “small” followingAdams and Ponthieu(1978, p. 2) and plants with more than 1000 employ-ees as “large” followingWhite et al. (1999)and theplants between 250 and 999 as medium sized plants.Survey items appear inAppendix A.

3.1.3. Lean manufacturing practicesLean manufacturing practices are measured on a

three-point scale ((1) no implementation; (2) someimplementation; (3) extensive implementation). Scalelength has been the focus of considerable discussionin the survey methods literature. While some researchshows that reliability increases with scale length(Lissitz and Green, 1975; Jenkins and Taber, 1977),there is no clear-cut relationship between validity andscale length (Smith, 1994; Smith and Peterson, 1985).

136R

.Shah,

P.T.W

ard/Journal

ofO

perationsM

anagement

21(2003)

129–149


Specifically,Morrison (1972)shows that three-pointscales capture most real information about constructrelationships and are thus considered adequate. Acorrelation matrix for each of the contextual variablesand lean practice included in the analysis is shown inTable 3.

3.1.4. Lean bundlesThe 22 individual lean practices were combined

into 4 lean bundles. For instance, all practices relatedto production flow were combined to form the JITbundle. The underlying rationale is that JIT is a man-ufacturing program with the primary goal of contin-uously reducing, and ultimately eliminating all formsof waste (Sugimori et al., 1977). Two major formsof waste are work-in-process (WIP) inventory andunnecessary delays in flow time. Both can be reducedby implementing practices related to production flowsuch as lot size reduction, cycle time reduction, quickchangeover techniques to reduce WIP inventory andby implementing cellular layout, reengineering pro-duction processes, and bottleneck removal to reduceunnecessary delays in the production process.

Practices related to continuous improvement andsustainability of quality products and process werecombined to form the TQM bundle. It includes prac-tices such as quality management programs, formalcontinuous improvement programs and process capa-bility measurement capability.

The TPM bundle includes practices primarily de-signed to maximize equipment effectiveness throughplanned predictive and preventive maintenance of theequipment and using maintenance optimization tech-niques. More generally, emphasis on maintenance mayalso be reflected by the emphasis given to new pro-cess equipment or technology acquisition (Cua et al.,2001).

The HRM bundle has significant theoretical andempirical support. The most commonly cited prac-tices are job rotation, job design, job enlargement,formal training programs, cross-training programs,work teams, problems solving groups, and employeeinvolvement (Ichniowski et al., 1994; MacDuffie,1995; Osterman, 1994). We include two practicesin the HRM bundle—flexible, cross-functional workforce, and self-directed work teams. Flexible, cross-functional work force, and self-directed work teamsare higher level practices that include many lower

level practices. Conceptually, in order for an organiza-tion to have a flexible cross-functional work force, itneeds to have a job-rotation program, it needs to con-sider job design, and formal, cross-functional trainingprograms have to be in place. Similarly, self-directedwork teams require that employees are organized inwork teams and involved in problem solving groups.Boyer and Pagell (2000)have suggested eliminat-ing lower level variables when these are included inhigher level constructs. Thus, we combine these twopractices into HRM bundle to represent a much widerset of underlying HRM practices.

Each of the bundles was formed by adding thescores for each of the practices included in the bundlefor each responding plant. The bundles were empir-ically validated using reliability analysis (Cronbachcoefficientα) and principal components analysis withvarimax rotation. All 22 lean practices were enteredfor principal component analysis (PCA). Varimaxrotation was used to extract orthogonal components(seeTable 4) and four components were extracted.The components conform almost perfectly to theconceptual bundles with the sole exception of “com-petitive benchmarking.” Competitive benchmarkingis conceptually linked to TQM practices and was inthe TQM bundle. The rotated component matrix re-sults indicate that it has a higher loading on the TPM(0.364) bundle compared to TQM (0.361). However,because the difference in the loadings is not statisti-cally significant we opted to include it in the TQMbundle.

The results were replicated using oblique rotationas a check for orthogonality and the extracted compo-nents were identical. Further, we checked the unidi-mensionality of each of the components by conductingPCA at the component level. More specifically, weentered only the items that load on one component toensure that they do indeed load on one component. Allfour components are unidimensional. Cronbach’s co-efficientα is above the acceptable limit except for hu-man resource management bundle, which is a two-itemfactor. The value of Cronbachα depends upon thenumber of items in the scale and the average inter-itemcorrelation (Carmines and Zeller, 1979, p. 45). Thus,keeping the average inter-item correlation constant, asthe number of items increase, the value ofα also in-creases. In this instance, if the number of items in theHRM practice bundle were increased from 2 to 4 while


Table 4Principle components analysis to validate lean bundles–rotated component matrix

Factor loadings

JIT TPM TQM HRM

Lot size reductions 0.659 0.062 0.007 0.031JIT/continuous flow production 0.649 0.081 0.213 0.116Pull system 0.647 −0.147 0.256 0.118Cellular manufacturing 0.631 −0.234 0.180 0.105Cycle time reductions 0.586 0.248 0.014 0.054Focused factory production systems 0.562 0.051 0.170 0.164Agile manufacturing strategies 0.552 0.327 0.075 0.146Quick changeover techniques 0.537 0.336 0.030 −0.064Bottleneck/constraint removal 0.501 0.349 0.126 0.151Reengineered production processes 0.440 0.288 0.138 0.023Predictive or preventive maintenance −0.001 0.715 0.198 0.116Maintenance optimization 0.038 0.681 0.168 0.176Safety improvement programs 0.012 0.552 0.240 0.089Planning and scheduling strategies 0.314 0.458 0.050 0.141New process equipment or technologies 0.248 0.418 0.147 −0.197Competitive benchmarking 0.256 0.364 0.361 0.073Quality management programs 0.024 0.178 0.741 0.079Total quality management 0.177 0.219 0.705 0.160Process capability measurements 0.211 0.101 0.660 −0.079Formal continuous improvement program 0.179 0.271 0.605 0.206Self-directed work teams 0.138 0.128 0.208 0.758Flexible, cross-functional workforce 0.259 0.177 0.042 0.710

Eigenvalue 5.88 1.98 1.25 1.05Initial percent of variance explained 26.74 9.01 5.67 4.72Rotation sum of squared loadings (total) 3.79 2.58 2.39 1.39Percent of variance explained 17.23 11.74 10.85 6.33Cronbachα (sampleN) 0.81 (1508) 0.74 (1579) 0.74 (1588) 0.51 (1709)

Extraction method: principal component analysis. Rotation method: varimax with Kaiser normalization.

keeping the average inter-item correlation constantat 0.51, the Cronbachα value will increase to above0.79.

3.1.5. Operational performanceA six-item scale is used to measure the operational

performance of a manufacturing plant. The items in-clude 5-year changes in manufacturing cycle time,scrap and rework costs, labor productivity, unit man-ufacturing costs, first pass yield, and customer leadtime. A scale was constructed for operational perfor-mance measure based on principal components anal-ysis of these items and the factor scores were used asthe dependent variable. As shown inTable 5, all sixitems load on one factor, with an eigenvalue of 2.37explaining almost 40% of the variation. Cronbach’scoefficientα is 0.69.

The construct validity issue of whether operationalperformance is unidimensional in a conceptual senseis a separate question from the empirical issue. Twocompeting and equally valid conceptual views onperformance are the tradeoff view and the productioncompetence view popularized bySkinner (1969)andVickery (1991), respectively. We argue that firms arecapable of developing multiple internal competences.Further, global competition requires that firms developand possess these multiple competences in order tosucceed. Our measure focuses on three underlyingaspects that are highly related: lead time, cost, andconformance quality. The very basis of JIT is thatmanufacturing processes that are faster and moreprecise with regard to first-time-through quality arealso inherently less costly. Thus, we contend thatthese six measures are conceptually and empirically

R.

Shah,P.T.

Ward

/Journalof

Operations

Managem

ent21

(2003)129–149

139


related and represent a single dimension of operationalperformance.

4. Results

4.1. Context and implementation of lean practices

We use likelihood ratioχ2-test statistic to evaluatethe relationship of association between contextual fac-tors and implementation of lean practices. For exam-ple, for plant size and cellular manufacturing, the 3×3(small/medium/large size× no/some/extensive imple-mentation) matrix is decomposed into three separate2 × 2 component matrices. The likelihood ratios arethen compared to evaluate the direction of the overallassociation. Thus, we computed a likelihood ratioχ2

statistic for each of the 22 lean manufacturing prac-tices and each individual contextual factor. The resultsare discussed below for each of the contextual factors.

4.1.1. Union representationOf the 22 lean manufacturing practices, 6 are signi-

ficantly associated with unionization. These practicesare cellular manufacturing, cross-functional workforce, cycle time reduction, maintenance optimiza-tion, process capability measurements, and self-direc-ted work teams. The evaluation of 2× 2 matricesfor each of the six practices indicates that union-ization is negatively related to implementation. Thissuggests that unionized plants are less likely to im-plement these six practices. It is interesting to notethat both practices related to managing work force(i.e. cross-functional work force and self-directedwork teams) are negatively related to unionization.No statistically significant relationship was found forthe other 16 practices. Contrary to popular belief, theresults show that unionization status does not havea significant impact on extensive implementation ofmost lean practices. This result is somewhat surprisingin light of conventional wisdom about the difficulty ofimplementing improvement programs in a unionizedenvironment. However, the findings do lend supportto observations about the difficulty of changing workforce rules in a union environment (Table 6).

4.1.2. Age of plantThe likelihood ratioχ2 indicates that 8 out of the 22

lean practices are significantly influenced by age of the

plant. Five of the eight have a significant negative asso-ciation between age of plant and implementation of thepractice, implying that old plants are less likely to im-plement these practices than newer plants. These fivepractices include cross-functional work force, cycletime reduction, JIT/continuous flow production, main-tenance optimization, reengineered production processand self-directed work teams. Three lean practices,planning and scheduling strategies, safety improve-ment programs, and total quality management pro-grams have significant positive association with ageof plant. This implies that old plants are more likelyto implement these practices relative to new plants.The results are consistent with the debate in the lit-erature about “liability of newness” and “liability ofobsolescence” (Hannan and Freeman, 1984; Baum,1989). No significant relationship was found betweenplant age and implementation in 14 other instances.These results indicate that although plant age has someimpact on the implementation of lean practices, thedirection of the effect is not always as predicted.

4.1.3. Size of the plantOf the 22 lean practices examined, plant size sig-

nificantly impacts all but two practices. No significantrelationship was found between size and implemen-tation of two practices: cross-functional work force,quality management programs. As predicted all theeffected practices have a significant positive associ-ation between plant size and implementation. Thissuggests that large plants are likely to implement thetwenty practices more extensively compared to smallplants. The findings with respect to size are generallyconsistent with the findings ofWhite et al. (1999).

4.1.4. Evaluation of Propositions 1–3Likelihood ratio χ2 statistic and the individual

2 × 2 matrix analyses provide unambiguous supportfor Proposition 3 that large size has a significantimpact in the implementation of 20 out of 22 leanpractices. However, the evidence is not as apparent forProposition 1 (implementation is negatively relatedto unionization status) and for Proposition 2 (imple-mentation is negatively related to age of the plant).Unionized plants are less likely to implement only 6of the 22 practices compared to non-unionized plants.Otherwise, unionization has no significant effect onlikelihood of implementing any other lean practices.


Table 6Pearson’sχ2 to test for association between contextual variables and implementation of lean practicesa

Unionization Age Size

Agile manufacturing strategies 2.85 0.92 14.30∗∗Bottleneck removal 0.01 1.39 20.73∗∗∗Cellular manufacturing (17.89)∗∗∗ 8.18 23.86∗∗∗Competitive benchmarking 7.62 1.32 70.32∗∗∗Continuous improvement programs 6.24 8.54 44.32∗∗∗Cross-functional work force (39.99)∗∗∗ (18.07)∗∗∗ 0.50Cycle time reduction (10.68)∗ 7.23 21.86∗∗∗Focused factory production systems 0.85 8.35 40.10∗∗∗JIT/continuous flow production 1.37 (12.69)∗ 35.44∗∗∗Lot size reductions 2.72 2.58 18.96∗∗∗Maintenance optimization (13.75)∗∗ (10.75)∗ 22.35∗∗∗New process equipment or technologies 7.24 4.35 48.73∗∗∗Planning and scheduling strategies 5.13 13.03∗∗ 13.04∗∗Preventive maintenance 3.24 6.72 33.19∗∗∗Process capability measurements (11.06)∗ 6.44 75.06∗∗∗Pull system/kanban 4.78 7.70 42.52∗∗∗Quality management programs 1.56 6.04 6.75Quick changeover techniques 2.53 5.07 22.54∗∗∗Reengineered production process 0.88 (10.72)∗∗ 16.34∗∗Safety improvement programs 8.61 13.92∗∗ 25.90∗∗∗Self-directed work teams (18.50)∗∗∗ (15.87)∗∗ 20.74∗∗∗Total quality management 6.50 12.00∗ 13.72∗∗

a All significant associations are positive except for those in parentheses, which indicate negative association.∗ P < 0.05.∗∗ P < 0.01.∗∗∗ P < 0.001.

Similarly, Proposition 2 suggests that plant age isnegatively associated with implementation. We findmixed support for it. Our analyses indicate that 5 ofthe 22 lean practices are less likely to be implementedin older plants. However, three additional practicesare found to be significant but not in the predicteddirection.

To show that our findings do not result merely fromchance, we calculate the probability of obtaining ourspecific results. For instance, large size has a signif-icant impact on implementation 20 out of 22 times.Thus, we wish to calculate the probability of twentysignificant results by calculating the probability of 20Type I errors, assuming there is no size effect. A gen-eral Ho can be stated as:

Ho: Organizational context has no impact on imple-mentation status of a lean practice.

We use a binomial probability distribution and cal-culatep(x), the probability of “x” (number of) Type

I errors for each of the contextual variables. Theprobability of getting 20 significant results by chance[p(20) Type I errors] is almost zero. Thus, we cansay with nearly complete confidence that large sizemakes a significant impact on implementation statusof lean practices. Similarly, six lean practices weremore likely to be implemented by non-unionizedfacilities. The probability of getting six significant re-sults by chance [p(6) Type I errors] is almost 0.00008.Impact of plant age can be evaluated by examiningthe results of a trinomial distribution. Because fivelean practices are more likely to be implementedby newer plants, we look at the probability of ob-taining these results by chance. The probability ofgetting five significant results by chance [p(5) TypeI errors] is almost 0.00003 while the probability ofobtaining three significant results is 0.006, indicat-ing that one can be slightly more than 99% certainthat plant age influences implementation of leanpractices.


4.2. Context, lean bundles, and performance

A hierarchical regression model is used to test theincremental effect of the lean practices on operationalperformance (Cohen and Cohen, 1975; Boyer et al.,1997). We use this technique to assess the incrementaleffects of lean bundles by controlling for the effects ofindustry, production process type, and organizationalcontext. Operational performance is the dependentvariable in each of the regression models.

First, 19 dummies for 2 digit SIC codes are enteredas the regressor into the regression model to represent20 industries present in the dataset. Various authorshave noted the effect of industry on performance(Mauri and Michaels, 1998; McGahan and Porter,1997; Dess et al., 1990). Therefore, this allows usto evaluate the incremental effect of organizationalcontext and lean manufacturing practice indepen-dent of industry effects. Second, plant size, plantage, and unionization are entered into the regressionmodel as regressors. This model establishes a base-line to serve as an effective control for the effectsof context on performance. Finally, the factor scoresfor the four bundles are entered into the regressionequation to establish their direct and incrementaleffect on operational performance after industryand context has been considered. We conducted thehierarchical regression analysis using the additivescores for the bundles as well, with almost identicalresults.

4.2.1. Hierarchical regression with operationalperformance as the dependent variable

Table 7shows the results of a hierarchical regres-sion with operational performance as the dependentvariable and industry effects, contextual variables, andlean practices entered in the sequential manner de-scribed above. Industry effects account for a smallbut significant amount of variance (adjustedR2 =0.024, P < 0.001). This result is in conformancewith McGahan and Porter (1997), who also foundthat the industry effects accounted for a relativelysmaller portion of profit variance in manufacturing.Contextual variables as a group also account for asignificant but small amount of variance (incremen-tal R2 = 0.006,P < 0.051), and plant age is foundto have a significant negative impact on operationalperformance.

Table 7Results from hierarchical regression analysis-dependent variableis factor score for operational performance

Standardized� coefficients

Model 1 Model 2 Model 3

Dummy 20 −0.120 −0.123∗ −0.065Dummy 22 −0.009 −0.013 0.009Dummy 23 −0.029 −0.035 −0.023Dummy 24 −0.116∗∗ −0.120∗∗ −0.070Dummy 25 −0.045 −0.046 −0.034Dummy 26 −0.065 −0.068 −0.026Dummy 27 −0.024 −0.025 0.014Dummy 28 −0.070 −0.074 −0.029Dummy 29 −0.037 −0.039 −0.023Dummy 30 −0.049 −0.051 −0.027Dummy 31 −0.068∗ −0.071∗ −0.035Dummy 32 −0.056 −0.055 −0.014Dummy 33 −0.010 −0.017 0.026Dummy 34a −0.007 −0.003 −0.013Dummy 35a −0.091 −0.096 −0.058Dummy 36a 0.023 0.015 0.017Dummy 37a −0.028 −0.046 −0.097Dummy 38 0.003 −0.003 0.023

Union 0.017 0.027Size 0.026 −0.068∗Age −0.079∗∗ −0.072∗∗

JIT 0.431∗∗∗TPM 0.176∗∗∗TQM 0.190∗∗HRM 0.097∗∗

R2 0.040 0.046 0.277AdjustedR2 0.024 0.028 0.261Change inR2 0.040 0.006 0.231P-value of F statistic 0.001 0.051 0.000P-value of overall model 0.001 0.001 0.000

a The variance inflation factor is greater than 5 for each ofthese variables in each of the models indicating multicollinearity.

∗ P < 0.05.∗∗ P < 0.01.∗∗∗ P < 0.000.

The inclusion of bundles representing lean prac-tices (Model 3) results in a change inR2 of 0.231,which is significant (P < 0.000). Overall, the modelexplains 27.7% of the variance in operational perfor-mance, with an associated significance atP < 0.000as indicated by theP-value of theF-test statistic. Wealso ran a fourth model, not shown, that included alltwo-way interactions between all pairs of contextualvariables and lean bundles. These interaction termswere not significant.


4.2.2. Evaluation of Proposition 4The analysis suggests that Proposition 4 (lean

bundles are associated with higher performance) issupported. Each of the four bundles is positively as-sociated with operational performance. As a group,they explain about 23% of the variation in operationalperformance even after accounting for the effects ofindustry and organizational context. The signs of allsignificant coefficients are positive, as expected. Thisprovides support for the synergistic effects of imple-menting practices representing multiple aspects oflean manufacturing.

We also find no effect from unionization on perfor-mance after accounting for the effects of industry. Thismeans that being unionized is unrelated to achievingsuperior values on the cost, time and quality measuresthat constitute operational performance. Plant age, onthe other hand, is negatively associated with opera-tional performance, so that older plants do appear tobe at a performance disadvantage.

We find that in Model 3 (all variables included),plant size is negatively associated with operational per-formance. The size coefficient was not significant inModel 2 (lean bundles not included). There is no evi-dence that size is collinear with other regressors. Thisfinding indicates that large size does not translate intoan advantage in operational performance and whenthe effects of lean bundles are considered, large sizeis actually a disadvantage with respect to operationalperformance.

Tolerance values, which are used to indicate thedegree of multicollinearity, can range from 0 to 1 withvalues near 1 indicating the presence of a high degreeof collinearity. The tolerance values of the contextualfactors with lean bundles in the model range from0.833 to 0.907 indicating that some degree of multi-collinearity affects these variables. To assess if mul-ticollinearity between variables could cause potentialproblems in attributing performance effects to individ-ual regressors, we calculated variance inflation factors(VIF) for all coefficients (26 in all) in the regressionequation for Model 3 to determine the existence ofmulticollinearity. Some experts in the field suggestthat values below five indicate that multicollinearityproblems do not affect the coefficient (Freund andLittell, 1986; Neter et al., 1990). VIF for four co-efficients, Dummy 34, Dummy 35, Dummy 36, andDummy 37 (fabricated metal, machinery, electronics,

and transportation equipment) exceeded the value of5, and VIF for all other coefficients was found tobe below 5. Thus, the overall regression model al-lows us to attribute performance effects to individualvariables except for the four industry dummies notedabove.

4.3. The effect of industry

The popular roots of lean manufacturing lie in theToyota production system and its many imitators indiscrete part manufacturers around the world. How-ever, lean practices clearly exist in every industry atleast, in isolated pockets (Womack and Jones, 1996).The question of the extent of implementation of leanpractices across industries has not been exploredempirically. The question is salient in this researchbecause it would not be productive to investigate in-dustries where lean practices are not implemented orare only implemented to a limited extent.

We address the issue of extent of implementationof lean practices across industries by selecting twosets of industries, one that is representative of discretemanufacturing and the other that is largely processmanufacturing. We then compare mean values ofeach of the four lean bundles for the two industrygroups. Results of the comparison are reported inTable 8.

Significant differences between process and dis-crete industries are found in two of the four bundles.As might be expected, plants in discrete industries aremore likely to implement JIT than those in the pro-cess industries where kanbans and small productionlots are hard to imagine. Interestingly, TPM practicesare more likely to be implemented by plants in pro-cess industries than in discrete industries. The meanvalues of TPM bundle between discrete and processindustry are almost a perfect mirror image of meanvalues of JIT bundle. This finding makes sense whenone considers the high degree of importance placedon capacity utilization in process industries (Hayesand Wheelwright, 1984). The remaining two bundles,TQM and HRM, present no significant differencesbetween discrete and process industries.

Our conclusion is that bundles of lean practicesare implemented across the industrial landscape. JITpractices are implemented more commonly in discreteindustries but are fairly commonly implemented in


Table 8Industry effects: analysis of variance comparing mean values of implementation of each of the lean bundles by industry category (processor discrete part)a

JIT TPM TQM HRM

Meanb

Process (n = 251) −0.596 0.546 0.032 −0.029Discrete (n = 829) 0.268 −0.214 −0.001 0.034

Total (n = 1080) 0.067 −0.037 0.003 0.020

ANOVASum of squares

Between groups 143.84 111.35 0.28 0.77Within groups 970.89 945.84 1038.61 1076.50

Total 1114.73 1057.19 1038.89 1077.27

d.f.

Between groups 1 1 1 1Within groups 1078 1078 1078 1078

Total 1079 1079 1079 1079

Mean square

Between groups 143.84 111.35 0.28 0.77Within groups 0.90 0.88 0.96 0.99

F statistic 159.71 126.91 0.29 0.77

Significance 0.000 0.000 0.591 0.381

a Process industries include SICs (20) food products, (22) textile mill products, (26) paper, and (28) chemicals. Discrete part industriesinclude SICS (34) fabricated metals, (35) machinery, (36) electronics, (37) transportation equipment, and (38) instruments.

b Standardized mean values of each of the lean bundles (JIT, TPM, TQM, HRM).

process industries as well. The opposite relationshipexists between TPM and industry type. TQM andHRM are implemented to about the same extent acrossboth discrete and process manufacturing. These find-ings suggest that lean practices are prevalent in allindustries and that studies of lean manufacturing neednot be restricted to industries associated with discretepart manufacturing.

5. Conclusion

This research suggests two major findings. First, or-ganizational context, i.e. plant size, unionization andplant age, matters with regard to implementation oflean practices, although not all aspects matter to thesame extent. Second, applying synergistic bundles oflean practices concurrently appears to make a substan-tial contribution to operational performance over and

above the small but significant effects of context. Wediscuss each of these findings in turn.

5.1. The importance of context

Each of the contextual variables under study is as-sociated with a significant lore about their impact onthe ability of manufacturers to implement various leanmanufacturing practices. The conventional wisdomassociated with unionization is that it is encumbering;that progress toward lean production is very difficultin the presence of unionized work force. Our findingsdo not bear out this contention. In fact, for 16 out of22 lean practices there was no significant differencein the likelihood of implementation between unionand non-union plants. However, the six practices thatunionized plants are less likely to implement include“cross-functional workforce,” a particularly thornyissue for plants with inflexible work rules imposed


by collective bargaining agreements. Although thefindings that we report suggest that having a unionizedworkforce is generally not a good reason for neglect-ing lean practices, certain practices are observed lessfrequently in a union environment.

Similar to the case of unionization, the expecta-tion is that older plants, often employing older workforces, are more resistant to the changes requiredto implement lean practices. Again, the findings donot fully substantiate this contention. Our findingssuggest that older plants are less likely to implementonly five practices relative to newer plants. Includedin this list are such important elements of a leanproduction system as cross-functional work force,cycle time reduction, JIT/continuous flow production,maintenance optimization, reengineered productionprocess and self-directed work teams. However, like-lihood of implementation of 14 of the lean practicesis unaffected by plant age, and older plants are actu-ally more likely to implement three practices relativeto newer plants. Thus, the evidence is mixed. Thedata indicate that plant age is a more important in-hibitor to implementing lean practices than is union-ization, but the findings do not provide much coverfor managers of older plants seeking a reason whylean practices cannot work in their plant. In fact, formany lean practices plant age is just not a significantfactor.

More evidence exists supporting the idea that largeplants are more likely to possess the resources toimplement lean practices than smaller plants. Thesefindings are consistent with the literature (e.g.Whiteet al., 1999). Larger plants are more likely thansmaller plants to extensively implement all but five ofthe lean practices under study. Interestingly, no dif-ferences in implementation likelihood were found foreither cross-functional workforce or quality manage-ment programs. An explanation may be that both ofthese practices are extremely important to achievingeconomies in smaller, high mix, plants and there-fore these practices receive high level of attention insmaller plants.

We also find that implementation of lean bundlesis ubiquitous across industries with reasonable highlevel of implementation of each of the lean bundlesoccurring in both process and discrete part industries.Although plants in discrete parts industries are some-what more likely to implement JIT and plants in pro-

cess industries are somewhat more likely to implementTPM, both types of plants implement both practicebundles. TQM and HRM bundles were about as likelyin process or discrete parts plants.

Overall, the evidence presented here suggests thatthe organizational context significantly affects the like-lihood of implementing lean practices. In particular,the influence of plant size appears to be substantialacross a wide mix of practices. The influence of union-ization and plant age, however, appear to be less per-vasive than conventional wisdom suggests.

5.2. Lean bundles and operational performance

The results suggest that implementation of each ofthe bundles of lean practices under study contributesubstantially to the operating performance of plants.Proponents of lean production frequently point out thatlean practices should be considered together as a sys-tem and that benefits accrue from all of the practices(e.g.Womack and Jones, 1996). Taking this point ofview, consider the impact of all four bundles understudy. These bundles explain about 23% of the vari-ation in operational performance after controlling forthe effects of industry and accounting for the effectsof plant size, plant age, and unionization. It is alsonotable that both size and age exert a negative influ-ence on performance when lean bundles are consid-ered. Unionization, on the other hand, has no suchnegative influence.

Each of the individual lean practice bundles alsocontributes to performance (P < 0.0001). This meansthat a separate and identifiable incremental effect canbe attributed to the four major lean practices areas thatwe have termed lean bundles.

A contribution of this research is that we identifyfour specific practice bundles from the literature, wevalidate the bundles empirically by extracting orthog-onal components consistent with the literature and weshow that all four bundles are significantly related toperformance.

Although the finding that each of the bundles con-tributes to performance may seem intuitive, such a re-sult has not been reported previously in the literature.In fact, Flynn et al. (1995)report that JIT and com-mon infrastructural practices have a positive effect onperformance but that TQM has no significant effect.On the other hand,McKone et al. (2001)find that JIT,


TQM and TPM all contribute to their weighted per-formance index.

These findings provide unambiguous evidence thatthe synergistic effects of all lean practices are asso-ciated with better manufacturing performance. Theimplication for managers of plants that are not im-plementing lean practices is also fairly clear. To notimplement lean bundles is likely to put plants at aperformance disadvantage compared to plants thatdo implement, regardless of size, age or level ofunionization of the plant in question.

5.3. Limitations and future research

This research is carried out using secondary datamade available by Penton Publishing. Although thedata present a great opportunity, they also presentsome limitations. In particular, the low response rateallows the representativeness of the sample to be calledinto question despite the large sample size. There arealso issues with self-reported performance data andpossible single respondent bias. Despite these limita-tions, the data provide a rich picture of the manufac-turing practices across many companies.

Based on the findings reported above, future re-search related to lean manufacturing clearly shouldcontrol for the effects of size and industry. The pos-itive findings with respect to the impact of contexton the implementation of lean practices suggest thatother environmental measures should also be consid-ered in future research. Specifically, the effects of en-vironmental dynamism, complexity and munificencemight be considered in future research on lean man-ufacturing in context. Separate industry level analysiswill also provide interesting insights, although leanpractices are found in plants in all industries.

Acknowledgements

The authors would like to thank three anonymousreviewers for thoughtful, constructive suggestionsthat greatly improved this paper. We also thank theCenter for Excellence in Manufacturing Management(CEMM) and Fisher College of Business, both atOhio State University, for financial support. We aregrateful to Penton Media, publishers ofIndustryWeek,for allowing access to their data.

Appendix A. Survey items

Contextual variables:How many employees are at this location?(less than 100; 100–250; 250–499; 500–999; 1000

or more)

How many years has it been since plant start-up?(less than 5 years; 5–10 years; 11–20 years; more

than 20 years)

Approximately what percent of plant productionworkers is represented by a union(s)?

(none; 1–25%; 26–50%; 51–75%; 76–99%; 100%)

Operational performance:How has thefollowing changed over the past

5 years?

Finished-product first-pass quality yield(reverse coded):

(improved more than 40%; improved 21–40%;improved 1–20%; stayed the same; declined1–20%; declined more than 20%)

Scrap and rework costs:(increased more than 20%; increased 1–20%;

stayed the same; decreased 1–20%; decreased21–40%; decreased more than 40%)

Productivity, defined as dollar volume of shipmentsper employee (reverse coded):

(increased more than 80%; increased 41–80%;increased 21–40%; increased 11–20%; increased1–20%; stayed the same; decreased 1–10%;decreased more than 10%)

Per unit manufacturing costs, excluding purchasedmaterial:

(increased more than 20%; increased 11–20%;increased 1–10%; stayed the same; decreased1–20%; decreased more than 20%)

Manufacturing cycle time:(no reduction; decreased 1–10%; decreased

11–20%; decreased 21–50%; decreased 51–75%;decreased more than 75%)

Customer lead-time:(increased more than 20%; increased 1–20%;

stayed the same; decreased 1–20%; decreased21–40%; decreased more than 40%)


References

Adam Jr., E.E., 1994. Alternative quality improvement practicesand organization performance. Journal of Operations Manage-ment 12 (1), 331–347.

Adams, S., Ponthieu, L.D., 1978. Administrative Policy andStrategy: A Casebook, 2nd ed. Grid Inc., Columbus, OH.

Ahire, S.L., Goldhar, D.Y., Waller, M.A., 1996. Developmentand validation of TQM implementation constructs. DecisionSciences 27 (1), 23–56.

Ahmed, N.U., Tunc, E.A., Montagno, R.V., 1991. A comparativestudy of US manufacturing firms at various stages of just-in-time implementation. International Journal of Production Re-search 29 (4), 787–802.

Aldrich, H., 1979. Organizations and Environments. Prentice-Hall,Englewood Cliffs, NJ.

Baldrige, V.A., Burnham, R.A., 1975. Organization innovation:individual, organizational, and environmental impacts. Adminis-trative Science Quarterly 20 (2), 165–176.

Baum, J.A.C., 1989. Liabilities of newness, adolescence, andobsolescence: the dissolution of organizational relationshipsand organizations. Proceedings of the Administrative ScienceAssociation of Canada 10 (5), 1–10.

Bicheno, J.R., 1989. Cause and Effect of JIT: A Pocket Guide.PICSIE Books, Buckingham, UK.

Boyer, K.K., Leong, G.K., Ward, P., Krajewski, L., 1997. Unlock-ing the potential of advanced manufacturing technologies.Journal of Operations Management 15 (1), 331–347.

Boyer, K.K., Pagell, M., 2000. Measurement issues in empiricalresearch: improving measures of operations strategy and advan-ced manufacturing strategy. Journal of Operations Management18 (3), 361–374.

Bronars, S.G., Deere, D.R., Tracy, J.S., 1994. The effects of unionson firm behavior: an empirical analysis using firm-level data.Industrial Relations 33 (4), 426–451.

Cappelli, P., Scherer, P.D., 1989. Spanning the union/nonunionboundary. Industrial Relations 28 (2), 188–205.

Carmines, E.C., Zeller, R.A., 1979. Reliability and Validity Assess-ment. Sage, Beverly Hills, CA.

Chan, J.S., Samson, D.A., Sohal, A.S., 1990. An integrative modelof Japanese manufacturing techniques. International Journal ofOperations and Production Management 10 (9), 37–56.

Chandler Jr., A.D., 1962. Strategy and Structure: Chapters in theHistory of the Industrial Enterprise. MIT Press, Cambridge,MA.

Child, J., 1972. Organization structure and strategies of control—areplication of the Aston Study. Administrative Science Quarterly17 (2), 163–177.

Choi, T., Eboch, K., 1998. The TQM paradox: relations amongTQM practices, plant performance, and customer satisfaction.Journal of Operations Management 17 (1), 59–75.

Cohen, J., Cohen, P., 1975. Applied Multiple Regression/Corre-lation Analysis for the Behavioral Sciences. Wiley, Hillsdale,NJ.

Cua, K.O., 2000. A theory of integrated manufacturing practicesrelating total quality management, just-in-time, and total pro-ductive maintenance. A Dissertation Thesis from University

of Minnesota. Ann Arbor, MI, UMI, Publication no. AAT9975753.

Cua, K.O., McKone, K.E., Schroeder, R.G., 2001. Relationshipsbetween implementation of TQM, JIT, and TPM and manufac-turing performance. Journal of Operations Management 19 (2),675–694.

Davy, J.A., White, R.E., Merritt, N.J., Gritzmacher, K., 1992. Aderivation of the underlying constructs of Just-In-Time manage-ment systems. Academy of Management Journal 35 (3), 653–670.

Dess, G.G., Ireland, R.D., Hitt, M.A., 1990. Industry effects andstrategic management research. Journal of Management 16 (1),7–27.

Drucker, P.F., 1987. Workers hands bound by tradition. Wall StreetJournal August 2, 18.

Finch, B.J., Cox, J.F., 1986. An examination of just-in-timemanagement for the small manufacturer with an illustration.International Journal of Production Research 24 (2), 329–342.

Flynn, B.B., Schroeder, R.G., Sakakibara, S., 1994. A frame-work for quality management research and an associatedmeasurement instrument. Journal of Operations Management11 (4), 339–366.

Flynn, B.B., Sakakibara, S., Schroeder, R.G., 1995. Relationshipbetween JIT and TQM: practices and performance. Academyof Management Journal 38 (5), 1325–1360.

Flynn, B.B., Schroeder, R.G., Flynn, E.J., 1999. World classmanufacturing: an investigation of Hayes and Wheelwright’sFoundation. Journal of Operations Management 17 (2), 249–269.

Freund, R.J., Littell, R.C., 1986. SAS System for Regression. SASInstitute Inc., Cary, NC.

Galbraith, J.R., 1977. Organization Design. Addison-Wesley,Philippines.

Germain, R., Droge, C., Daugherty, P.J., 1994. A cost and impacttypology of logistics technology and the effects of its adoptionon organizational practice. Journal of Business Logistics 15 (2),227–248.

Giffi, C., Roth, A.V., Seal, G.M., 1990. Competing in World-ClassManufacturing: America’s 21st Century Challenge. BusinessOne Irwin, Homewood, IL.

Gopalakrishnan, S., Damanpour, F., 1997. A review of innovationresearch in economics, sociology, and technology management.Omega 25 (1), 15–28.

Hackman, J.R., Wageman, R., 1995. Total quality management:empirical, conceptual, and practical issues. Administrative Sci-ence Quarterly 40 (2), 309–342.

Hall, R.W., 1987. Attaining Manufacturing Excellence: Just-in-Time, Total Quality, Total People Involvement. Dow Jones-Irwin, Homewood, IL.

Hannan, M.T., Freeman, J., 1984. Structural inertia and organi-zational change. American Sociological Review 49 (2), 149–164.

Hay, E.J., 1988. The Just-in-Time Breakthrough: Implementing theNew Manufacturing Basics. Wiley, New York.

Hayes, R.H., Wheelwright, S.C., 1984. Restoring our CompetitiveEdge: Competing Through Manufacturing. Wiley, New York.


Im, J.H., Lee, S.M., 1989. Implementation of JIT systems in USmanufacturing firms. International Journal of Operations andProduction Management 9 (1), 5–14.

Ichniowski, C., Shaw, K., Prennushi, G., 1994. The Effect ofHuman Resource Management on Productivity. Working Paper,Columbia University.

Jenkins, G.D., Taber, T.D., 1977. A Monte-Carlo study of factorsaffecting three indices of composite scale reliability. Journal ofApplied Psychology 62 (4), 292–298.

Katz, H., 1985. Shifting Gears: Changing Labor Relations in theUS Auto Industry. MIT Press, Cambridge, MA.

Koufteros, X.A., Vonderembse, M.A., Doll, W.J., 1998. Developingmeasures of time based manufacturing. Journal of OperationsManagement 16 (1), 21–41.

Lee, S.M., Ebrahimpour, M., 1984. Just-in-time production system:some requirements for implementation. International Journal ofOperations and Production Management 4 (4), 3–15.

Lissitz, R.W., Green, S.B., 1975. Effect of the number of scalepoints on reliability: a Monte Carlo approach. Journal ofApplied Psychology 60 (1), 10–13.

MacDuffie, J.P., 1995. Human resource bundles and manufacturingperformance: organizational logic and flexible production sys-tems in the world auto industry. Industrial and Labor RelationsReview 48 (2), 197–221.

Machin, S., 1995. Plant closures and unionization in Britishestablishments. British Journal of Industrial Relations 33 (1),55–68.

Mauri, A.J., Michaels, M.P., 1998. Firm and industry effects withinstrategic management: an empirical examination. StrategicManagement Journal 19 (2), 211–219.

McGahan, A.M., Porter, M.E., 1997. How much does industrymatter, really? Strategic Management Journal 18 (6/7), 15–30.

McKone, K.E., Weiss, E.N., 1999. Total productive maintenance:bridging the gap between practice and research. Production andOperations Management 7 (4), 335–351.

McKone, K.E., Schroeder, R.G., Cua, K.O., 1999. Total productivemaintenance: a contextual view. Journal of Operations Manage-ment 17 (2), 123–144.

McKone, K.E., Schroeder, R.G., Cua, K.O., 2001. The impact oftotal productive maintenance on manufacturing performance.Journal of Operations Management 19 (1), 39–58.

McLachlin, R., 1997. Management initiatives and just-in-timemanufacturing. Journal of Operations Management 15 (4), 271–292.

Meador, M.W., Walters, J.K., 1994. Unions and productivity: evi-dence from academe. Journal of Labor Research 15 (4), 373–386.

Meredith, J.R., McTavish, R., 1992. Organized manufacturing forsuperior market performance. Long Range Planning 25 (6),63–71.

Moch, M.K., 1976. Structure and organizational resource alloca-tion. Administrative Science Quarterly 21 (4), 661–674.

Monden, Y., 1981. What makes the Toyota production systemreally tick? Industrial Engineering 13 (1), 13–16.

Morrison, D.G., 1972. Regression with discrete dependent varia-bles: the effect onR2. Journal of Marketing Research 9 (3),338–340.

Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theoryof Economic Change. Harvard University Press, Cambridge,MA.

Neter, J., Wasserman, W., Kutner, M.H., 1990. Applied LinearStatistical Models, 3rd ed. Irwin, Homewood, IL.

Ng, I., Maki, D., 1994. Trade union influence on human resourcemanagement practices. Industrial Relations 33 (1), 121–135.

Osterman, P., 1994. How common is workplace transformation andwho adopts it? Industrial and Labor Relations Review 47 (2),173–188.

Pegels, C.C., 1984. The Toyota production system: lessons forAmerican management. International Journal of Operations andProduction Management 4 (1), 3–11.

Pil, F.K., MacDuffie, J.P., 1996. The adoption of high-involvementwork practices. Industrial Relations 35 (3), 423–455.

Piper, C.J., McLachlin, R.D., 1990. Just-in-time production: elevenachievable dimensions. Operations Management Review 7 (3/4),1–8.

Powell, T.C., 1995. Total quality management as competitive ad-vantage: a review and empirical study. Strategic ManagementJournal 16 (1), 15–27.

Roth, A.V., van der Velde, M., 1991. Operations as marketing: acompetitive service strategy. Journal of Operations Management10 (3), 303–328.

Sakakibara, S., Flynn, B.B., Schroeder, R.G., Morris, W.T., 1997.The impact of just-in-time manufacturing and its infrastructureon manufacturing performance. Management Science 43 (9),1246–1257.

Samson, D., Terziovski, M., 1999. The relationship between totalquality management practices and operational performance.Journal of Operations Management 17 (5), 393–409.

Schonberger, R.J., 1982. Japanese Manufacturing Techniques: NineHidden Lessons in Simplicity. Free Press, New York.

Schonberger, R.J., 1986. World Class Manufacturing: The Lessonsof Simplicity Applied. Free Press, New York.

Schroeder, R.G., Flynn, B.B., 2001. High Performance Manufac-turing: Global Perspectives. Wiley, New York.

Shingo Prize Guidelines, 1996. Richey, D., The Shingo Prize forexcellence in manufacturing. Journal for Quality and Partici-pation 19 (4), 28–31.

Skinner, W., 1969. Manufacturing—missing link in corporate stra-tegy. Harvard Business Review 50 (1), 136–145.

Smith, T.W., 1994. A comparison of two confidence intervals.GSS Methodological Report No. 80. National Opinion ResearchCenter, Chicago, IL.

Smith, T.W., Peterson, B.L., 1985. The impact of number of res-ponse categories on inter-item associations: experimental andsimulated studies. In: Proceedings of the American SociologicalAssociation Meetings, Convention Center, Washington, DC,26–30 August 1985.

Stinchcombe, A., 1965. Social structure and organizations. In:March, J. (Ed.), Handbook of Industrial Organization. RandMcNally, Chicago, IL, pp. 142–193.

Stock, G.N., Greis, N.P., Kasarda, J.D., 2000. Enterprise logisticsand supply chain structure. Journal of Operations Management18 (5), 531–547.

Sugimori, Y., Kusunoki, F., Cho, F., Uchikawa, S., 1977. Toyotaproduction system and kanban system: materialization of


just-in-time and respect for human systems. InternationalJournal of Production Research 15 (6), 553–564.

Suzaki, K., 1985. Japanese manufacturing techniques: their impor-tance to US manufacturers. Journal of Business Strategy 5 (3),10–19.

US Census of Manufacturers, 1997.http://www.census.gov/.Vickery, S.K., 1991. A theory of production competence revisited.

Decision Sciences 22 (3), 635–643.Voss, C.A., Robinson, S.J., 1987. Application of just-in-time manu-

facturing techniques in the United Kingdom. International Jour-nal of Operations and Production Management 7 (4), 46–52.

Wantuck, K.A., 1983. The Japanese approach to productivity, In:Chase, R., Aquilano, N.J. (Eds.). Production and Operations

Management: A Life Cycle Approach, 6th ed. Irwin, Home-wood, IL.

White, R.E., 1993. An empirical assessment of JIT in US manufac-turers. Production and Inventory Management Journal 34 (2),38–42.

White, R.E., Pearson, J.N., Wilson, J.R., 1999. JIT Manufacturing:a survey of implementation in small and large US manufac-turers. Management Science 45 (1), 1–15.

Womack, J.P., Jones, D.T., Roos, D., 1990. The Machine ThatChanged the World. Harper Perennial, New York.

Womack, J.P., Jones, D.T., 1996. Lean Thinking: Banish Wasteand Create Wealth in Your Corporation. Simon & Schuster,New York.

http://www.census.gov/

Documents

Lean manufacturing: context, practice bundles, and …files.ferraz8.webnode.com/200000065-e27a6e46de/Artig… · · 2013-10-04Journal of Operations Management 21 (2003) 129–149