process-technology fit and its implications for manufacturing performance

20
Journal of Operations Management 19 (2001) 521–540 Process-technology fit and its implications for manufacturing performance Ajay Das a,, Ram Narasimhan b a Department of Management, Zicklin School of Business, Baruch College, City University of New York, Box F-1813, 17 Lexington Avenue, New York, NY 10010, USA b Department of Marketing and Supply Chain Management, Eli Broad Graduate School of Business, Michigan State University, East Lansing, MI 48824, USA Received 21 September 1999; accepted 9 April 2001 Abstract This study investigates the issue of fit between process environment and advanced manufacturing technology, and its impact on manufacturing and business performance. We find that different process environments tend to align advanced manufacturing technology investments in distinct profiles, which are associated with superior performance. Deviations from these ‘ideal’ profiles are shown to have a negative impact on manufacturing performance. The findings also suggest that firms are not fully exploiting the potential afforded by AMT investments to compete in off-diagonal positions in the Hayes–Wheelwright framework. © 2001 Elsevier Science B.V. All rights reserved. JEL classification: M11 (production management) Keywords: Technology management; Operations strategy; Empirical research 1. Introduction The concept of ‘fit’ or alignment has been dis- cussed and investigated in many disciplines. The business strategy literature has examined fit between environment and strategy in different contexts and found that lack of fit has significant effects on perfor- mance (Hofer, 1975; Hambrick, 1980; Venkatraman and Prescott, 1990). The supply chain management literature has studied fit as a function of alignment be- tween sourcing practices and product life cycle (PLC) stage (Birou et al., 1998). Similarly, frameworks link- ing process type to market characteristics and PLC Corresponding author. Tel.: +1-212-802-6920; fax: +1-212-802-6873. E-mail address: ajay [email protected] (A. Das). have been developed in the operations management literature (Hayes and Wheelwright, 1979; Hill, 1994). While addressing different issues, these investiga- tions have all stressed the link between the extent of alignment achieved by a firm and its performance outcomes. Hayes and Wheelwright (1979) in their semi- nal paper, suggested that superior manufacturing performance is contingent on the degree of align- ment between the process environment and the volume–variety characteristics of the market. Differ- ent process environments would, therefore, empha- size different goals. A jobshop would likely compete on the basis of customization and flexibility, and a flow shop on delivery and cost performance. Accord- ingly, investments in machinery and personnel can be expected to differ across process environments. 0272-6963/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII:S0272-6963(01)00064-X

Upload: ajay-das

Post on 03-Jul-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Process-technology fit and its implications for manufacturing performance

Journal of Operations Management 19 (2001) 521–540

Process-technology fit and its implications formanufacturing performance

Ajay Das a,∗, Ram Narasimhan b

a Department of Management, Zicklin School of Business, Baruch College, City University of New York,Box F-1813, 17 Lexington Avenue, New York, NY 10010, USA

b Department of Marketing and Supply Chain Management, Eli Broad Graduate School of Business,Michigan State University, East Lansing, MI 48824, USA

Received 21 September 1999; accepted 9 April 2001

Abstract

This study investigates the issue of fit between process environment and advanced manufacturing technology, and its impacton manufacturing and business performance. We find that different process environments tend to align advanced manufacturingtechnology investments in distinct profiles, which are associated with superior performance. Deviations from these ‘ideal’profiles are shown to have a negative impact on manufacturing performance. The findings also suggest that firms are notfully exploiting the potential afforded by AMT investments to compete in off-diagonal positions in the Hayes–Wheelwrightframework. © 2001 Elsevier Science B.V. All rights reserved.

JEL classification: M11 (production management)

Keywords: Technology management; Operations strategy; Empirical research

1. Introduction

The concept of ‘fit’ or alignment has been dis-cussed and investigated in many disciplines. Thebusiness strategy literature has examined fit betweenenvironment and strategy in different contexts andfound that lack of fit has significant effects on perfor-mance (Hofer, 1975; Hambrick, 1980; Venkatramanand Prescott, 1990). The supply chain managementliterature has studied fit as a function of alignment be-tween sourcing practices and product life cycle (PLC)stage (Birou et al., 1998). Similarly, frameworks link-ing process type to market characteristics and PLC

∗ Corresponding author. Tel.: +1-212-802-6920;fax: +1-212-802-6873.E-mail address: ajay [email protected] (A. Das).

have been developed in the operations managementliterature (Hayes and Wheelwright, 1979; Hill, 1994).While addressing different issues, these investiga-tions have all stressed the link between the extentof alignment achieved by a firm and its performanceoutcomes.

Hayes and Wheelwright (1979) in their semi-nal paper, suggested that superior manufacturingperformance is contingent on the degree of align-ment between the process environment and thevolume–variety characteristics of the market. Differ-ent process environments would, therefore, empha-size different goals. A jobshop would likely competeon the basis of customization and flexibility, and aflow shop on delivery and cost performance. Accord-ingly, investments in machinery and personnel canbe expected to differ across process environments.

0272-6963/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S0272 -6963 (01 )00064 -X

Page 2: Process-technology fit and its implications for manufacturing performance

522 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

As is well known, jobshops utilize general-purposemachines and multi-skilled workforce, and assemblylines use dedicated equipment with less-skilled la-bor. Hayes and Wheelwright (1979) argue that firmsnot aligned in terms of market, process technologyand labor choices, i.e. firms not ‘on the diagonal’would suffer performance setbacks. However, recentadvances in technology have enabled firms to oper-ate in off-diagonal positions without the performancepenalties predicted by the Hayes and Wheelwright(1979) framework. These developments have height-ened research interest in the role and use of advancedmanufacturing technologies and practices, in differentprocess environments. Safizadeh et al. (1996) foundthat companies in line flow industries are success-fully operating in “off-diagonal” positions in Hayesand Wheelwright’s (1979) framework, meeting cus-tomization requirements by deploying flexible manu-facturing systems and practices. Similarly, there areexamples of firms using cellular manufacturing lay-outs with small production lots in a jobshop environ-ment to achieve efficiencies comparable to line-flowoperations (Markland et al., 1998). Computer-aideddesign and manufacturing techniques are being em-ployed to reduce product development and productioncycle times.

A growing body of evidence thus suggests thatoff-diagonal positions in Hayes and Wheelwright’s(1979) framework are indeed viable through the useof advanced manufacturing technologies and prac-tices. However, these studies have not attempted todevelop a prescriptive model to address the question:are there ideal configurations of advanced manufac-turing technology investments that result in superiormanufacturing performance for a given process en-vironment? The answer to this question will helpfirms in two ways: enable them to evaluate and ad-dress gaps in their advanced technology investmentpatterns vis-à-vis an ‘ideal’ configuration; and deter-mine whether advanced technology would facilitatea move to an “off-diagonal” position, to pursue whatmight be non-traditional competitive goals for theirprocess environment. This paper attempts to developan answer to this central research question of strategicimportance.

The next section discusses past research on man-ufacturing process and manufacturing technology.Based on the literature, a conceptual model is devel-

oped leading to a set of propositions related to thecentral research question of interest. The succeedingsections describe scale development, data collectionand analysis, and statistical testing of the proposi-tions. The paper concludes with a discussion of thefindings and their implications for manufacturingstrategy theory and practice.

2. Overview of the literature

Choosing a manufacturing process type is a strate-gic decision in operations management, with atten-dant implications for performance. The literature onprocess choice presents a strong rationale for align-ing process type with PLC stage and product/marketcharacteristics (Hayes and Wheelwright, 1979; Hill,1994). High variety, low volume market environmentswould demand a process capable of small produc-tion run sizes and high set-up frequencies, with rel-atively low set-up costs. A jobshop would addressthese needs allowing a firm to compete on flexibility,customization and new product development capabil-ities. Low variety, high volume market environmentswould require extended production runs, and usuallyentail relatively high set-up costs. An assembly linebest fits this environment, where cost and quality areprimary strategic capabilities. Empirical evidence inthe literature supports these perspectives. Ward et al.(1992) found that “typical producers” align processchoice decisions with PLC stages. Safizadeh et al.(1996) allude to the strategic importance of align-ment between process choice and a firm’s competitivepriority.

Firms pursue competitive goals by adoptingwhat they deem to be appropriate technology ini-tiatives. New technologies in the form of pro-grammable machines, flexible manufacturing systems,computer-aided design (CAD), computer-aided engi-neering (CAE), and computer-aided manufacturing(CAM) practice innovations such as worker teams andcellular manufacturing have allowed high volume pro-ducers to offer more variety, and created opportunitiesfor efficient small lot manufacturing for these produc-ers. Advanced manufacturing technologies have alsobeen adopted by jobshops to reduce costs. Computernumerical controlled (CNC) machines, CAD, CAM,and operator teams are widely used across different

Page 3: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 523

manufacturing environments (Markland et al., 1998;Boyer et al., 1996). Given that different process en-vironments have different manufacturing goals, it isreasonable to argue that some differences can be ex-pected in the way that different process environmentswould configure their technology-based initiatives.For example, cellular manufacturing may be morecompatible with a low volume jobshop environment,than a repetitive assembly line situation. In contrast,set-up time reduction programs are more likely to befound in assembly line systems. Similarly, the use ofmachine robots would be more common in assem-bly line settings than in jobshops. Past research hasexamined technology investment preferences amongdifferent firms (Boyer et al., 1996). What has notbeen investigated is how advanced manufacturingtechnologies can be “best” deployed in different pro-cess environments. Do ‘ideal’ profiles of technologyinitiatives exist for different process environments?This research investigates this question across twodifferent process environments — jobshop and as-sembly lines — representing contrasting points in thevolume–variety continuum.

3. Conceptual framework

Fig. 1 shows the conceptualized relationshipsamong process environment, technology-based ini-tiatives, and manufacturing performance. The frame-work shows advanced manufacturing technologyinitiatives as a multi-dimensional construct, encom-passing initiatives in manufacturing design, manufac-turing technology, manufacturing infrastructure, andhuman resource management practices. The researchproposition is that jobshop and assembly environ-ments will configure resource investments in thesetechnology dimensions in different patterns, to maxi-mize manufacturing performance. Alignment or lackthereof, between process environment and technologyinvestments is hypothesized to be a determinant ofmanufacturing performance. By implication, it canbe argued that the technology investments of thosefirms that achieve superior manufacturing perfor-mance constitute an “ideal profile”, in the sense thatthese investments exhibit a high degree of fit with theprocess environment. This study examines two keypropositions underlying the framework.

• Different process environments tend to align ad-vanced manufacturing technology investments inidentifiable ‘ideal’ profiles.

• That deviations from these ‘ideal’ profiles have anegative impact on manufacturing performance.

In regard to the first proposition, little research ex-ists on process-based differences in the adoption ofadvanced manufacturing technologies and practices.Interestingly, the few studies that have touched onsuch differences emerged with contending perspec-tives. McDermott et al. (1997) found that both mainmarket as well as niche firms pursued advanced man-ufacturing technologies and practices, such as invest-ments in CNC equipment and work cell organization.They concluded that there was essentially no differ-ence in the rate and scope of adoption of advancedmanufacturing technologies and practices across dif-ferent operational environments. In a sense, jobshopsand assembly line processes aspired to mimic eachother’s unique capabilities, while maintaining theirown process-based advantages. Jobshops invest inCNCs and re-organize work-flow processes to achievesome of the efficiencies of large-scale production,while line processes do the same in order to reducechangeover costs and increase responsiveness (Mc-Dermott et al., 1997; Safizadeh et al., 1996). As such,one would not expect to see substantive differences be-tween jobshop and assembly line processes, in regardto the types of advanced manufacturing technologiesand practices adopted. The distinction would lie inthe different nature of the goals pursued by differentprocesses, and not the state of use of advanced man-ufacturing technology or practice, per se. However,Swamidass and Kotha (1998) report otherwise. In astudy of 160 manufacturing firms, they distinguishedbetween “high volume automation technology” and“low volume automation technology”. “Low volumeautomation technology” comprised of CNC ma-chines, programmable controllers, CAD/CAM andFMS, technologies that facilitate flexibility and re-sponsiveness. Such technologies would be compatiblewith a jobshop process environment, characterized bylow volumes and customization responsiveness. Anassembly line would use “high volume automationtechnology”, comprised of computer-aided inspectionand testing, robotics and manufacturing automationprotocol. Swamidass and Kotha (1998) extended their

Page 4: Process-technology fit and its implications for manufacturing performance

524 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

Fig. 1. Process environments and manufacturing technologies/practices.

investigation to examine the use of advanced technol-ogy in high performers in different strategic segments(Kotha and Swamidass, 2000). They reported positiveassociations between cost leaders’ performances andthe use of design and low volume technologies, and,between differentiators’ performances and the use ofdesign, information, low volume and high volumetechnologies. These latter findings are striking in thatthey reinforce the conclusions of McDermott et al.(1997). The premise is that cost leaders use low vol-

ume technologies to reduce the transition penaltiesin acquiring jobshop like flexibilities, while jobshopsemploy high volume technologies to pursue line flowefficiencies. Albeit, as noted by Kotha and Swami-dass (2000), differentiators with low volume, highflexibility operational demands (i.e. jobshops) profitfrom a broader spectrum of advanced manufactur-ing technologies and practices, relative to companiesthat seek cost leadership through volume economies(i.e. assembly lines). It bears noting that the use of

Page 5: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 525

design and low volume technologies are prevalentin both process environments. This study conceptu-alizes ‘ideal profiles’ of technology investments aspatterns of advanced manufacturing technologies andpractices associated with high performing firms indifferent processes.

Based on the above discussion, it was anticipatedthat the advanced manufacturing technologies andpractices employed in high performing jobshops andassembly lines would not differ radically. The ex-ceptions would be for individual variables such asset-up time reduction practices and robotics that canbe expected to be more useful for assembly lines,and cellular manufacturing that can re-structure thejumbled work flow of a jobshop into more regularpatterns. Accordingly, an a priori ideal profile of ad-vanced manufacturing technologies and practices for

Table 1Envisaged AMT items in ‘ideal’ profiles

Item measure Technologies/practices in ideal profile Used in

Jobshops Assembly line

Manufacturing technology (MT) Use of CNC technology Yes (Y)a Yes (Y)Use of computer-aided manufacturing Yes (Y) Yes (Y)Use of flexible manufacturing systems No (Y) Yes (Y)Use of set-up time reduction methods No (N)a Yes (Y)Use of robotics No (N) Yes (N)

Human resources management (HRM) Worker cross-training Yes (Y) Yes (N)Use of operator teams Yes (Y) Yes (Y)Decentralized decision making for production schedluing Yes (Y) Yes (Y)Decentralized decision making for operator tasks Yes (Y) Yes (Y)

Manufacturing design (MD) Use of computer-aided engineering Yes (Y) Yesb

Use of computer-aided design Yes (Y) YesUse of modularization in product design Yes (Y) YesUse of computer-aided testing Yes (N) Yes

Manufacturing infrastructure (MI) Use of Kanban systems Nob YesUse of automated material handling systems No YesUse of in-plant EDI systems Yes YesUse of bar coding Yes YesUse of real-time process controls Yes YesUse of preventive maintenance Yes YesJIT supplier deliveries No YesCellular manufacturing/group technology Yes NoSafety stock only for unique components Yes Yes

a ‘Y’ indicates that the item found inclusion in the final ideal profile developed from the sample data. ‘N’ indicates the item did notfind inclusion in the final post-analysis ideal profile.

b Manufacturing design was not significantly associated with performance for assembly lines. Hence, following the prescribed method-ology, items from this particular AMT dimension were not included while developing the ideal profile for assembly lines. Similarly,manufacturing infrastructure was not significantly associated with performance in either jobshops or assembly lines, and items for thisAMT dimension were excluded from the ideal profiles.

best performers in both jobshops and assembly lineswas developed for the study (Table 1).

There are two observations that need to be madewith respect to Table 1. By definition, ‘ideal profiles’should be comprised of only those technologies andpractices that impact performance. However, it wasnot possible to categorically identify those technolo-gies/practices that impact performance in advance, forthe sample of firms in this study. Therefore, all fourAMT dimensions, manufacturing technology, man-ufacturing design, manufacturing infrastructure andhuman resources management were considered as po-tential influences on performance, and items relatingto each dimension were included in the envisaged,a priori ideal profile. Second, process-contingentdifferences in the intensity of use of each manufac-turing technology or practice were not hypothesized

Page 6: Process-technology fit and its implications for manufacturing performance

526 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

a priori-the primary research objective was to exploreif patterns of AMT usage exist-since the extent ofinter-pattern differences across process environments,was of secondary interest. Generally speaking andbased on previous studies (Kotha and Swamidass,2000), it was expected that jobshops may use theadvanced manufacturing technologies/practices listedin the envisaged ideal profiles to a greater extent thanassembly lines, for performance gains.

This research differs from previous studies in threedistinct ways. First, this research examines tech-nology initiatives in a broad context, employing aholistic configuration approach that seeks to iden-tify patterns of technology investments related toperformance. This is in sharp contrast to the tradi-tional examination of bi-variate interaction effects onperformance. The product–process matrix suggestsspecific machine technology types for specific pro-cess environments (Hayes and Wheelwright, 1979;Markland et al., 1998). We explore the use of ad-vanced manufacturing technologies across differentprocess environments. Second, this study examinesthe alignment between process environment and tech-nology investments, whereas previous research hasfocused on process-market congruence issues (Wardet al., 1992; Safizadeh et al., 1996). A final differencepertains to a methodological issue. Previous inves-tigations have employed analytical methods such ascluster analysis to identify technology investmentpatterns. Cluster analysis and similar categorizationtechniques, however, do not allow any calibration ofthe degree of differences among or within differentgroups. For instance, Boyer et al. (1996) use clusteranalysis to group firms in terms of differences in theirmean scores on the advanced manufacturing tech-nology dimensions: “design”, “manufacturing” and“administrative”. Although this approach is usefulfor identifying clusters of technology investments, itdoes not provide a composite measure of the degreeof inter-group difference. This study employs a com-posite measure of inter-group differences and relatesthat to manufacturing performance.

4. Methodology

This section describes the development of con-structs, measurement scales and the sampling proce-dure followed in the study.

4.1. Sampling

Data for this analysis was collected from purchas-ing executives, as part of a larger study that examinedthe interface between purchasing and manufacturing.The National Association of Purchasing Management(NAPM)’s Title 1 list of senior members from manu-facturing industries was consulted to prepare the sam-ple frame, since high-ranking respondents tend to bemore reliable sources of information than their subor-dinate ranks (Philips, 1981). The SIC codes includedin the sample were as follows:

SIC 34 fabricated metal products, exceptmachinery and transportation equipment

SIC 35 industrial and commercial machineryand computing equipment

SIC 36 electronic and other electrical equipmentand components

SIC 37 transportation machinery and itemsSIC 38 measuring, analyzing and controlling

instruments, photographic, medical andoptical goods

These SICs are established users of AMT in dis-crete product manufacturing (Snell and Dean, 1992;US Department of Commerce, 1988), represent con-trasting process environments, and together with SIC39, comprise over 40% of the sales of the US manu-facturing sector (US Department of Commerce, 1988).The unit of analysis was the plant.

Data collection was conducted in two phases. Phase1 involved preliminary site visits, interviews andsurveying of executive management in the sourcingand manufacturing areas across different industries.Pilot-tests of the instrument and initial interviewswith purchasing and manufacturing executives (atthe same location) did not reveal any significant dif-ferences between responses. The item measures didnot require detailed technical knowledge of equip-ment or practices. It was found that senior materialsmanagement were relatively conversant with man-ufacturing demands and situations. As a post hoctest of inter-rater reliability, manufacturing managersfrom a random sub-sample of 20 responding firmswere contacted to obtain responses to the manufac-turing related items in the questionnaire. Telephoneinterviews were conducted with seven responding

Page 7: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 527

manufacturing managers. Paired comparisons ofscores were made between the purchasing and man-ufacturing respondents for the manufacturing relatedquestionnaire items, for each of the seven firms. Theaverage inter-rater reliability was 0.96, evidencing ahigh degree of agreement between purchasing andmanufacturing respondents on manufacturing relatedissues (James et al., 1984). Further, respondents wereasked to confirm whether they had consulted withmanufacturing in responding to manufacturing re-lated questions — 38% of the respondents replied inthe affirmative. Phase 2 of the data collection pro-cess, following Dillman’s (1978) guidelines, involvedmailing the survey to senior level NAPM membersselected at random from the NAPM member list.The mailing package consisted of a cover letter, thesurvey and a reply paid return envelope. Assuminga conservative 15–20% response rate, the mailingwas addressed to around 1700 potential respondents.A great deal of time and effort was invested in thedata collection. A multi-faceted strategy was imple-mented, that included an explanation to potential re-spondents of the adverse consequences of incompleteresponses, three follow-up mailings (two with sur-veys attached and one post card reminder), and phonecalls and faxes. A reminder post card was mailed toall non-respondents after a week of mailing. Writ-ten follow-ups (with duplicate questionnaires) weremailed to all non-respondents approximately 3 weeksafter the initial mailing. Respondents were promised abenchmarking report of their operating characteristicsas an incentive to participate.

Two mailings of the survey instrument resulted ina total of 322 responses, excluding returns, refusalsand unusable responses. The response rate of 19%compares well with past studies (Gupta and Somers,1996). Table 2 shows the respondent profile. Tests fornon-response bias across the first (n = 200) and sec-ond (n = 122) wave of respondents did not show anysignificant results. An ANOVA test did not reveal anystatistically significant differences among the categorymeans for company sales, plant sales or number ofemployees across the different SIC groups. The dataappeared to be biased towards larger plants, as evidentfrom the comparisons with census data. One explana-tion may be that respondents were asked to focus onthe key plant in their company in their responses. Inview of the possible bias, caution is urged in general-

izing the results of the study to smaller organizationsin the manufacturing sector.

4.2. Measures

Respondents were required to identify the produc-tion processes that accounted for the most time in themanufacture of their major product(s), and categorizethat as a jobshop or assembly process environment.Pilot tests and interviews had confirmed that managershad an adequate understanding of the volume and va-riety connotations of jobshop and assembly processes.Further 82% of the responding plants from the posthoc inter-rater reliability sample had identical clas-sifications by purchasing and manufacturing respon-dents on the primary type of manufacturing processemployed in their plants. Measures were developedfor the remaining constructs, based on the literature.The responses were measured on a 1–5-Likert scale,consistent with past practice in many studies (Swami-dass and Kotha, 1998; Swamidass and Newell, 1987;Venkatraman, 1990).

Past research has described advanced manufactur-ing technology as a multi-dimensional construct thatincludes the use of ‘hard’ machine related aspects:robotics, CNC hardware, CAD/CAM, etc.; and ‘soft’practice/process related aspects: Kanban, set-up timereduction techniques, JIT, etc. (Burgess and Gules,1998; Clark, 1996; Dean and Snell, 1996; Roth andGiffi, 1995; McCutcheon et al., 1994; Tranfield et al.,1991). The construct definitions provided by Tranfieldet al. (1991) and Boyer et al. (1996) were of interestto this research because of their breadth of scope,transcending automated machine typology to enfoldCAD/CAM, information technology, configurationalprocesses, use of JIT/Kanban, and administrativeprocesses in their description of an advanced man-ufacturing system. Boyer et al. (1996) empiricallyderived three separate dimensions of AMT: design,manufacturing and administrative. Design includeda mix of design and process technologies such asCAD, CAE, CAM, computer-aided process planning,and the use of CNC equipment. Manufacturing cov-ered technology elements such as FMS, real timeprocess control systems and robotics, while the ad-ministrative dimension included MRPII, EDI, andknowledge and decision support systems. Swamidassand Kotha’s (1998) dimensionalization of the AMT

Page 8: Process-technology fit and its implications for manufacturing performance

528 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

Table 2Sample profile of respondents

Number of valid 322responses

Respondent titles VP’s/directorpurchasing/materials

50

Purchasing/commodity/materials managers

179

Senior buyers/buyers 7Other titles(operations manager,purchasing engineer)

6

No response 80

Total no. of plants with 240identifiable SIC codes

SIC Plant sales number ≤US$ 1million

US$ 1–10million

US$10–50million

US$50–100million

US$100–500million

US$>500million

Plant sales distributiona

34 No. of plants 0 3 16 4 4 235 No. of plants 0 8 36 30 23 536 No. of plants 0 6 14 10 18 537 No. of plants 0 1 14 10 5 338 No. of plants 0 3 8 4 6 2

Total 0 21 88 58 56 17

Average no.of employees(data)

Average no.of employees(census)

No. of employees distributionb

34 334 (range 30–2000) 4135 430 (range 30–1700) 3536 400 (range 31–2000) 9337 464 (range 30–2000) 12638 504 (range 20–2000) 67

Processcharacteristics

Jobshop Batch Repetitive Continuous

No. of respondents(total sample)c

109 71 117 24

a Average plant sales as manually estimated from US census 1997 — SIC 34: US$ 6.10 million; SIC 35: US$ 7.25 million; SIC36: US$ 20.38 million; SIC 37: US$ 41.65 million; SIC 38: US$ 11.96 million. Census figures are estimates and may not be strictlycomparable with sample data; www.census.gov/epcd.

b Outliers eliminated — in excess of 90% of responding plants in these SIC classifications reported no. of employees ≤2000.c Figures do not include missing responses.

construct was not very different from Boyer et al.’s(1996) classification. They (the former) developedfour dimensions: information exchange and planningtechnologies (MRPII, EDI, etc.), and production de-sign technology (CAD/CAE), and distinguished be-tween high-volume automation technology (robotics,

manufacturing automation, computer-aided inspec-tion, etc.) and low-volume automation technology(CNC, CAD. CAM, FMS). A careful examinationof these conceptualizations reveals three clear AMTdomains: a design domain that is concerned largelywith design technologies, a manufacturing domain

Page 9: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 529

that involves mainly process related technologies,and an infrastructural domain that comprises infor-mation and production planning and control techno-logies.

One key aspect of successful AMT systems thatresearchers have neglected to integrate under therubric of advanced manufacturing technologies andpractices is the human element. It is now widelyrecognized that the benefits of hard investments inadvanced technology are largely moderated by con-current investments in soft technologies and practices(Cagliano and Spina, 2000; McDermott and Stock,1999; Parthasarthy and Sethi, 1992). Highly skilledand trained workers are believed to be an essentialattribute of successful AMT systems, because ofthe increasing complexity of the technology and theplant-wide integration to which such technologiesoften lead (Upton and McAfee, 1988; Cohen andApte, 1997; Adler, 1988). Increased complexity isalso associated with increased operational uncertain-ties, requiring an organic structure that allows morefront-line decisions (Burns and Stalker, 1961). Infact, Ward et al. (1994) developed an infrastructuralfactor, involving worker training, empowerment, andjob-enrichment. Based on the above discussion andrecognizing the inseparable role of human resourcemanagement practices in advanced manufacturingtechnology implementation (Snell and Dean, 1992),advanced manufacturing technology was defined asthe collective use of advanced manufacturing, design,infrastructural, and human resource managementpractices and systems in a plant. Specifically, theconstruct encompassed the following.

• Manufacturing technology (MT): The use of ad-vanced manufacturing systems — CNC machines.CAM, and flexible manufacturing systems.

• Manufacturing design (MD): The use of computer-aided design, engineering and testing.

• Manufacturing infrastructure (MI): The use of in-frastructural support systems — JIT manufacturing,automated material handling, Kanban, minimal in-ventories, preventive equipment maintenance, ac-celerated die changes and set-ups, parts bar coding,and EDI usage in manufacturing.

• Human resource management (HRM): The use ofinnovative human resource management practicesand structures in manufacturing — cross-functional

teams, decentralized decision-making and workercross training.

Specific item measures for the AMT construct wereadapted from the technology scales developed bySwamidass and Kotha (1998), Ward et al. (1994) andBoyer et al. (1996). Table 3 shows the list of initialitem measures.

A Bartlett Chi-square test resulted in statisticallysignificant Chi-squares for both jobshop and assemblyline data matrices, thus permitting factor analysis onthe data. There were between 3 and 8 missing valueson 16 variables in the jobshop data and between 1 and7 missing values on 17 variables in the assembly linedata, out of the total of 40 variables involved in theanalysis. These missing values were replaced by themean values of the relevant variables, which is conven-tional when the number of missing cases is not exces-sive. There were 117 and 109 cases for the assemblyline and jobshop process environments, respectively.

Table 3List of initial item measures for AMT

Likert scale: 1 (very low) to 5 (very high)

Manufacturing technologyThe use of flexible integrated manufacturing systemsThe use of CNC technologyThe use of computer-aided manufacturingThe use of robotics

Manufacturing designThe use of computer-aided designThe use of computer-aided engineeringThe use of computer-aided testingModularization in design

Manufacturing infrastructureThe use of automated material handling systemsThe use of Kanban/similar manufacturing practicesThe use of in-plant electronic data interchange systemsThe use of real-time process controlsThe use of bar-codingThe use of set-up time reduction techniquesThe use of preventive maintenanceJIT supplier deliveriesCellular manufacturing organizationsGroup technologyMaintaining safety-stock only for unique components

Human resources managementThe use of cross-trained employeesThe use of teams in manufacturingThe use of decentralized decision-making in manufacturing

Page 10: Process-technology fit and its implications for manufacturing performance

530 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

A confirmatory factor analysis for advanced manu-facturing technology factors was performed on bothjobshop and assembly line process data. The maxi-mum number of variables in the CFAs was 23 — thesample sizes of 117 and 109 were deemed adequate

Table 4Factor analysis results for the AMT constructa

Factor Item Loading Loadings(varimax) (promax)

JobshopManufacturing technology (MT) (α = 0.819) Use of CNC technology 0.770 0.678

Use of flexible manufacturing systems 0.742 0.775Use of computer-aided manufacturing (CAM) 0.807 0.824

Manufacturing design (MD) (α = 0.687) Use of computer-aided design (CAD) 0.763 0.753Use of computer-aided engineering (CAE) 0.762 0.825Modularization in designb 0.598 0.443

Manufacturing infrastructure (MI) (α = 0.747) Use of bar coding 0.798 0.670Use of EDI 0.559 0.604Use of material handling equipment 0.576 0.608Use of cellular manufacturingb 0.534 0.575

Human resources management (HRM)(α = 0.820)

Decentralized decisionmaking for distributingoperator tasks for the day

0.885 0.848

Decentralized decision making(operator teams/individuals) for(micro) production scheduling

0.875 0.859

Operator teams in manufacturing 0.658 0.630Cross-trained employeesb 0.616 0.622

Assembly lineManufacturing technology (α = 0.819) Use of CNC technology 0.870 0.844

Use of flexible manufacturing systems 0.832 0.840Use of computer-aided manufacturing 0.544 0.578Use of set-up time reduction techniquesc 0.624 0.682

Manufacturing design (α = 0.739) Use of computer-aided design 0.858 0.851Use of computer-aided engineering 0.786 0.739Use of computer-aided testingc 0.562 0.561

Manufacturing infrastructure (α = 0.730) Use of bar coding 0.694 0.666Use of EDI 0.639 0.551Use of material handling equipment 0.814 0.746Use of roboticsc 0.648 0.599

Human resources management (α = 0.824) Decentralized decision makingfor distributing operator tasksfor the day

0.780 0.721

Decentralized decision making(operator teams/individuals) for(micro) production scheduling

0.824 0.799

Operator teams in manufacturing 0.718 0.801

a Cronbach’s α-values shown in parenthesis.b Items specific to jobshop process.c Items specific to assembly line process.

vis-à-vis the recommended guideline of 5–10 subjectsper variable (Tinsley and Tinslay, 1987). Initial com-munalities were set at 1.00, and a varimax rotationwith Kaiser normalization was performed on all fac-tors. As an additional step, a promax oblique rotation

Page 11: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 531

Table 5Factor analysis results for composite manufacturing performance (α = 0.872)a

Factors Items Loading

Manufacturing cost reduction (α = 0.783) Relative to internal goals 0.903Relative to primary competition 0.778

Quality performance (α = 0.825) No. of defects/product reduction relative 0.857to internal goalsNo. of defects/product reduction relative to relative to 0.895primary competition

New product introduction time reduction Relative to internal goals 0.865performance (α = 0.790)

Relative to primary competition 0.893

Delivery performance (α = 0.897) Delivery speed relative to internal goals 0.839Delivery speed relative to primary competition 0.754Delivery dependability relative to internal goals 0.871Delivery dependability relative to primary competition 0.819

Customization responsiveness performance Meeting customization requests relative to internal goals 0.821(α = 0.821)

Meeting customization requests relative 0.876to relative to primary competition

a The extent to which the plant has been able to meet its performance goals in.

procedure was performed on the data, using principalaxis factor extraction. Tinsley and Tinslay (1987) statethat an oblique rotation procedure (promax methodrecommended) will produce a solution comparable toa orthogonal rotation procedure, if the underlying fac-tor dimensions are essentially orthogonal. The resultsof the promax rotated analysis were found to be sim-ilar to those obtained using the varimax rotated anal-ysis (see Table 4).

Items loading on the AMT dimensions — MT, MD,MI and HRM — are shown in Table 3. Reliabilities(Cronbach’s α) for the scales exceeded the recom-mended 0.70 cut-off point (Nunnally, 1978; Churchill,1979).

As Table 4 shows, the factor loadings were essen-tially similar across both jobshop and assembly lineenvironments, with a few exceptions. Items specific tojobshops were ‘modularization in design’, ‘use of cel-lular manufacturing’ and ‘cross-trained employees’.Items specific to assembly lines were ‘use of set-uptime reduction techniques’, ‘use of computer-aidedtesting’ and ‘use of robotics’. The remaining itemsloaded on the same factors for both process types. Totest the conceptual validity of the AMT construct, asecond order confirmatory factor model was developedusing the first order factors. All the first order factor

loadings in the second order analysis were above 0.70and the composite reliability of the AMT constructwas well above 0.70.

Manufacturing performance goals have traditionallyincluded product quality, manufacturing cost, delivery,flexibility and innovation (Krajewski and Ritzman,1999). This research measured manufacturing per-formance along five dimensions: manufacturing costreduction, quality improvement (conformance), deliv-ery speed and reliability, customization responsive-ness (flexibility), and new product introduction time.Item measures were adapted from existing scales formanufacturing performance (Miller and Roth, 1994;Dean and Snell, 1996). The item measures requiredrespondents to evaluate their firm’s manufacturingperformance against internal goals and competitorperformance. A factor analysis was performed re-sulting in five distinct factors corresponding to theconceptualized performance dimensions (Table 5).

Convergent validity for the all the CFAs was de-termined by the strong and significant (0.534–0.903;P < 0.5) item loadings (see Tables 4 and 5). 1 All

1 The means and variances of the variables used in the factoranalysis, associated correlation matrices, and the initial and finalcommunalities are available from the corresponding author.

Page 12: Process-technology fit and its implications for manufacturing performance

532 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

inter-factor correlations were found significantly dif-ferent from 1.00, and the average variance extractedfor each factor in a CFA model was found greaterthan the squared correlation between that factor andany other factor in that model, providing a rigor-ous verification of discriminant validity (Fornell andLarcker, 1981). Content validity was ensured by theadaptation of item measures from previous scalesin the literature and through pre-tests of the instru-ment.

5. Data analysis

The data analysis procedure was based on the busi-ness strategy literature on fit and alignment (Drazinand Van de Ven, 1985; Venkatraman and Prescott,1990). The sequence of analysis for each process en-vironment was as follows:

1. identify the strategic manufacturing capabilities;2. identify specific AMT domains that influence these

capabilities;3. develop ‘ideal’ profiles of AMT resource alloca-

tions;4. develop a measure of deviation from these ‘ideal’

profiles;5. relate the measure of deviation to manufacturing

performance.

First, each manufacturing performance dimensionwas related to firm level profit, sales growth, ROAand market growth performance. Earlier studies havelinked site-based factors such as buyer–supplier re-lationship, manufacturing process improvement, andproduction competence with broader firm-level per-formance (Carr and Pearson, 1999; Samson andTerziovski, 1999; Choe et al., 1997). Firm perfor-mance along the financial and market dimensions wasassessed through self-reported measures that askedrespondents to indicate their company’s performancerelative to competition, on a ‘high–low’ Likert scale(Venkatraman, 1990). Past research indicates thatmanagerial evaluations correspond closely to objec-tive data obtained from both internal and externalsources (Dess and Robinson, 1984; Venkatraman andRamanujam, 1987). The majority of respondents be-longed to companies for which relevant published

financial or market data could not be found. However,the profit measures reported by respondents in 25 pub-licly traded companies were compared with objectivefinancial data (E∗Trade Inc., 2000). Approximately80% of the self-reported measures were consistentwith the objective data. Significant correlations iden-tified those manufacturing performance dimensionsthat related to firm level performance. Table 6 showsthe results of the correlation analysis.

Customization responsiveness and new productintroduction time reduction were found to have astatistically significant association with market shareand sales growth, respectively for jobshops. Qual-ity of conformance emerged as a strategic priorityfor assembly line environments. Manufacturing costreduction related to three firm level performance out-comes, for both process environments. Given theseprocess-specific strategic manufacturing capabili-ties, the next step was to identify AMT domains ofstrategic value in different process environments. Aregression analysis was done using the four AMT do-mains as independent variables, and process-specificstrategic manufacturing capabilities as the dependentvariables. Tables 7 and 8 report the results of theregression analysis.

The results indicated that the AMT domains ofhuman resources management and manufacturingtechnology significantly influence several strategicmanufacturing capabilities, in both jobshop and as-sembly line processes. Manufacturing design wasfound to influence manufacturing customization ca-pabilities. Manufacturing infrastructure did not havea significant relationship with any manufacturingpriority in either type of process.

Next, ‘ideal’ profiles of AMT resource allocationswere developed for each process environment, fol-lowing Venkatraman and Prescott’s (1990) procedure.For this purpose, a composite measure of manufac-turing performance was developed for each processenvironment, aggregating scores on significant manu-facturing performance dimensions. Past studies havedeveloped and used similar composite measures ofperformance (Ward et al., 1994). To illustrate, a com-posite manufacturing performance measure for a firmoperating as a jobshop was computed by taking themean of its summed scores on manufacturing costreduction, customization responsiveness and newproduct introduction time reduction — manufacturing

Page 13: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 533

Table 6Significant correlations between manufacturing performance and firm performancea

Net profits Sales growth ROA Market share

Jobshop process (n = 109)Manufacturing cost reduction 0.2736∗∗ 0.2868∗∗ 0.2189∗∗ n.s.b

Customization responsiveness n.s. n.s. n.s. 0.2125∗∗New product introduction time n.s. 0.2195∗ n.s. n.s.

Assembly line process (n = 117)Quality 0.2298∗ n.s. n.s. n.s.Manufacturing cost reduction n.s. 0.2290∗ 0.2199∗ 0.1977∗

a All financial, market and manufacturing performance measures are on a 1 (very low) to 5 (very high) Likert scale, and are relativeto primary competition.

b n.s.: not significant at P < 0.05.∗ P < 0.05.∗∗ P < 0.01.

performance dimensions significantly related to firmperformance for jobshop environments. Firms in eachprocess type were then ranked on their respectivecomposite manufacturing performance score, and thetop 10% companies identified. The mean score ofthis group of high performing companies on each

Table 7Regression runs-AMT domains vs. strategic manufacturing capabilities jobshop process (n = 109)

Independent variables B S.E. B β P

Dependent variable: manufacturing cost reduction performancea

Manufacturing technology 0.199 0.070 0.314 0.006Manufacturing design (−)0.073 0.092 (−)0.083 0.428Manufacturing infrastructure 0.007 0.102 0.008 0.945Human resources management 0.207 0.082 0.260 0.013

Dependent variable: customization responsiveness performanceb

Manufacturing technology (−)0.116 0.082 (−)0.154 0.157Manufacturing design 0.353 0.107 0.339 0.001Manufacturing infrastructure 0.147 0.119 0.138 0.220Human resources management 0.106 0.095 0.112 0.263

Dependent variable: new product introduction time reduction performancec

Manufacturing technology (−)0.128 0.077 (−)0.177 0.102Manufacturing design 0.160 0.102 0.160 0.119Manufacturing infrastructure 0.113 0.113 0.110 0.322Human resources management 0.299 0.090 0.327 0.001

Dependent variable = delivery performance (not related to firm performance in jobshops)d

Manufacturing design 0.241 0.084 0.273 0.005Human resources management 0.068 0.091 0.084 0.455Manufacturing infrastructure 0.123 0.135 0.125 0.363Manufacturing technology (−)0.038 0.078 (−)0.059 0.629

a R2 = 0.179, adjusted R2 = 0.146, significant at P = 0.001.b R2 = 0.168, adjusted R2 = 0.136, significant at P = 0.001.c R2 = 0.180, adjusted R2 = 0.149, significant at P = 0.003.d R2 = 0.074, adjusted R2 = 0.065, significant at P = 0.005.

critical AMT dimension was computed to developan ‘ideal’ profile of AMT resource configuration, foreach process environment. Table 9 presents the ‘idealprofiles’ developed based on this procedure for job-shop and assembly line processes, respectively. The‘ideal profiles’ show the distinct combinations (and

Page 14: Process-technology fit and its implications for manufacturing performance

534 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

Table 8Regression runs-AMT domains vs. strategic manufacturing capabilities assembly line process (n = 117)

Independent variables B S.E. B β P

Dependent variable: manufacturing cost reduction performancea

Manufacturing technology 0.058 0.075 0.085 0.441Manufacturing design 0.102 0.071 0.143 0.153Manufacturing infrastructure 0.002 0.079 0.003 0.981Human resources management 0.152 0.070 0.226 0.032

Dependent variable: quality performanceb

Manufacturing technology 0.202 0.101 0.214 0.048Manufacturing design 0.058 0.096 0.058 0.548Manufacturing infrastructure 0.086 0.105 0.082 0.416Human resources management 0.117 0.095 0.125 0.221

Dependent variable: new product introduction timereduction performance (not related to firm performancein assembly line firms)c

Human resources management (only significant variable) 0.251 0.068 0.331 0.000

a R2 = 0.126, adjusted R2 = 0.094, significant at P = 0.005.b R2 = 0.135, adjusted R2 = 0.105, significant at P = 0.003.c R2 = 0.110, adjusted R2 = 0.102, significant at P = 0.0.

intensities of use) of manufacturing technologiesand practices, that were found to be associated withsuperior performance in jobshop and assembly lineenvironments.

The identification of ‘ideal profiles’ for differentprocesses validated the first research proposition ofthis paper: different process environments tend to

Table 9‘Ideal’ profiles of AMT resource allocations

Significant AMT domaina Item measure Mean score (and S.D.) of ideal profile

Jobshop Assembly line

MT Use of CNC technology 2.83 (1.40) 3.27 (1.85)Use of computer-aided manufacturing 3.30 (1.34) 3.36 (1.69)Use of flexible manufacturing systems 3.20 (1.40) 2.91 (1.87)Use of set-up time reduction methods – 3.45(1.63)

HRM Worker cross-training 3.80 (1.03) –Use of operator teams 2.60 (1.90) 3.73 (1.42)Decentralized decision making for scheduling 2.90 (1.52) 2.91 (1.57)Decentralized decision making for operator tasks 2.90 (1.60) 3.00 (1.48)

MD Use of computer-aided engineering 3.83 (1.15) Not significantly related tostrategic manufacturingcapabilities

Use of computer-aided design 3.75 (0.82) –Use of modularization in product design 3.50 (1.27) –

a AMT domains found to significantly influence strategic manufacturing capabilities.

align advanced manufacturing technology investmentsin identifiable ‘ideal’ profiles. The composition of the‘ideal profiles’ show that processes may differ in theirchoice and intensity of use of manufacturing tech-nologies and practices associated with superior man-ufacturing and business performance. For instance,jobshops were seen to deploy CNC technology and

Page 15: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 535

manufacturing operator teams to a lesser extent, ascompared to assembly lines. On the other hand, bothprocess types emphasized decentralized decision mak-ing about equally. The ‘ideal profiles’ also identifiedseveral technologies and practices that were exclusiveto a specific process type. Worker cross-training, theuse of CAE and CAD, and the use of modularizationin design were found to be associated with improvedstrategic manufacturing performance in jobshops. Thisdoes not necessarily mean that successful assemblyline process firms do not deploy these technologies.Rather, the use of these technologies in assembly linesdid not appear to contribute to those manufacturingperformance dimensions that were found to affect firmperformance in assembly line environments.

The second research proposition of this paperposited that deviations from the ‘ideal’ profiles willhave a negative impact on manufacturing perfor-mance. Accordingly, a measure of deviation from theideal profile, ‘lack of fit’ (LOF), was calculated toassess the extent to which the remaining firms di-verged from the ideal profile of mean scores in AMTinvestments. ‘LOF’ represents a composite, weightedmeasure based on the deviation of a firm’s AMT itemmeasure scores from the mean item scores of thegroup of firms comprising the ideal profile. Follow-ing Venkatraman and Prescott (1990), weights werederived from the beta coefficients of particular AMT

dimensions, obtained from their regressions on strate-gic manufacturing capabilities (see Tables 7 and 8).

LOFi =∑

j∈J

(bj (xij − µpj))2

where i is the ith firm in sample (excluding firms inideal profile), bj , the standardized beta weight for thejth significant AMT dimension, J the set of item mea-sures for the critical AMT dimensions for a particu-lar process environment, xij the score for firm i forjth item measure and µpj the mean score for jth itemmeasure in the ideal profile.

A ‘LOF’ measure was calculated for each firm be-longing to the two process environments. A represen-tative computation is shown as follows:

Assembly process

Strategic manufacturing capabilities Manufacturing cost reduction and quality performance (see Table 6)AMT dimensions affecting cost and MT (β = 0.214); HRM (β = 0.226) (see Table 7)

quality capabilitiesItem measures of MT and HRM MT: Use of CNC machines, use of CAM, use of FMS, use of set-up

time reduction techniquesHRM: Use of operator teams in manufacturing, decentralized decisionmaking for distributing operator tasks, decentralized decision makingfor production scheduling

Illustrative computation for a firm2 LOF = ∑[0.214{(firm’s score on CNC − 3.27)2 +

(firm’s score on CAM − 3.36)2 + (firm’s score on FMS − 2.91)2 +(firm’s score on set-up time reduction techniques − 3.45)2} +0.226{(firm’s score on operator team − 3.73)2 +(firm’s score on decentralized decision making for scheduling−2.91)2+(firm’s score on decentralized decision making for opeator tasks −3.00)2}]

The top 10% cases and the bottom 10% compos-ite manufacturing performance cases were eliminatedfrom the sample prior to this computation. The latterwas done to avoid the possibility of restriction of rangebias (Venkatraman and Prescott, 1990). Finally, ‘LOF’for each process environment was related to compos-ite manufacturing performance. It was anticipated thatdeviations from the ideal profile would have an ad-verse effect on performance. Accordingly, the corre-lations between ‘LOF’ and composite manufacturing

2Values inside brackets indicates mean score on that item inthe ideal profile of assembly line firms.

Page 16: Process-technology fit and its implications for manufacturing performance

536 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

Table 10Correlations between ‘lack of fit’ and composite manufacturingperformancea

JobshopLOF↔(−)0.261∗

Composite manufacturing performanceAssemblyLine

LOF↔(−)0.226∗Composite manufacturing performance

a Asterisk (∗) indicates significant at P < 0.01.

performance were expected to be significant and neg-ative. Table 10 shows the results of the correlationanalysis.

‘LOF’ was found to have a significant and negativeimpact on composite manufacturing performance, inboth jobshop and assembly line processes.

6. Discussion of results

This section discusses the implications of the idealprofiles and the use of AMT for strategic ends.

6.1. Ideal profiles

The results of the analysis validate the utility ofthe ‘ideal profile’ methodology in production envi-ronments. Development of an ideal profile representswhat the strategy literature calls a ‘holistic’ approachto assessing fit (Venkatraman and Prescott, 1990). Theholistic perspective is based on the premise that con-figurations, rather than bi-variate examinations (as inregular interaction analysis), are important to com-pletely describe a synergistic profile or system. Thetechnology items in the ideal profiles in this studycollectively describe and define the strategic technol-ogy profile of the best performing (manufacturing per-formance) companies. It is not necessary that meaninvestments in all the technology items in an idealprofile be uniformly higher than (or different from)that in the remaining firms. For example, the meansscores for CNC technology use CAE and CAD, threecritical technology items for the low performing job-shop firms, were higher than the mean scores for theseitems in the ideal profile of the high performing job-shops (3.00, 3.91 and 4.14 versus 2.83, 3.83 and 3.75,respectively). This suggests that indiscriminate invest-ment in all technology practices may not be associatedwith improved performance. It can be speculated that

the firms in the ideal profile derive superior manufac-turing performance from the synergies implicit in thearrangement of their technology investments — theissue is one of whether a specific pattern of invest-ments is associated with performance gains, and notof the relative magnitude of investments in individualtechnology practices. It is not our intention to general-ize or prescribe an established ‘ideal profile’ of AMTinvestments for individual jobshop or assembly lineprocess plants. Product-based differences and equifi-nality considerations make it conceivable that otherideal profiles will exist. Further study is required tofully investigate these configurational synergies, per-haps most appropriately through a case-based in-depthanalysis of manufacturing plants in the same productmarket. In this study, the ‘ideal’ profile concept offersa perspective on investment implications that can pro-vide more clarity on synergies to be had in technologychoice decisions.

The relationship between LOF and performance un-derscores the need to recognize patterns in technologyinvestments for competitive benchmarking and anal-ysis. The ‘ideal’ profile methodology could provide avalid composite, multivariate benchmarking tool fordecision-makers. An ideal profile could provide man-agers with an opportunity to utilize the ‘LOF’ measureto assess the degree of deviation from the benchmarkpattern of investments, identify gaps, consider avail-able resources and associated risks, and implementcorrective strategies. To illustrate, a company operat-ing multiple plants in a single industry could constructan ‘ideal profile’ of technology investments from thebest performing plant in its industry, and assess theperformance implications of plant deviations from thisprofile. If deviations from the best plant profile affectperformance significantly, management needs to iden-tify the manufacturing levers to adjust them for im-proved performance, i.e. closing the gaps in the plantprofile to reduce LOF and achieve better fit with theideal profile. For example, if comparison of a plant’sprofile with the ideal profile reveals gaps in opera-tor teaming, use of CNC machines and use of CAD,management should identify the reasons for thesegaps or deficiencies (lack of resources, lack of equip-ment, plant culture, worker skills), consider availableresources, and establish and sequence targets for im-provement in these areas. For practical reasons, it maybe easier for managers to initiate action in the softer,

Page 17: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 537

less capital intensive operational dimensions of AMT,such as HRM, and develop an improved profile, usingthese dimensions. They can then quantify the extent ofdeviation of the new improved profile from the ‘idealprofile’, following the methodology outlined here,re-assess (correlate) the performance consequences ofthe deviation, and subsequently make the capital-basedequipment MT and MD investments, if warrantedfrom a cost/resource availability perspective.

A comparison of the ideal profiles developed fromthe data with the envisaged ideal profiles (see Table 1)revealed that most of the AMT items (belonging tostrategic AMT domains) in the a priori profiles foundinclusion in the final profile. Few exceptions are notedon Table 1, and can be explained as an outcome ofthe factor analysis, or as a result of a particular AMTdimension not having a significant association withperformance. The results generally support earlierfindings of the broad use of AMT across different pro-cess environments, albeit to achieve different strategicobjectives (Kotha and Swamidass, 2000; McDermottet al., 1997).

6.2. Advanced manufacturing technology and‘off-diagonal’ positions

6.2.1. AMT’s potentialThe results show that the general purpose/dedicated

purpose technology investment divide between job-shop and assembly environments (Hill, 1994), is beingbridged by advanced manufacturing technology. Theuse of advanced manufacturing technology appearsto influence strategic capabilities across disparateprocess environments. Theoretically, advanced man-ufacturing technology should enable firms to operatewith success in off-diagonal positions, especially ifthe technology elements are aligned with an ‘idealprofile’ for that particular product industry. In suchcases, an ‘aligned’ firm would have an enhanced abil-ity to compete with non-traditional capabilities. Toillustrate, a regression analysis showed that manufac-turing design, a dimension of AMT, significantly af-fected delivery performance in jobshops (see Table 7).This suggests that it may be feasible for jobshops toventure into delivery oriented markets based on exist-ing technology capabilities. Similarly, regression runsin assembly environments showed a significant link-

age between human resources management practices,and new product introduction time reduction perfor-mance (see Table 8), a manufacturing capability thatwas not related to firm performance in this processtype. The data suggest that HRM practices such asdecentralization and teaming could help assemblyline manufacturers make successful lateral extensionsinto “off-diagonal” positions in customization andinnovation driven markets.

6.2.2. AMT’s realityWith the exception of cost, companies in different

process environments used similar advanced manu-facturing technology investments for very differentstrategic manufacturing goals. This finding disaffirmssuggestions of a growing shift to “off-diagonal” posi-tions, where for instance, continuous flow shops useFMS to serve customization driven markets (Safizadehet al., 1996; Markland et al., 1998). The nature ofthe manufacturing related “order winners” for eachprocess environment (see Table 6) implies that firmsoperate within their traditional market environmentsusing compatible capabilities, without attempting tore-define competition by pursuing non-traditionalcapabilities. The strategic opportunities offered byadvanced manufacturing technology to dilute the in-fluence of the volume–variety tradeoff seems to haveremained unexploited. The notion that firms remainwithin their traditional markets is supported by thedata. The data show that the proportion of produc-tion changes in jobshops (after release of productionschedules) due to changes in design and volume, wasin the range of 30–60%. Evidently, this suggests avolatile market environment for jobshops, where useof customization and new product introduction ca-pabilities would be a preferred strategy for pursuingmarket performance. The results show (see Table 6)that firm performance in jobshops was indeed pos-itively related to these manufacturing capabilities.In contrast, delivery performance in jobshops didnot relate to their financial or market growth perfor-mance, suggesting that jobshops were not operating intime sensitive markets. The proportion of productionchanges in assembly line firms was less than 10%for design changes, and 10–20% for volume changes,indicating a much more stable and mature market en-vironment for these firms. Accordingly, it can be ex-pected that these companies would emphasize quality,

Page 18: Process-technology fit and its implications for manufacturing performance

538 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

cost and other mature-market competitive goals. Theresults validate this view — quality and cost reductionperformance in assembly line firms were found to bepositively related to financial and market growthperformance, whereas customization or new prod-uct development was not. Inferentially, utilizationof advanced manufacturing technology in assemblyenvironments appears to center on controlling costsand improving quality, whereas advanced manufac-turing technology in jobshops seems to be utilizedfor cost, customization and new product introductionobjectives.

Cost has not been a pressing concern for jobshops,which have traditionally competed on the basis ofcustomization and product innovation. Manufacturingtechnology and human resources management prac-tices, two key dimensions of AMT, were associatedwith cost reduction performance in jobshops (see Ta-ble 7). Advanced manufacturing technology wouldthus appear to have enabled jobshop firms in our sam-ple to add cost-based strategies to their competitivearsenal, an advantage lacking with general-purposetechnology.

6.2.3. AMT’s paradoxThe findings of this investigation present an inter-

esting paradox in the use of advanced manufacturingtechnology. Clearly, AMT has the potential to legit-imize off-diagonal positions. Yet it is evidently notbeing utilized for such objectives. Jobshops in ourstudy did not seem to use their AMT-based capabil-ities to pursue markets with non-traditional capabili-ties such as delivery performance. Similarly, assemblyfirms in our sample did not appear to be deriving fi-nancial or market gains from their technology enabledability to accelerate new product development cycle-times.

A plausible reason for the lack of use of AMT togain entry into markets with different competitive re-quirements could be that managers who acquire andutilize AMT for traditional objectives, do not spendtime identifying the additional competitive opportuni-ties available from the technology. AMT’s capabilityto promote customization and product developmentmay not find articulation or application in assemblyenvironments. Another more reasonable explanationfor the paradox could lie in the frequent separationbetween the manufacturing function and strategy for-

mulation process seen in many companies. Strategyplanners may lack knowledge of manufacturing capa-bilities, especially if such capabilities are not readilyapparent to them. Marketing’s decision to competein specific markets may in fact determine the use ofAMT, in the absence of manufacturing’s input in busi-ness strategy formulation. Regardless of the reasonsfor not exploiting the potential of AMT, it is pertinentto note that there is a growing body of evidence thatreports convergence of markets in terms of customerexpectations: customization, cost and quality (Pine,1993; Kotha, 1995). In view of this emerging trend,both jobshop and assembly line firms would be wellserved if they identify and exploit the full capabili-ties of advanced manufacturing technologies. Man-ufacturing can ill afford to neglect AMT’s strategiccapability to introduce new forms of competition.

Although this study takes a systemic perspectivein understanding technology investments, the statis-tical relationships between specific AMT dimensionsand strategic manufacturing capabilities merit men-tion (see Tables 7 and 8). Manufacturing technologywas found to benefit manufacturing cost reduction injobshops. One can rationalize this relationship. For in-stance, CNC and CAM technologies should result indecreased set-up frequencies, improved set-up times,more accurate scheduling and increased productivity,all leading to reductions in manufacturing cycle times.Lower manufacturing cycle times should lead to lowermanufacturing costs. To verify this, MT was regressedon manufacturing cycle time reduction in jobshops,with significant and positive results (P < 0.05). Man-ufacturing design (CAD and CAE systems) was foundto relate positively to customization performance injobshops. Human resources management was foundto be an overarching influence on the achievement ofstrategic manufacturing capabilities, supporting cur-rent notions of the strategic importance of knowledgefactors. Manufacturing managers should be trained todevelop, foster, evaluate and reward the human ele-ment in operations.

7. Conclusion

The development and testing of ‘ideal’ profiles ofadvanced manufacturing technology resource invest-ments underscores the strategic importance of aligning

Page 19: Process-technology fit and its implications for manufacturing performance

A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540 539

technology investments with process requirements.Theoretically, the existence of ideal profiles highlightsthe concept of ‘domain navigation’ by companies —the design of operational strategies to match com-petitive environmental conditions (Venkatraman andPrescott, 1990). The results validate the feasibility ofoff-diagonal positions in the Hayes and Wheelwright(1979) matrix, yet concurrently underscore the failureof firms to adopt non-traditional positions for com-petitive advantage. Contrary to recent speculations,the results suggest that the use of AMT is still con-fined to defined process goals and does not appear toencourage firms to assume “off diagonal” positions inthe Hayes and Wheelwright (1979) framework. Thisparadox bears further exploration.

This study validates a methodology for assessingdeviations from best-in-class patterns in a specific in-dustry, and offers evidence of the existence of perfor-mance linked patterns in technology investments. Theresults can provide a stimulus for product focused,case-based research that could lead to a more com-plete understanding of how and why configurationalsynergies develop in AMT investments. The impact oftransitions in process environments on technology in-vestments also requires investigation. A longitudinalinvestigation would allow the study of such changes.It would also be interesting to examine whether in-vestments in advanced manufacturing technology do-mains follow particular implementation sequences indifferent process environments.

References

Adler, P.S., 1988. Managing flexible automation. CaliforniaManagement Review 30 (3), 34–56.

Birou, L.M., Fawcett, S.E., Magnan, G.M., 1998. The product lifecycle: a tool for functional strategic alignment. InternationalJournal of Purchasing and Materials Management 34 (2), 37–51.

Boyer, K., Ward, P.T., Leong, K.G., 1996. Approaches to thefactory of the future: an empirical taxonomy. Journal ofOperations Management 14, 297–313.

Burgess, T.F., Guler, H.K., 1998. Buyer supplier relationshipsin films adopting advanced manufacturing and technology— an empirical analysis of the implementation of hard andsoft technologies. Journal of Engineering and TechnologyManagement 15, 127–152.

Burns, B., Stalker, G.M., 1961. The Management of Innovation.Tavistock Publications, London.

Cagliano, R., Spina, G., 2000. Advanced manufacturingtechnologies and strategically flexible production. Journal ofOperations Management 18 (2), 169–190.

Carr, A.S., Pearson, J.H., 1999. Strategically managedbuyer–supplier relationships and performance outcomes. Journalof Operations Management 17, 497–519.

Choe, K., Booth, D., Hu, M., 1997. Production competence andits impact on business performance. Journal of ManufacturingSystems 16 (6), 409–421.

Churchill, G.A., 1979. A paradigm for developing better measuresof marketing constructs. Journal of Marketing Research 16,64–73.

Clark, K.B., 1996. Competing through manufacturing and the newmanufacturing paradigm: is manufacturing passes? Productionand Operations Management 5 (1), 42–58.

Cohen, M.A., Apte, U.A., 1997. Manufacturing Automation. Irwin,Chicago, IL.

Dean Jr., J.W., Snell, S.A., 1996. The strategic use ofintegrated manufacturing: an empirical examination. StrategicManagement Journal 17, 459–480.

Dess, G.G., Robinson, R.B., 1984. Measuring organizationalperformance in the absence of objective measures: the case ofprivately-held firms and conglomerate business units. StrategicManagement Journal 5, 265–273.

Dillman, D.A., 1978. Mail and telephone surveys. The total designmethod. Wiley Interscience, New York.

Drazin, R., Van de Ven, A.H., 1985. Alternative forms of fit incontingency theory. Administrative Science Quarterly 30, 514–539.

E∗Trade Inc., 2000. Market Guide Inc., 2001, Marcus Avenue,Lake Success, NY 11042.

Fornell, C., Larcker, D.F., 1981. Evaluating structural modelsin unobservable variables and measurement error. Journal ofMarketing Research XVIII, 39–50.

Gupta, Y.P., Somers, T.M., 1996. Business strategy, manufacturingflexibility, and organizational performance relationships: a pathanalysis approach. Production and Operations Management5 (3), 204–233.

Hambrick, D.C., 1980. Operationalizing the concept ofbusiness-level strategy in research. Academy of ManagementReview 5, 567–575.

Hayes, R.H., Wheelwright, S.C., 1979. Linking manufacturingprocess and product life cycles. Harvard Business Review57 (1), 133–140.

Hill, T.J., 1994. Manufacturing Strategy: Text and Cases. Irwin,Burr Ridge, IL.

Hofer, C.W., 1975. Towards a contingency theory of businessstrategy. Academy of Management Journal 18, 210–784.

James, L.R., Demaree, R.G., Wolf, G., 1984. Estimatingwithin-group inter-rater reliability with and without responsebias. Journal of Applied Psychology 69 (1), 85–98.

Kotha, S., 1995. Mass customization: implementing the emergingparadigm for competitive advantage. Strategic ManagementJournal 16 (Special Issue), 21–22.

Kotha, S., Swamidass, P.M., 2000. Strategy, advancedmanufacturing technology and performance: empirical evidencefrom US manufacturing firms. Journal of OperationsManagement 18, 257–277.

Page 20: Process-technology fit and its implications for manufacturing performance

540 A. Das, R. Narasimhan / Journal of Operations Management 19 (2001) 521–540

Krajewski, L.J., Ritzman, L.P., 1999. Operations Management:Strategy and Analysis. Addison-Wesley, Longman, New York.

Markland, R.E., Vickery, S.K., Davis, R.A., 1998. OperationsManagement: Concepts in Manufacturing and Services.South-Western College Publishing, Cincinnati, OH.

McCutcheon, D.M., Raturi, A.S., Meredith, J.R., 1994. Thecustomization-responsiveness squeeze. Sloan ManagementReview 35 (2), 89–99.

McDermott, C.M., Stock, G.N., 1999. Organizational culture andadvanced manufacturing technology implementation. Journal ofOperations Management 17, 521–533.

McDermott, C.M., Greis, N.P., Fischer, W.A., 1997. Thediminishing utility of the product–process matrix — a study ofthe US power tool industry. International Journal of Operations& Production Management 17 (1), 65–84.

Miller, J.G., Roth, A.V., 1994. A taxonomy of manufacturingstrategies. Management Science 40 (3), 285–304.

Nunnally, J., 1978. Psychometric Theory. McGraw-Hill, New York.Parthasarthy, R., Sethi, S.P., 1992. The impact of flexible

manufacturing automation on business strategy andorganizational structure. The Academy of Management Review17 (1), 86–105.

Philips, J., 1981. Assessing measurement error in key informantreports: a methodological note on organizational analysis inmarketing. Journal of Marketing Research 18 (4), 395–415.

Pine, J.B., 1993. Mass Customization: The New Frontier inBusiness Competition. Harvard Business School Press, Boston,MA.

Roth, A.V., Giffi, C.A., 1995. Critical factors for achieving worldclass manufacturing. OM Review 10 (2), 1–29.

Safizadeh, M.H., Ritzman, L.P., Sharma, D., Wood, C., 1996. Anempirical analysis of the product–process matrix. ManagementScience 42 (11), 1576–1591.

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

Snell, S.A., Dean Jr., J.W., 1992. Integrated manufacturing andhuman resource management: a human capital perspective.Academy of Management Journal 35 (3), 467–504.

Swamidass, P.M., Newell, W.T., 1987. Manufacturing strategy,environmental uncertainty and performance: a path analyticmodel. Journal of Operations Management 17 (1), 23–37.

Swamidass, P.M., Kotha, S., 1998. Explaining manufacturingtechnology use, firm size and performance using a multi-dimensional view of technology. Journal of OperationsManagement 17 (1), 23–37.

Tinsley, H.E.A., Tinslay, D.J., 1987. Uses of factor analysisin counseling psychology research. Journal of CounselingPsychology 34 (4), 414–424.

Tranfield, D., Smith, S., Ley, C., Bessant, J., Levy, P., 1991.Changing organizational design and practices for computerintegrated technologies. International Journal of TechnologyManagement 6 (3/4), 211–221.

US Department of Commerce, 1988. Manufacturing Technology,1988. A Bureau of Census Report, US Government PrintingOffice, Washington, DC.

Upton, D.M., McAfee, A.P., 1988. Computer integration andcatastrophic process failure. Production and OperationsManagement 7 (3), 265–281.

Venkatraman, N., 1990. Performance implications of strategicco-alignment: a methodological perspective. Journal ofManagement Studies 27 (1), 19–41.

Venkatraman, N., Ramanujam, V., 1987. Measurement ofbusiness economic performance: an examination of methodconvergence. Journal of Management 13 (1), 109–112.

Venkatraman, N., Prescott, J., 1990. Environment strategyco-alignment: an empirical test of its performance implications.Strategic Management Journal 11, 1–23.

Ward, P.T., Berger, P.D., Miller, J.G., Rosenthal, S.R.,1992. Manufacturing process technology and support staffcomposition: an empirical view of industry evidence. Productionand Operations Management 1, 5–21.

Ward, P.T., Leong, G.K., Boyer, K.K., 1994. Manufacturingproactiveness and performance. Decision Sciences 25 (3), 337–355.